Grant Sanderson: Math, Manim, Neural Networks & Teaching with 3Blue1Brown

Grant Sanderson (02:15.880)

looking at how many people go to their site and get their stuff in order to determine

Grant Sanderson (02:19.320)

if they want to support us for the long term, so you know what to do.

Lex Fridman (02:22.960)

It's the best way to support this podcast, as always.

Grant Sanderson (02:26.680)

This show is sponsored by Dollar Shave Club.

Grant Sanderson (02:28.960)

Try them out with a one time offer for only $5 and free shipping at DollarShaveClub.com

Grant Sanderson (02:34.560)

slash Lex.

Grant Sanderson (02:36.600)

The starter kit comes with a six blade razor, refills, and all kinds of other stuff that

Grant Sanderson (02:41.200)

makes shaving feel great.

Grant Sanderson (02:43.160)

I've been a member of Dollar Shave Club for over five years now, and actually signed up

Grant Sanderson (02:47.040)

when I first heard about them on the Joe Rogan podcast, and now we have come full circle.

Grant Sanderson (02:53.080)

I feel like I've made it, now that I can do a read for them just like Joe did all those

Grant Sanderson (02:57.720)

years ago.

Grant Sanderson (02:58.720)

For the most part, I've just used the razor and the refills, but they encouraged me to

Grant Sanderson (03:02.360)

try the shave butter, which I've never used before, so I did, and I love it.

Grant Sanderson (03:07.880)

Not sure how the chemistry of it works out, but it's translucent somehow, which is a

Grant Sanderson (03:11.920)

cool new experience.

Grant Sanderson (03:13.400)

Again, try the Ultimate Shave Starter Set today for just $5 plus free shipping at DollarShaveClub.com

Grant Sanderson (03:19.520)

slash Lex.

Lex Fridman (03:21.440)

This show is also sponsored by DoorDash.

Grant Sanderson (03:24.480)

Get $5 off and zero delivery fees on your first order of $15 or more when you download

Lex Fridman (03:29.520)

the DoorDash app and enter code LEX.

Grant Sanderson (03:32.440)

I have so many memories of working late nights for a deadline with a team of engineers and

Grant Sanderson (03:37.640)

eventually taking a break to argue about which DoorDash restaurant to order from, and when

Grant Sanderson (03:42.160)

the food came, those moments of bonding, of exchanging ideas, of pausing to shift attention

Lex Fridman (03:48.280)

from the programs to the humans were special.

Grant Sanderson (03:53.180)

These days, for a bit of time, I'm on my own, sadly, so I miss that camaraderie.

Lex Fridman (03:58.360)

But actually DoorDash is still there for me.

Grant Sanderson (04:00.640)

There's a million options that fit into my keto diet ways.

Lex Fridman (04:04.320)

Also it's a great way to support restaurants in these challenging times.

Grant Sanderson (04:07.800)

Once again, download the DoorDash app and enter code LEX to get $5 off and zero delivery

Lex Fridman (04:13.020)

fees on your first order of $15 or more.

Grant Sanderson (04:16.520)

Finally, this show is presented by Cash App, the number one finance app in the App Store.

Lex Fridman (04:21.760)

When you get it, use code LEX PODCAST.

Grant Sanderson (04:24.760)

Cash App lets you send money to friends, buy bitcoin, and invest in the stock market with

Lex Fridman (04:28.940)

as little as $1.

Grant Sanderson (04:31.040)

It's one of the best design interfaces of an app that I've ever used.

Lex Fridman (04:34.760)

To me, good design is when everything is easy and natural.

Grant Sanderson (04:38.280)

Bad design is when the app gets in the way either because it's buggy or because it tries

Lex Fridman (04:42.640)

too hard to be helpful.

Grant Sanderson (04:44.880)

I'm looking at you, Clippy.

Grant Sanderson (04:46.360)

Anyway, there's a big part of my brain and heart that love to design things and also

Grant Sanderson (04:50.400)

to appreciate great design by others.

Lex Fridman (04:52.800)

So again, if you get Cash App from the App Store or Google Play and use code LEX PODCAST,

Grant Sanderson (04:57.920)

you get $10 and Cash App will also donate $10 to FIRST, an organization that is helping

Lex Fridman (05:03.200)

to advance robotics and STEM education for young people around the world.

Lex Fridman (05:08.560)

And now here's my conversation with Grant Sanderson.

Lex Fridman (05:13.880)

You've spoken about Richard Feynman as someone you admire.

Grant Sanderson (05:17.360)

I think last time we spoke, we ran out of time.

Lex Fridman (05:21.420)

So I wanted to talk to you about him.

Lex Fridman (05:24.760)

Who is Richard Feynman to you in your eyes?

Lex Fridman (05:27.280)

What impact did he have on you?

Grant Sanderson (05:29.400)

I mean, I think a ton of people like Feynman.

Lex Fridman (05:31.800)

It's a little bit cliche to say that you like Feynman, right?

Grant Sanderson (05:34.280)

That's almost like when you don't know what to say about sports and you just point to

Lex Fridman (05:38.720)

the Super Bowl or something or something you enjoy watching.

Lex Fridman (05:41.320)

But I do actually think there's a layer to Feynman that sits behind the iconography.

Grant Sanderson (05:45.600)

One thing that just really struck me was this letter that he wrote to his wife two years

Grant Sanderson (05:50.540)

after she died.

Lex Fridman (05:51.780)

So during the Manhattan Project, she had polio.

Grant Sanderson (05:54.320)

Tragically she died.

Lex Fridman (05:55.500)

They were just young, madly in love.

Lex Fridman (05:58.080)

And the icon of Feynman is almost this mildly sexist, womanizing philanderer, at least on

Lex Fridman (06:06.000)

the personal side.

Lex Fridman (06:08.160)

But you read this letter, and I can try to pull it up for you if I want.

Lex Fridman (06:10.840)

And it's just this absolutely heartfelt letter to his wife saying how much he loves her even

Grant Sanderson (06:15.840)

though she's dead and what she means to him, how no woman can ever measure up to her.

Lex Fridman (06:21.760)

And it shows you that the Feynman that we've all seen in Surely You're Joking is different

Grant Sanderson (06:26.820)

from the Feynman in reality.

Lex Fridman (06:28.860)

And I think the same kind of goes in his science, where he sometimes has this output of being

Grant Sanderson (06:35.340)

this aw shucks character, like everyone else is coming in with these fancyfalutin formulas,

Lex Fridman (06:39.960)

but I'm just going to try to whittle it down to its essentials, which is so appealing because

Grant Sanderson (06:43.240)

we love to see that kind of thing.

Lex Fridman (06:45.120)

But when you get into it, what he was doing was actually quite deep, very much mathematical.

Grant Sanderson (06:52.060)

That should go without saying, but I remember reading a book about Feynman in a cafe once,

Lex Fridman (06:55.160)

and this woman looked at me and saw that it was about Feynman.

Grant Sanderson (06:58.720)

She was like, oh, I love him.

Lex Fridman (06:59.720)

I read Surely You're Joking.

Lex Fridman (07:01.220)

And she started explaining to me how he was never really a math person.

Lex Fridman (07:04.880)

And I don't understand how that can possibly be a public perception about any physicist,

Lex Fridman (07:10.040)

but for whatever reason, that worked into his art that he shooed off math in place of

Lex Fridman (07:14.660)

true science.

Grant Sanderson (07:15.660)

The reality of it is he was deeply in love with math and was much more going in that

Grant Sanderson (07:18.880)

direction and had a clicking point into seeing that physics was a way to realize that and

Grant Sanderson (07:22.880)

all the creativity that he could output in that direction was instead poured towards

Grant Sanderson (07:27.560)

things like fundamental, not even fundamental theories, just emergent phenomena and everything

Grant Sanderson (07:31.680)

like that.

Lex Fridman (07:32.760)

So to answer your actual question, like what, what, what I like about, uh, his way of going

Grant Sanderson (07:37.880)

at things is this constant desire to reinvent it for himself.

Grant Sanderson (07:42.320)

Like when he would consume papers, the way he'd describe it, he's, he would start to

Grant Sanderson (07:45.320)

see what problem he was trying to solve and then just try to solve it himself to get a

Lex Fridman (07:48.340)

sense of personal ownership.

Lex Fridman (07:50.040)

And then from there, see what others had done.

Lex Fridman (07:52.560)

Is that how you see problems yourself?

Grant Sanderson (07:54.120)

Like that's actually an interesting point when you first are inspired by a certain idea

Lex Fridman (08:00.360)

that you maybe want to teach or visualize or just explore on your own.

Grant Sanderson (08:05.040)

I'm sure you're captured by some possibility and magic of it.

Lex Fridman (08:08.420)

Do you read the work of others?

Lex Fridman (08:10.600)

Like do you go through the proofs or do you try to rediscover everything yourself?

Lex Fridman (08:15.000)

So I think the things that I've learned best and have the deepest ownership of are the

Grant Sanderson (08:19.120)

ones that have some element of rediscovery.

Lex Fridman (08:21.200)

The problem is that really slows you down.

Lex Fridman (08:23.260)

And this is for my, for my part, it's actually a big fault.

Lex Fridman (08:25.400)

Like this is part of why I'm, I'm not an active researcher.

Grant Sanderson (08:28.240)

I'm not like at the depth of the field.

Lex Fridman (08:29.880)

A lot of other people are the stuff that I do learn.

Grant Sanderson (08:32.160)

I try to learn it really well.

Lex Fridman (08:33.880)

Um, but other times you do need to get through it at a certain pace.

Grant Sanderson (08:37.480)

You do need to get to a point of a problem you're trying to solve.

Lex Fridman (08:39.680)

So obviously you need to be well equipped to read things, uh, without that reinvention

Grant Sanderson (08:43.940)

component and see how others have done it.

Lex Fridman (08:45.960)

But I think if you choose a few core building blocks along the way and you say, I'm really

Grant Sanderson (08:49.080)

going to try to approach this, um, before I see how this person went at it, I'm really

Lex Fridman (08:54.920)

going to try to approach it for myself.

Grant Sanderson (08:56.880)

No matter what you gain, all sorts of inarticulatable intuitions about that topic, which aren't

Lex Fridman (09:02.020)

going to be there.

Grant Sanderson (09:03.020)

If you simply go through the proof, for example, you're going to be, um, trying to come up

Lex Fridman (09:06.600)

with counter examples.

Grant Sanderson (09:07.880)

You're going to try to come up with, um, intuitive examples, all sorts of things where you're

Lex Fridman (09:11.920)

populating your brain with data.

Lex Fridman (09:13.560)

And the ones that you come up with are likely to be different than the one that the text

Lex Fridman (09:16.120)

comes up with and that like lends at a different angle.

Lex Fridman (09:19.240)

So that aspect also slowed Feynman down in a lot of respects.

Grant Sanderson (09:22.640)

I think there was a period when like the rest of physics was running away from him.

Grant Sanderson (09:26.480)

Um, but in so far as got, it got him to where he was, uh, I, I kind of resonate with that.

Grant Sanderson (09:32.240)

I just, I would, I would be nowhere near it cause I not like him at all, but it's like

Grant Sanderson (09:37.240)

a, uh, state to aspire to.

Grant Sanderson (09:41.360)

You know, just to link in a small point you made that you're not a quote unquote active

Grant Sanderson (09:46.800)

researcher, do you, you're swimming often in reasonably good depth about a lot of topics.

Lex Fridman (09:55.400)

Do you sometimes want to like dive deep at a certain moment and say, like, cause you

Grant Sanderson (10:01.280)

probably built up a hell of an amazing intuition about what is and isn't true within these

Lex Fridman (10:07.080)

worlds.

Lex Fridman (10:08.080)

Do you ever want to just dive in and see if you can discover something new?

Lex Fridman (10:14.000)

Yeah.

Grant Sanderson (10:15.000)

I, I think one of my biggest regrets from undergrad is not having built better relationships

Lex Fridman (10:20.440)

with the professors I had there.

Lex Fridman (10:21.800)

And I think a big part of success and research is that element of like mentorship and like

Grant Sanderson (10:26.400)

people giving you the kind of scaffolded problems to carry along for my own like goals right

Grant Sanderson (10:30.760)

now.

Grant Sanderson (10:31.760)

I feel like, um, I'm pretty good at exposing math to others and like want to continue doing

Grant Sanderson (10:37.320)

that for my personal learning.

Lex Fridman (10:39.800)

I, are you familiar with like the hedgehog Fox dynamic?

Grant Sanderson (10:44.720)

I think this was, um, either the ancient Greeks came up with it or it was pretended to be

Grant Sanderson (10:50.280)

something drawn from the ancient Greeks that I don't know who to point it to, but the probably

Grant Sanderson (10:53.560)

Mark Twain.

Grant Sanderson (10:54.560)

It is that you've got two types of people or especially two types of researchers.

Grant Sanderson (10:59.400)

There's the Fox that knows many different things and then the hedgehog that knows one

Lex Fridman (11:03.140)

thing very deeply.

Lex Fridman (11:04.700)

So like von Neumann would have been the Fox.

Grant Sanderson (11:06.840)

Obviously someone who knows many different things, just very foundational, a lot of different

Grant Sanderson (11:10.040)

fields.

Grant Sanderson (11:11.040)

Um, Einstein would have been more of a hedge thinking really deeply about one particular

Grant Sanderson (11:14.240)

thing and both are very necessary for making progress.

Grant Sanderson (11:17.520)

Um, so between those two, I would definitely see myself as like the Fox where, uh, I'll

Grant Sanderson (11:22.600)

try to get my pause in like a whole bunch of different things.

Lex Fridman (11:24.720)

And at the moment I just think I don't know enough of anything to make like a significant

Grant Sanderson (11:28.660)

contribution to any of them.

Lex Fridman (11:30.080)

But I do see value in, um, like having a decently deep understanding of a wide variety of things.

Grant Sanderson (11:36.760)

Like most people who, uh, know computer science really deeply don't necessarily know physics

Grant Sanderson (11:42.200)

very deeply or, uh, many of the aspects, like different fields in math, even let's say you

Grant Sanderson (11:47.120)

have like an analytic number theory versus an algebraic number theory.

Lex Fridman (11:49.760)

Like these two things end up being related to very different fields.

Grant Sanderson (11:52.440)

Like some of them more complex analysis, some of them more like algebraic geometry.

Lex Fridman (11:56.240)

And then when you just go out so far as to take those adjacent fields, place one, you

Grant Sanderson (12:00.000)

know, PhD student into a seminar of another ones, they don't understand what the other

Lex Fridman (12:03.040)

one's saying at all.

Grant Sanderson (12:04.040)

Like you take the complex analysis specialist inside the algebraic geometry seminar, they're

Lex Fridman (12:08.480)

as lost as you or I would be.

Lex Fridman (12:10.320)

But I think, uh, going around and like trying to have some sense of what this big picture

Lex Fridman (12:14.320)

is certainly has personal value for me.

Grant Sanderson (12:16.560)

I don't know if I would ever make like new contributions in those fields, but I do think

Grant Sanderson (12:20.880)

I could make new like expositional contributions where there's kind of a notion of, uh, things

Grant Sanderson (12:26.600)

that are known, but like haven't been explained very well.

Grant Sanderson (12:29.400)

Well, first of all, I think most people would agree your videos, your teaching the way you

Grant Sanderson (12:34.600)

see the world is fundamentally often new, like you're creating something new and it

Grant Sanderson (12:42.040)

almost feels like research, even just like the visualizations, uh, the multidimensional

Grant Sanderson (12:47.280)

visualization we'll talk about.

Grant Sanderson (12:48.480)

I mean, you're revealing something very interesting that, uh, yeah, just feels like research feels

Grant Sanderson (12:54.840)

like science feels like the cutting edge of the very thing of which like new ideas and

Lex Fridman (13:01.440)

new discoveries are made of.

Grant Sanderson (13:03.160)

I do think you're being a little bit more generous than is necessarily.

Lex Fridman (13:06.800)

And I promise that's not even false humility because I sometimes think when I research

Grant Sanderson (13:10.960)

a video, I'll learn like 10 times as much as I need for the video itself and it ends

Lex Fridman (13:14.320)

up feeling kind of elementary.

Grant Sanderson (13:15.960)

Um, so I have a sense of just how far away like the stuff that I cover is from the actual

Lex Fridman (13:20.960)

depth.

Grant Sanderson (13:22.440)

I think that's natural, but I think that could also be a mathematics thing.

Grant Sanderson (13:26.320)

I feel like in the machine learning world, you like two weeks in, you feel like you've

Grant Sanderson (13:30.680)

basically mastered in mathematics.

Lex Fridman (13:34.040)

It's like, well, everything is either trivial or impossible.

Lex Fridman (13:37.520)

And it's like a shockingly thin line between the two where you can find something that's

Lex Fridman (13:41.200)

totally impenetrable.

Lex Fridman (13:42.200)

And then after you get a feel for it, it's like, Oh yeah, that whole, that whole subject

Lex Fridman (13:45.880)

is actually trivial in some way.

Lex Fridman (13:48.180)

So maybe that's what goes on.

Grant Sanderson (13:50.360)

Every researcher is just on the other end of that hump and it feels like it's so far

Grant Sanderson (13:53.240)

away, but one step actually gets them there.

Lex Fridman (13:56.120)

What do you think about, uh, sort of Feynman's teaching style or another perspective of use

Lex Fridman (14:02.720)

of visualization?

Grant Sanderson (14:05.720)

Well his teaching style is interesting because people have described like the Feynman effect

Grant Sanderson (14:09.640)

where while you're watching his lectures or while he reading his lectures, everything

Lex Fridman (14:13.520)

makes such perfect sense.

Lex Fridman (14:15.420)

So as an entertainment session, it's wonderful because it gives you this, um, this intellectual

Grant Sanderson (14:21.920)

satisfaction that you don't get from anywhere else that you like finally understand it.

Lex Fridman (14:26.440)

But the Feynman effect is that you can't really recall what it is that gave you that insight,

Lex Fridman (14:31.080)

you know, even a week later.

Lex Fridman (14:32.680)

And this is, um, this is true of a lot of books and a lot of lectures where the retention

Lex Fridman (14:36.640)

is never quite what we hope it is.

Grant Sanderson (14:38.760)

Um, so there is a risk that, uh, the stuff that I do also fits that same bill where at

Grant Sanderson (14:46.680)

best it's giving this kind of intellectual candy on giving a glimpse of feeling like

Grant Sanderson (14:50.320)

you understand something, but unless you do something active, like reinventing it yourself,

Grant Sanderson (14:55.040)

like doing problems, um, to solidify it, um, even things like space repetition memory to

Grant Sanderson (15:00.280)

just make sure that you have like the building blocks of what do all the terms mean.

Lex Fridman (15:04.420)

Unless you're doing something like that, it's not actually going to stick.

Lex Fridman (15:07.080)

So the very same thing that's so admirable about Feynman's lectures, which is how damn

Grant Sanderson (15:12.720)

satisfying they are to consume might actually also reveal a little bit of the flaw that

Grant Sanderson (15:18.680)

we should as educators all look out for, which is that that does not correlate with long

Lex Fridman (15:22.840)

term learning.

Grant Sanderson (15:23.840)

We'll talk about it a little bit.

Lex Fridman (15:25.600)

I think you've done some interactive stuff.

Grant Sanderson (15:28.280)

I mean, even in your videos, the awesome thing that Feynman couldn't do at the time is you

Lex Fridman (15:34.960)

could, since it's programmed, you can like tinker, like play with stuff.

Grant Sanderson (15:40.840)

You could take this value and change it.

Grant Sanderson (15:42.760)

You can like heroes, take the value of this variable and change it to build up an intuition,

Grant Sanderson (15:47.720)

to move along the surface or to, to change the shape of something.

Lex Fridman (15:51.960)

I think that's almost an equivalent of you doing it yourself.

Grant Sanderson (15:57.120)

It's not quite there, but you as a viewer, um, yeah, do you think there's some value

Lex Fridman (16:02.040)

in that interactive element?

Grant Sanderson (16:04.000)

Yeah, well, so what's interesting is you're saying that, and the videos are non interactive

Lex Fridman (16:08.280)

in the sense that there's a play button and a pause button.

Grant Sanderson (16:10.520)

Um, and you could ask like, Hey, while you're programming these things, why don't you program

Lex Fridman (16:13.940)

it into an interactable version?

Grant Sanderson (16:15.880)

You know, make it a Jupiter notebook that people can play with, which I should do.

Lex Fridman (16:19.320)

And that like would be better.

Grant Sanderson (16:20.760)

I think the thing about interactives though is most people consuming them, um, just sort

Lex Fridman (16:25.440)

of consume what the author had in mind.

Grant Sanderson (16:27.520)

Uh, and that's kind of what they want.

Lex Fridman (16:29.680)

Like I have a ton of friends who make interactive explanations.

Lex Fridman (16:33.360)

And when you look into the analytics of how people use them, there's a small sliver that

Lex Fridman (16:37.800)

genuinely use it as a playground to have experiments.

Lex Fridman (16:40.360)

And maybe that small sliver is actually who you're targeting and the rest don't matter.

Grant Sanderson (16:43.520)

Um, but most people consume it just as a piece of, um, like well constructed literature that

Grant Sanderson (16:48.580)

maybe you tweak with the example a little bit to see what it's getting at.

Lex Fridman (16:51.960)

But in that way, I do think like a video can get most of the benefits of the interactive,

Grant Sanderson (16:56.760)

like the interactive app, as long as you make the interactive for yourself and you decide

Lex Fridman (17:01.160)

what the best narrative to spin is.

Grant Sanderson (17:03.480)

Um, as a more concrete example, like my process with, I made this video about, um, SIR models

Grant Sanderson (17:08.600)

for epidemics and it's like this agent based bottling thing where you tweak some things

Grant Sanderson (17:13.660)

about how the epidemic spreads and you want to see how that affects its evolution.

Grant Sanderson (17:17.360)

Um, my, my, uh, format for making that was very different than others where rather than

Grant Sanderson (17:21.720)

scripting it ahead of time, I just made the playground and then I played a bunch, uh,

Lex Fridman (17:26.800)

and then I saw what stories there were to tell within that.

Grant Sanderson (17:29.480)

Um,

Lex Fridman (17:30.480)

that's cool.

Lex Fridman (17:31.480)

So your, your video had that kind of structure, it had, uh, like five or six stories or whatever

Lex Fridman (17:36.040)

it was.

Lex Fridman (17:37.160)

And like, it was basically, okay, here's a simulation, here's a model.

Lex Fridman (17:42.040)

What can we discover with this model?

Lex Fridman (17:44.440)

And here's five things I found after playing with it.

Grant Sanderson (17:46.960)

Well, cause here, the thing is a way that you could do that project is you make the

Grant Sanderson (17:51.080)

model and then you put it out and you say, here's a thing for the world to play with,

Lex Fridman (17:54.640)

like come to my website where you interact with this thing.

Grant Sanderson (17:57.160)

Um, and, and people did like sort of remake it in a, um, JavaScript way so that you can

Lex Fridman (18:01.560)

go to that website and you can test your own hypotheses.

Lex Fridman (18:04.720)

But I think a meaningful part of the value to add is not just the technology, but to

Lex Fridman (18:08.640)

give the story around it as well.

Lex Fridman (18:11.000)

And like, that's kind of my job.

Grant Sanderson (18:12.960)

It's not just to like make the, uh, the visuals that someone will look at it's to be the one

Grant Sanderson (18:17.680)

to decide what's the interesting thing to walk through here.

Grant Sanderson (18:20.320)

Um, and even though there's lots of other interesting paths that one could take, that

Grant Sanderson (18:24.280)

can be kind of daunting when you're just sitting there in a sandbox and you're given this

Grant Sanderson (18:27.120)

tool with like five different sliders and you're told to like play and discover things.

Lex Fridman (18:32.240)

Where do you do?

Lex Fridman (18:33.240)

What do you start?

Lex Fridman (18:34.240)

What are my hypotheses?

Lex Fridman (18:35.240)

What should I be asking?

Grant Sanderson (18:36.240)

Like a little bit of guidance in that direction can be what actually sparks curiosity to make

Lex Fridman (18:40.040)

someone want to, um, imagine more about it.

Grant Sanderson (18:43.120)

A few videos I've seen you do, I don't know how often you do it, but there's almost a

Grant Sanderson (18:47.720)

tangential like pause where you, here's a cool thing you say like, here's a cool thing,

Lex Fridman (18:54.280)

but it's outside the scope of this video essentially, but I'll leave it to you as homework essentially

Lex Fridman (18:59.560)

to like figure out it's a cool thing to explore.

Grant Sanderson (19:02.760)

I wish I could say that wasn't a function of laziness and that's like, you've worked

Lex Fridman (19:06.960)

so hard on making the 20 minutes already that to extend it out even further, it would take

Grant Sanderson (19:11.320)

more time.

Lex Fridman (19:12.320)

And one of your cooler videos, the homomorphic, like from the Mobius strip to this, yeah,

Grant Sanderson (19:19.640)

that's the super and you're like, yeah, you can't, uh, you can't transform the Mobius

Grant Sanderson (19:26.160)

strip into a, into a surface without it intersecting itself, but I'll leave it to you to see why

Grant Sanderson (19:35.800)

that is.

Grant Sanderson (19:36.800)

Well, I hope that's not exactly how I phrase it because I think what my hope would be is

Grant Sanderson (19:41.240)

that I leave it to you to think about why you would expect that to be true and then

Grant Sanderson (19:45.280)

to want to know what aspects of a Mobius strip do you want to formalize such that you can

Grant Sanderson (19:50.040)

prove that intuition that you have because at some point now you're starting to invent

Lex Fridman (19:55.160)

algebraic topology.

Grant Sanderson (19:56.840)

If you have these vague instincts like I want to get this Mobius strip, I want to, um, fit

Lex Fridman (1:00:01.360)

Exactly.

Grant Sanderson (1:00:02.360)

You're always going to write it as the best version of yourself.

Grant Sanderson (1:00:05.640)

You're not going to explain the phenomenon of like wanting to get out of the bed, but

Grant Sanderson (1:00:10.240)

not really wanting to get out of the bed and all of that.

Lex Fridman (1:00:13.520)

And just like zoning out for random reasons or the one that people probably don't touch

Grant Sanderson (1:00:18.720)

at all is I try to check social media once a day, but I'm like only.

Lex Fridman (1:00:24.160)

So I post and that's it.

Grant Sanderson (1:00:26.440)

When I post, I check the previous days.

Lex Fridman (1:00:28.840)

That's like my, what I try to do.

Grant Sanderson (1:00:31.880)

That's what I do like 90% of the days.

Lex Fridman (1:00:34.080)

But then I'll go, I'll have like a two week period where it's just like, I'm checking

Grant Sanderson (1:00:38.680)

the internet like, I mean, it's some, probably some scary number of times and a lot of people

Lex Fridman (1:00:44.320)

can resonate with that.

Grant Sanderson (1:00:45.320)

I think it's a legitimate addiction.

Grant Sanderson (1:00:47.480)

It's like, it's a dopamine addiction and it's, I don't know if it's a problem because as

Grant Sanderson (1:00:52.740)

long as it's the kind of socializing, like if you're actually engaging with friends and

Lex Fridman (1:00:55.920)

engaging with other people's ideas, uh, I think it can be really useful.

Grant Sanderson (1:01:00.080)

Well, I don't know.

Lex Fridman (1:01:01.480)

So like for sure I agree with you, but I'm, it's a, it's definitely an addiction because

Grant Sanderson (1:01:07.080)

for me, I think it's true for a lot of people.

Lex Fridman (1:01:09.640)

I am very cognizant of the fact I just don't feel that happy.

Grant Sanderson (1:01:14.560)

If I look at a day where I've checked social media a lot, like if I just aggregate, I did

Grant Sanderson (1:01:20.280)

a self report, I'm sure I would find that I'm just like literally on like less happy

Grant Sanderson (1:01:26.160)

with my life and myself after I've done that check.

Grant Sanderson (1:01:29.880)

When I check it once a day, I'm very like, I'm happy I even like, cause I've seen it.

Grant Sanderson (1:01:36.000)

Okay.

Grant Sanderson (1:01:37.000)

One way to measure that is when somebody says something not nice to you on the internet

Grant Sanderson (1:01:42.360)

is like when I check it once a day, I'm able to just like, like I smile, like, like I virtually,

Lex Fridman (1:01:48.600)

I think about them positively, empathetically, I send them love.

Grant Sanderson (1:01:51.520)

I don't, I don't ever respond, but I just feel positively about the whole thing.

Lex Fridman (1:01:56.080)

If I check it, if I check like more than that, it starts eating at me.

Grant Sanderson (1:02:01.520)

Like it start, there, there's an eating thing that, that happens like anxiety.

Lex Fridman (1:02:07.600)

It occupies a part of your mind that's not, doesn't seem to be healthy.

Grant Sanderson (1:02:11.160)

Same with, I mean, you, you, you put stuff out on YouTube.

Lex Fridman (1:02:15.760)

I think it's important.

Grant Sanderson (1:02:17.520)

I think you have a million dimensions that are interesting to you, but yeah, one of,

Grant Sanderson (1:02:21.260)

one of the interesting ones is the study of education and the psychological aspect of

Grant Sanderson (1:02:26.800)

putting stuff up on YouTube.

Lex Fridman (1:02:28.880)

I like now have completely stopped checking statistics of any kind.

Grant Sanderson (1:02:34.560)

I've released an episode a 100 with my dad, conversation with my dad.

Lex Fridman (1:02:39.920)

He checks, he's probably listening to this stop.

Grant Sanderson (1:02:44.920)

He checks the number of views on his, on his video, on his conversation.

Lex Fridman (1:02:49.560)

So he discovered like a reason he's new to this whole addiction and he just checks and

Grant Sanderson (1:02:54.480)

he like, he'll text me or write to me, I just passed Dawkins and I love that so much.

Lex Fridman (1:03:04.160)

Yeah.

Lex Fridman (1:03:05.160)

So he's, uh, can I tell you a funny story in that effect of like parental use of YouTube?

Lex Fridman (1:03:09.800)

Uh, early on in the channel, uh, my mom would like text me.

Grant Sanderson (1:03:14.480)

She's like, uh, the channel, the channel has had 990,000 views.

Lex Fridman (1:03:19.200)

The channel has had 991,000 views.

Grant Sanderson (1:03:20.720)

I'm like, oh, that's cute.

Grant Sanderson (1:03:22.040)

She's going to the little part on the about page where you see the total number of channel

Grant Sanderson (1:03:24.880)

views.

Lex Fridman (1:03:25.880)

No, she didn't know about that.

Grant Sanderson (1:03:27.880)

She had been going every day through all the videos and then adding them up and she thought

Grant Sanderson (1:03:33.200)

she was like doing me this favor of providing me this like global analytic that, uh, otherwise

Grant Sanderson (1:03:38.000)

wouldn't be visible.

Lex Fridman (1:03:39.000)

That's awesome.

Grant Sanderson (1:03:40.000)

It's just like this addiction where you have some number you want to follow and like, yeah,

Lex Fridman (1:03:43.680)

it's funny that your dad had this.

Grant Sanderson (1:03:44.960)

I think a lot of people have it.

Grant Sanderson (1:03:46.960)

I think that's probably a beautiful thing for like parents cause they're legitimately,

Grant Sanderson (1:03:52.080)

they're proud.

Lex Fridman (1:03:53.080)

Yeah.

Grant Sanderson (1:03:54.080)

It's, it's born of love.

Lex Fridman (1:03:55.080)

It's great.

Grant Sanderson (1:03:56.620)

The downside, I feel one, one of them is this is one interesting experience that you probably

Lex Fridman (1:04:03.800)

don't know much about cause comments on your videos are super positive.

Grant Sanderson (1:04:07.560)

Uh, but people judge the quality of how something went.

Lex Fridman (1:04:12.320)

Like I see that with these conversations by the comments.

Grant Sanderson (1:04:16.200)

Yeah.

Grant Sanderson (1:04:17.200)

Like, I'm not talking about like, you know, people in their twenties and their thirties.

Grant Sanderson (1:04:22.440)

I'm talking about like CEOs of major companies who don't have time.

Lex Fridman (1:04:27.120)

They basically, they literally, this is their evaluation metric.

Grant Sanderson (1:04:31.620)

They're like, Ooh, the comments seem to be positive and that's really concerning to me.

Grant Sanderson (1:04:35.820)

Most important lesson for any content creator to learn is that the commenting public is

Grant Sanderson (1:04:40.320)

not representative of the actual public.

Lex Fridman (1:04:42.800)

And this is easy to see.

Lex Fridman (1:04:44.100)

Ask yourself, how often do you write comments on YouTube videos?

Lex Fridman (1:04:47.400)

Most people will realize I never do it.

Grant Sanderson (1:04:49.680)

Some people realize they do, but the people who realize they never do it should understand

Lex Fridman (1:04:53.760)

that that's a sign.

Grant Sanderson (1:04:54.760)

The kind of people who are like you aren't the ones leaving comments.

Lex Fridman (1:04:58.120)

And I think this is important.

Grant Sanderson (1:04:59.120)

A number of respects, like, uh, in my case, I think I would think my content was better

Lex Fridman (1:05:03.000)

than it was if I just read comments cause people are super nice.

Grant Sanderson (1:05:06.360)

The thing is the people who are bored by it are, are put off by it in some way or frustrated

Lex Fridman (1:05:10.920)

by it.

Grant Sanderson (1:05:11.960)

Usually they just go away.

Grant Sanderson (1:05:13.420)

You're certainly not going to watch the whole video, much less leave a comment on it.

Lex Fridman (1:05:16.680)

So there's a huge under representation of like negative feedback, like well intentioned

Lex Fridman (1:05:20.720)

negative feedback because very few people actively do that.

Grant Sanderson (1:05:23.400)

Like watch the whole thing that they dislike, figure out what they disliked, articulate

Lex Fridman (1:05:26.820)

what they dislike.

Grant Sanderson (1:05:27.820)

Um, there's plenty of negative feedback that's not well intentioned, but um, for like that

Grant Sanderson (1:05:32.280)

golden kind, uh, I think a lot of YouTuber friends I have, uh, at least have gone through

Grant Sanderson (1:05:38.240)

phases of like anxiety about the nature of comments, um, that stem from basically just

Grant Sanderson (1:05:44.520)

this that it's like people who aren't necessarily representative of who they were going for

Grant Sanderson (1:05:48.600)

or misinterpreted what they're trying to say or whatever have you, or we're focusing on

Lex Fridman (1:05:52.640)

things like personal appearances as opposed to like substance.

Lex Fridman (1:05:55.880)

Um, and they come away thinking like, oh, that's what everyone thinks, right?

Lex Fridman (1:05:59.480)

That's what everyone's response to this video was.

Grant Sanderson (1:06:01.720)

Um, but a lot of the people who had the reaction you wanted them to have, like they probably

Lex Fridman (1:06:05.680)

didn't write it down.

Lex Fridman (1:06:07.120)

So very important to learn.

Grant Sanderson (1:06:09.640)

It also translates to, um, realizing that you're not as important as you might think

Lex Fridman (1:06:14.440)

you are, right?

Grant Sanderson (1:06:15.440)

Because all of the people commenting are the ones who love you the most and are like really

Grant Sanderson (1:06:19.020)

asking you to like create certain things or like mad that you didn't create like a past

Lex Fridman (1:06:22.280)

thing.

Grant Sanderson (1:06:23.280)

Um, I don't, I have such a problem.

Grant Sanderson (1:06:26.560)

Like I have a very real problem with making promises about a type of content that I'll

Grant Sanderson (1:06:30.060)

make and then either not following up on it soon or just like never following up on it.

Lex Fridman (1:06:34.680)

Yeah.

Grant Sanderson (1:06:35.680)

Like the last time we talked, I think prom, I'm not sure a promise to me that you'll have

Grant Sanderson (1:06:38.840)

music incorporated into your, like, uh, I'll share it with you a private link, but there's

Grant Sanderson (1:06:44.040)

an example of like what I had in mind.

Grant Sanderson (1:06:45.720)

I like did a version of it, um, and I'm like, Oh, I think there's a better version of this

Grant Sanderson (1:06:50.120)

that might exist one day.

Lex Fridman (1:06:52.100)

So it's now on the, like the back burner, it's like, it's sitting there.

Grant Sanderson (1:06:55.800)

It was like a live performance at this one thing, I think next, next circumstance that

Grant Sanderson (1:06:59.400)

I'm like doing another recorded live performance that like fits having that then in a better

Grant Sanderson (1:07:03.740)

recording context, maybe I'll make it nice in public.

Lex Fridman (1:07:06.040)

Maybe a while, but exactly.

Grant Sanderson (1:07:08.040)

Right.

Grant Sanderson (1:07:09.040)

Um, the point I was going to make those, like, I know I'm bad about following up on stuff,

Grant Sanderson (1:07:12.400)

uh, which is an actual problem.

Grant Sanderson (1:07:14.800)

It's born of the fact that I have a sense of what will be like good content when it

Grant Sanderson (1:07:18.440)

won't be.

Grant Sanderson (1:07:19.440)

Um, but this can actually be credibly disheartening because a ton of comments that I see are people

Grant Sanderson (1:07:24.600)

who are like, uh, frustrated, usually in a benevolent way that like I haven't followed

Lex Fridman (1:07:29.160)

through on like X and X, which I get and I should do that.

Lex Fridman (1:07:32.680)

But what's comforting thought for me is that when there's a topic I haven't promised, but

Grant Sanderson (1:07:36.400)

I am working on and I'm excited about, it's like the people who would really like this

Grant Sanderson (1:07:40.600)

don't know that it's coming and don't know to like comment to that effect and like the

Grant Sanderson (1:07:44.240)

commenting public that I'm seeing is not representative of like who I think this other project will

Grant Sanderson (1:07:49.380)

touch meaningfully.

Lex Fridman (1:07:50.380)

Yeah.

Lex Fridman (1:07:51.380)

So focus on the future on the thing you're creating now, just like the, uh, yeah, the

Lex Fridman (1:07:54.240)

art of it.

Grant Sanderson (1:07:55.240)

One of the people is really inspiring to me in that regard because I've really seen it

Grant Sanderson (1:08:00.760)

in persons, um, Joe Rogan, he doesn't read comments, but not just that he doesn't give

Grant Sanderson (1:08:08.720)

a damn.

Lex Fridman (1:08:09.720)

Hmm.

Grant Sanderson (1:08:10.720)

He like legitimate, he's not like clueless about it.

Grant Sanderson (1:08:13.960)

He's like, just like the richness and the depth of a smile he has when he just experiences

Grant Sanderson (1:08:19.800)

the moment with you like offline, you can tell he doesn't give a damn about like, like

Grant Sanderson (1:08:28.420)

about anything, about what people think about whether if it's on a podcast, you talk to

Grant Sanderson (1:08:31.760)

them or whether offline about just, it's not there.

Grant Sanderson (1:08:36.480)

Like what other people think, how, how, um, even like what the rest of the day looks like

Grant Sanderson (1:08:41.420)

is just deeply in the moment, uh, or like, especially like is, is what we're doing going

Lex Fridman (1:08:48.040)

to make for a good Instagram photo or something like that?

Grant Sanderson (1:08:50.720)

It doesn't think like that at all.

Grant Sanderson (1:08:52.880)

It's I think for actually quite a lot of people, he's an inspiration in that way, but it was

Lex Fridman (1:08:58.360)

and in real life, I show that you can be very successful, not giving a damn about, um, about

Lex Fridman (1:09:06.760)

comments.

Lex Fridman (1:09:07.760)

And it sounds, it sounds bad not to read comments cause it's like, well, there's a huge number

Lex Fridman (1:09:12.940)

of people who are deeply passionate about what you do.

Lex Fridman (1:09:15.120)

So you're what ignoring them, but at the same time, the nature of our platforms is such

Grant Sanderson (1:09:20.760)

that the cost of listening to all the positive people who are really close to you, who are

Grant Sanderson (1:09:27.200)

incredible people have been, you know, I've made a great community that you can learn

Grant Sanderson (1:09:32.240)

a lot from the cost of listening to those folks is also the cost of your psychology

Grant Sanderson (1:09:40.840)

slowly being degraded by the natural underlying toxicity of the internet.

Grant Sanderson (1:09:47.560)

Engage with a handful of people deeply rather than like as many people as you can in a shallow

Grant Sanderson (1:09:51.720)

way.

Lex Fridman (1:09:52.720)

I think that's a good lesson for social media usage.

Grant Sanderson (1:09:55.000)

Um, like platforms in general, like choose, choose just a handful of things to engage

Grant Sanderson (1:10:00.160)

with and engage with it very well in a way that you feel proud of and don't worry about

Grant Sanderson (1:10:03.720)

the rest.

Lex Fridman (1:10:04.720)

Honestly, I think the best social media platform is texting.

Grant Sanderson (1:10:09.400)

That's my favorite.

Lex Fridman (1:10:10.400)

That's my go to social media platform.

Grant Sanderson (1:10:12.080)

Well, yeah, the best social media interactions like real life, not social media, but social

Lex Fridman (1:10:17.560)

interaction.

Grant Sanderson (1:10:18.560)

Oh yeah.

Lex Fridman (1:10:19.560)

No, no, no question there.

Grant Sanderson (1:10:20.560)

I think everyone should agree with that.

Grant Sanderson (1:10:21.560)

Which sucks because, uh, it's been challenged now with the current situation and we're trying

Grant Sanderson (1:10:26.440)

to figure out what kind of platform can be created that we can do remote communication

Lex Fridman (1:10:31.560)

that still is effective.

Grant Sanderson (1:10:32.560)

It's important for education.

Lex Fridman (1:10:34.480)

It's important for just the question of education right now.

Grant Sanderson (1:10:38.040)

Yeah.

Lex Fridman (1:10:39.120)

So on that topic, uh, you've done a series of live streams called lockdown math and you

Grant Sanderson (1:10:44.560)

know, you want live, which is different than you usually do.

Lex Fridman (1:10:48.920)

Maybe one, can you talk about how that feel?

Grant Sanderson (1:10:53.160)

What's that experience like like in your own, when you look back, like, is that an effective

Lex Fridman (1:10:58.080)

way?

Lex Fridman (1:10:59.080)

Did you find a being able to teach?

Lex Fridman (1:11:01.040)

And if so, is there a lessons for this world where all of these educators are now trying

Lex Fridman (1:11:07.720)

to figure out how the heck do I teach remotely?

Lex Fridman (1:11:11.120)

For me, it was very different, as different as you can get.

Grant Sanderson (1:11:13.280)

I'm on camera, which I'm usually not.

Lex Fridman (1:11:15.080)

I'm doing it live, which is nerve wracking.

Grant Sanderson (1:11:17.240)

Um, it was a slightly different like level of topics, although realistically I'm just

Lex Fridman (1:11:21.920)

talking about things I'm interested in no matter what.

Grant Sanderson (1:11:24.360)

I think the reason I did that was this thought that a ton of people are looking to learn

Grant Sanderson (1:11:28.520)

remotely the rate at which I usually put out content is too slow to be actively helpful.

Grant Sanderson (1:11:33.160)

Let me just do some biweekly lectures that if you're looking for a place to point your

Grant Sanderson (1:11:36.880)

students, if you're a student looking for a place to be edified about math, just tune

Grant Sanderson (1:11:39.920)

in at these times.

Grant Sanderson (1:11:40.920)

Um, and in that sense, I think it was, you know, a success for those who followed with

Grant Sanderson (1:11:45.000)

it.

Lex Fridman (1:11:46.000)

It was a really rewarding experience for me to see how people engaged with it.

Grant Sanderson (1:11:50.240)

Um, part of the fun of the live interaction was to actually like I do these live quizzes

Lex Fridman (1:11:54.800)

and see how people would answer and try to shape the lesson based on that or see what

Grant Sanderson (1:11:57.820)

questions people were asking in the audience.

Grant Sanderson (1:11:59.960)

I would love to, if I did more things like that in the future, kind of tighten that feedback

Grant Sanderson (1:12:03.880)

loop even more.

Grant Sanderson (1:12:05.440)

Um, I think for, you know, you asked about like if this can be relevant to educators,

Grant Sanderson (1:12:10.960)

like 100% online teaching is basically a form of live streaming now.

Lex Fridman (1:12:15.040)

Um, and usually it happens through zoom.

Grant Sanderson (1:12:17.280)

I think if teachers view what they're doing as a kind of performance and a kind of live

Grant Sanderson (1:12:22.440)

stream performance, um, that would probably be pretty healthy because zoom can be kind

Grant Sanderson (1:12:27.660)

of awkward.

Grant Sanderson (1:12:28.660)

Um, and I brought up this little blog post actually just on like just what our setup

Grant Sanderson (1:12:32.820)

looked like if you want to adopt it yourself and how to integrate, um, like the broadcasting

Lex Fridman (1:12:37.040)

software OBS with zoom or things like that.

Grant Sanderson (1:12:39.400)

It was really sorry to pause on that.

Lex Fridman (1:12:40.880)

I mean, yeah, maybe we could look at the blog post, but it looked really nice.

Grant Sanderson (1:12:45.560)

The thing is, I knew nothing about any of that stuff before I started.

Lex Fridman (1:12:48.440)

I had a friend who knew a fair bit.

Grant Sanderson (1:12:50.320)

Um, and so he kind of helped show me the routes.

Grant Sanderson (1:12:52.640)

One of the things that I realized is that you could, as a teacher, like it doesn't take

Grant Sanderson (1:12:57.080)

that much to make things look and feel pretty professional.

Grant Sanderson (1:12:59.680)

Um, like one component of it is as soon as you hook things up with the broadcasting software,

Grant Sanderson (1:13:04.160)

rather than just doing like screen sharing, you can set up different scenes and then you

Lex Fridman (1:13:07.840)

can like have keyboard shortcuts to transition between those scenes.

Lex Fridman (1:13:11.320)

So you don't need a production studio with a director calling like, go to camera three,

Lex Fridman (1:13:14.560)

go to camera two, like onto the screen capture.

Grant Sanderson (1:13:17.040)

Instead you can have control of that.

Lex Fridman (1:13:18.820)

And it took a little bit of practice and I would mess it up now and then, but I think

Grant Sanderson (1:13:21.800)

I had it decently smooth such that, you know, I'm talking to the camera and then we're doing

Lex Fridman (1:13:25.800)

something on the paper.

Grant Sanderson (1:13:26.800)

Then we're doing like a, um, playing with a Desmos graph or something.

Lex Fridman (1:13:31.080)

And something that I think in the past would have required a production team, you can actually

Grant Sanderson (1:13:34.240)

do as a solo operation, um, and in particular as a teacher.

Lex Fridman (1:13:38.000)

And I think it's worth it to try to do that because, uh, two reasons, one, you might get

Grant Sanderson (1:13:42.760)

more engagement from the students, but the biggest reason I think one of the like best

Grant Sanderson (1:13:46.400)

things that can come out of this pandemic education wise is if we turn a bunch of teachers

Grant Sanderson (1:13:50.440)

into content creators.

Lex Fridman (1:13:51.920)

And if we take lessons that are usually done in these one off settings and like start to

Grant Sanderson (1:13:55.920)

get in the habit of, um, sometimes I'll use the phrase commoditizing explanation where

Lex Fridman (1:14:01.300)

what you want is whatever a thing a student wants to learn.

Grant Sanderson (1:14:06.680)

It just seems inefficient to me that that lesson is taught millions of times over in

Lex Fridman (1:14:11.020)

parallel across many different classrooms in the world.

Grant Sanderson (1:14:14.160)

Like year to year, you've got a given algebra one lesson that's just taught like literally

Lex Fridman (1:14:18.120)

millions of times, um, by different people.

Lex Fridman (1:14:21.640)

What should happen is that there's the small handful of explanations online, uh, that exists

Lex Fridman (1:14:27.080)

so that when someone needs that explanation, they can go to it, that the time in classroom

Grant Sanderson (1:14:30.640)

is spent on all of the parts of teaching and education that aren't explanation, which is

Lex Fridman (1:14:34.040)

most of it.

Grant Sanderson (1:14:35.400)

Right.

Grant Sanderson (1:14:36.400)

Um, and the way to get there is to basically have more people who are already explaining,

Grant Sanderson (1:14:40.720)

publish their explanations and have it in a publicized forum.

Lex Fridman (1:14:43.960)

So if during a pandemic you can have people automatically creating online content cause

Grant Sanderson (1:14:49.340)

it has to be online, but getting into the habit of doing it in a, um, in a way that

Grant Sanderson (1:14:53.400)

doesn't just feel like a zoom call that happened to be recorded, but it actually feels like

Grant Sanderson (1:14:57.800)

a, a piece that was always going to be publicized to more people than just your students that

Lex Fridman (1:15:03.680)

can be really powerful.

Lex Fridman (1:15:05.480)

And there's an improvement process there, like so being self critical and growing, like,

Grant Sanderson (1:15:11.160)

you know, like I guess YouTubers go through this process of like putting out some content

Lex Fridman (1:15:17.680)

and like nobody caring about it and then trying to figure out like, and basically improving

Lex Fridman (1:15:24.000)

figure out like, why did nobody care?

Lex Fridman (1:15:28.000)

What can I, you know, and they come up with all kinds of answers, which may or may not

Lex Fridman (1:15:31.280)

be correct, but doesn't matter because the answer leads to improvement.

Lex Fridman (1:15:35.600)

So you're being constantly self critical, self analytical, it should be better to say.

Lex Fridman (1:15:40.520)

So you think of like, how can I make the audio better?

Grant Sanderson (1:15:43.060)

Like all the basic things.

Grant Sanderson (1:15:45.680)

Maybe one, one question to ask, cause, uh, well, by way of, uh, Russ Tedrick is a robotics

Grant Sanderson (1:15:52.240)

professor at MIT, one of my favorite people, a big fan of yours.

Lex Fridman (1:15:55.920)

Uh, he watched our first conversation.

Grant Sanderson (1:15:57.720)

I just interviewed him a couple of weeks ago.

Grant Sanderson (1:16:01.320)

He, uh, he teaches this course in the under actuated robotics, which is, um, like robotic

Grant Sanderson (1:16:08.320)

systems when you can't control everything, like when you're like, we as humans, when

Lex Fridman (1:16:13.560)

we walk, we're always falling forward, which means like it's gravity.

Grant Sanderson (1:16:17.680)

You can't control it.

Lex Fridman (1:16:18.680)

You just hope you can catch yourself, but that's not all guaranteed.

Grant Sanderson (1:16:21.800)

It depends on the surface.

Lex Fridman (1:16:23.000)

So like that's under actuated.

Grant Sanderson (1:16:24.480)

You can't control everything.

Grant Sanderson (1:16:26.360)

The number of actuators, uh, the degrees of freedoms you have is not enough to fully control

Grant Sanderson (1:16:31.760)

the system.

Lex Fridman (1:16:32.760)

So I don't know.

Grant Sanderson (1:16:33.760)

It's a really, I think, beautiful, fascinating class.

Lex Fridman (1:16:35.680)

He puts it online.

Grant Sanderson (1:16:37.240)

Um, it's quite popular.

Lex Fridman (1:16:39.040)

He does an incredible job teaching.

Grant Sanderson (1:16:40.440)

He puts it online every time, but he's kind of been interested in like crisping it up,

Lex Fridman (1:16:45.240)

like, you know, making it, uh, you know, innovating in different kinds of ways.

Lex Fridman (1:16:50.480)

And he was inspired by the work you do, because I think in his work, he can do similar kinds

Grant Sanderson (1:16:56.280)

of explanations as you're doing, like revealing the beauty of it and spending like months

Grant Sanderson (1:17:01.240)

in preparing a single video.

Lex Fridman (1:17:03.360)

Uh, and he's interested in how to do that.

Grant Sanderson (1:17:06.360)

That's why he listened to the conversation.

Grant Sanderson (1:17:07.840)

He's playing with manum, but he had this question of, you know, um, of, uh, you know, like in

Grant Sanderson (1:17:16.000)

my apartment where we did the interview, I have like curtains, like the, for like a black

Grant Sanderson (1:17:21.520)

curtain, not this, uh, this is, this is a adjacent mansion that we're in that I also,

Grant Sanderson (1:17:28.800)

uh, but you basically just have, I have like a black curtain, whatever that, you know,

Grant Sanderson (1:17:33.600)

makes it really easy to set up a filming situation with cameras that we have here, these microphones.

Lex Fridman (1:17:38.080)

He was asking, you know, what kind of equipment do you recommend?

Lex Fridman (1:17:41.560)

I guess like your blog post is a good one.

Grant Sanderson (1:17:43.640)

I said, I don't recommend this is excessive and actually really hard to work with.

Lex Fridman (1:17:49.000)

So I wonder, I mean, uh, is there something you would recommend in terms of equipment?

Grant Sanderson (1:17:55.160)

Like is, is it, do you re do you think like lapel mics, like USB mics, what do you, for

Grant Sanderson (1:18:00.880)

my narration, I use a USB mic for the streams that used to lapel mic, uh, the narration,

Grant Sanderson (1:18:06.280)

it's a blue Yeti.

Grant Sanderson (1:18:07.280)

Um, I'm forgetting actually the name of the lapel mic, but it was probably like a road

Grant Sanderson (1:18:12.120)

of some kind.

Lex Fridman (1:18:13.680)

Um, but is it hard to figure out how to make the audio sound good?

Grant Sanderson (1:18:17.120)

Oh, I mean, listen to all the early videos on my channel and clearly like I'm terrible

Lex Fridman (1:18:21.600)

at this for, for some reason.

Grant Sanderson (1:18:23.640)

Um, I just couldn't get audio for awhile.

Lex Fridman (1:18:25.760)

I think I, it's weird when you hear your own voice.

Lex Fridman (1:18:28.120)

So you hear it, you're like, this sounds weird and it's hard to notice it sound weird because

Grant Sanderson (1:18:31.820)

you're not used to your own voice or they're like actual audio artifacts at play.

Grant Sanderson (1:18:36.160)

Um, so, uh, and then video is just for the lockdown, just the camera, like you said,

Grant Sanderson (1:18:43.120)

it was probably streaming somehow through the, yeah, there were two GH five cameras.

Grant Sanderson (1:18:47.260)

One that was mounted overhead over a piece of paper.

Grant Sanderson (1:18:49.520)

You could also use like an iPad or a Wacom tablet to do your writing electronically,

Lex Fridman (1:18:53.640)

but I just wanted the paper feel, um, one on the face.

Lex Fridman (1:18:57.840)

There's two.

Grant Sanderson (1:18:58.840)

Um, again, I don't know, I'm like just not actually the one to ask this cause I like

Grant Sanderson (1:19:02.400)

animate stuff usually, but, uh, each of them like has a compressor object that makes it

Grant Sanderson (1:19:08.400)

such that the camera output goes into the computer USB, but like gets compressed before

Lex Fridman (1:19:12.880)

it does that.

Lex Fridman (1:19:13.880)

The, the live aspect of it, do you, do you regret doing it live?

Lex Fridman (1:19:20.120)

Not at all.

Grant Sanderson (1:19:21.120)

Um, I think I do think the content might be like much less sharp and tight than if it

Lex Fridman (1:19:26.360)

were something, even that I just recorded like that and then edited later.

Lex Fridman (1:19:30.360)

But I do like something that I do to be out there to show like, Hey, this is what it's

Lex Fridman (1:19:34.080)

like.

Grant Sanderson (1:19:35.080)

Raw.

Lex Fridman (1:19:36.080)

This is what it's like when I make mistakes.

Grant Sanderson (1:19:37.080)

Um, this is like the pace of thinking, um, I like the live interaction of it.

Lex Fridman (1:19:41.240)

I think that made it better.

Grant Sanderson (1:19:42.760)

Uh, I probably would do it on a different channel.

Grant Sanderson (1:19:45.480)

I think, um, if I did series like that in the future, just because it's, it's a different

Grant Sanderson (1:19:49.120)

style.

Grant Sanderson (1:19:50.120)

It's probably a different target audience and, um, kind of keep clean what three blue

Lex Fridman (1:19:53.240)

and brown is about versus, uh, the benefits of like live lectures.

Lex Fridman (1:19:58.080)

Do you, uh, suggest like in this time of COVID that people like Russ or other educators tried

Grant Sanderson (1:20:04.280)

to go like the, the shorter, like 20 minute videos that are like really well planned out

Lex Fridman (1:20:12.080)

or scripted.

Grant Sanderson (1:20:13.080)

You really think through, you slowly design.

Lex Fridman (1:20:15.680)

So it's not live.

Lex Fridman (1:20:16.680)

Do you see like that being an important part of, um, what they do?

Lex Fridman (1:20:20.360)

Yeah.

Grant Sanderson (1:20:21.360)

Well, what I think teachers like Russ should do is, um, choose the small handful of topics

Lex Fridman (1:20:25.880)

that they're going to do just really well.

Grant Sanderson (1:20:27.400)

They want to create the best short explanation of it in the world that will be one of those

Lex Fridman (1:20:31.320)

handfuls in a world where you have commoditized explanation, right?

Grant Sanderson (1:20:35.160)

Most of the lectures should be done just normally.

Lex Fridman (1:20:37.440)

Um, so put thought and planning into it.

Grant Sanderson (1:20:39.040)

I'm sure he's a wonderful teacher and like knows all about that, but maybe choose those

Lex Fridman (1:20:42.920)

small handful of topics.

Grant Sanderson (1:20:44.440)

Um, do what beneficial for me sometimes is I do sample lessons with people on that topic

Lex Fridman (1:20:49.560)

to get some sense of how other people think about it.

Grant Sanderson (1:20:52.560)

Let that inform how you want to, um, edit it or script it or whatever format you want

Lex Fridman (1:20:56.840)

to do.

Grant Sanderson (1:20:57.840)

Some people are comfortable just explaining it and editing later.

Lex Fridman (1:20:59.960)

I'm more comfortable like writing it out and thinking in that setting.

Grant Sanderson (1:21:02.640)

Yeah.

Lex Fridman (1:21:03.640)

It's kind of sad.

Grant Sanderson (1:21:04.640)

Sorry to interrupt.

Lex Fridman (1:21:05.640)

Uh, it's, it's a little bit sad to me to see how much knowledge is lost.

Grant Sanderson (1:21:10.960)

Like just, just like you mentioned, there's professors, like we can take my dad, for example,

Grant Sanderson (1:21:16.200)

to blow up his ego a little bit, but he's a, he's a great teacher and he knows plasma,

Grant Sanderson (1:21:21.800)

plasma chemistry, plasma physics really well.

Lex Fridman (1:21:23.900)

So he can very simply explain some beautiful, but otherwise, uh, complicated concepts.

Lex Fridman (1:21:31.520)

And it's sad that like, if you Google plasma or like for plasma physics, like there's no

Lex Fridman (1:21:37.520)

videos.

Lex Fridman (1:21:38.520)

And just imagine if every one of those excellent teachers like your father or like Russ, um,

Grant Sanderson (1:21:43.120)

even if they just chose one topic this year, they're like, I'm going to make the best video

Grant Sanderson (1:21:46.940)

that I can on this topic.

Grant Sanderson (1:21:48.160)

If every one of the great teachers did that, the internet would be replete and it's already

Grant Sanderson (1:21:52.360)

replete with great explanations.

Lex Fridman (1:21:53.800)

But it would be even more so with all the niche, great explanations and like anything

Grant Sanderson (1:21:56.840)

you want to learn.

Grant Sanderson (1:21:57.840)

Um, and there's a self interest to it for, in terms of teachers, in terms of even, so

Grant Sanderson (1:22:02.480)

if you take Russ, for example, it's not that he's teaching something like he teaches his

Lex Fridman (1:22:08.160)

main thing, his thing he's deeply passionate about.

Lex Fridman (1:22:11.600)

And from a selfish perspective, it's also just like, I mean, it's a, it's a, it's like

Grant Sanderson (1:22:20.360)

publishing a paper in a really, uh, like nature has like letters, like accessible publication.

Grant Sanderson (1:22:27.680)

It's just going to guarantee that your work, that your passion is seen by a huge number

Lex Fridman (1:22:35.500)

of people, whatever the definition of huge is, doesn't matter.

Grant Sanderson (1:22:39.080)

It's much more than it otherwise, uh, would be.

Lex Fridman (1:22:42.400)

And it's those lectures that tell early students what to be interested in at the moment.

Grant Sanderson (1:22:47.960)

I think students are disproportionately interested in the things that are well represented on

Lex Fridman (1:22:51.360)

YouTube.

Lex Fridman (1:22:52.440)

So to any educator out there, if you're wondering, Hey, I want more like grad students in my

Lex Fridman (1:22:56.000)

department, like what's the best way to recruit grad students?

Grant Sanderson (1:22:59.000)

It's like, make the best video you can and then wait eight years.

Lex Fridman (1:23:02.040)

And then you're going to have a pile of like excellent grad students for that department.

Lex Fridman (1:23:05.560)

And one of the lessons I think your channel teaches is there's appeal of explaining just

Grant Sanderson (1:23:12.920)

something beautiful, explaining it cleanly, technically not doing a marketing video about

Lex Fridman (1:23:19.680)

why topology is great.

Lex Fridman (1:23:21.120)

There's yeah, that's the, there's people interested in this stuff.

Grant Sanderson (1:23:24.000)

I mean, uh, one of the greatest channels like Matt, it's not even a math channel, but the

Lex Fridman (1:23:29.280)

channel with greatest math content is Vsauce, like interviewed.

Grant Sanderson (1:23:33.200)

If imagine you were to propose making a video that explains the Banach Tarski paradox substantively,

Lex Fridman (1:23:38.800)

right?

Grant Sanderson (1:23:39.800)

Like not shying around it, maybe not describing things in terms of, um, like the group theoretic

Grant Sanderson (1:23:45.440)

terminology that you'd usually see in a paper, but the actual results, um, that went into

Grant Sanderson (1:23:51.920)

this idea of like breaking apart a sphere, proposing that to like a network TV station

Grant Sanderson (1:23:56.160)

saying, yeah, I'm going to, I'm going to do this in depth talk of the Banach Tarski paradox.

Grant Sanderson (1:23:59.880)

I'm pretty sure it's going to reach 20 million people.

Lex Fridman (1:24:02.120)

It's like, get out of here.

Grant Sanderson (1:24:04.000)

Like no, no one cares about that.

Lex Fridman (1:24:05.760)

No one's interested in anything even anywhere near that.

Lex Fridman (1:24:08.720)

But then you have Michael's quirky personality around it.

Lex Fridman (1:24:11.060)

And just people that are actually hungry for that kind of depth, um, then you don't need

Grant Sanderson (1:24:16.200)

like the approval of some higher network.

Lex Fridman (1:24:19.280)

You can just do it and let the people speak for themselves.

Lex Fridman (1:24:22.080)

So I think, you know, if your father was to make something on plasma physics or, um, if

Grant Sanderson (1:24:26.720)

we were to have like, uh, underactualized robotics, underactuated, underactuated, yes,

Grant Sanderson (1:24:32.680)

not underactualized, plenty actualized underactuated robotics.

Lex Fridman (1:24:37.360)

Robotics is under actualized currently.

Lex Fridman (1:24:41.520)

So even if it's things that you might think are niche, I bet you'll be surprised by how

Lex Fridman (1:24:46.100)

many people, um, actually engage with it really deeply.

Grant Sanderson (1:24:49.360)

Although I just psychologically watching him, I can't speak for a lot of people.

Lex Fridman (1:24:52.720)

I can speak for my dad.

Grant Sanderson (1:24:53.880)

I think there's a, there's a little bit of a skill gap, but I think that could be overcome.

Lex Fridman (1:25:00.200)

That's pretty basic.

Grant Sanderson (1:25:01.200)

None of us know how to make videos when we start the first stuff I made was terrible

Lex Fridman (1:25:04.160)

in a number of respects.

Grant Sanderson (1:25:05.160)

Like look at the earliest videos I need in the YouTube channel, except for captain disillusion.

Lex Fridman (1:25:09.280)

And they're all like terrible versions of whatever they are now.

Lex Fridman (1:25:13.480)

But the thing I've noticed, especially like with world experts is it's the same thing

Lex Fridman (1:25:19.480)

that I'm sure you went through, which is like, um, fear of like embarrassment.

Grant Sanderson (1:25:25.320)

Like they, they definitely, it's, it's the same reason.

Grant Sanderson (1:25:29.720)

Like I feel that anytime I put out a video, I don't know if you still feel that.

Lex Fridman (1:25:35.120)

But like, I don't know, it's this imposter syndrome.

Lex Fridman (1:25:39.360)

Like who am I to talk about this?

Lex Fridman (1:25:41.040)

And that that's true for like even things that you've studied for like your whole life.

Lex Fridman (1:25:46.320)

Uh, I don't know.

Grant Sanderson (1:25:47.400)

It's scary to post stuff on YouTube.

Lex Fridman (1:25:50.160)

It is scary.

Grant Sanderson (1:25:51.160)

Uh, I honestly wish that more of the people who had that modesty to say, who am I to

Lex Fridman (1:25:57.700)

post this?

Grant Sanderson (1:25:58.700)

We're the ones actually posting it.

Lex Fridman (1:26:00.200)

That's right.

Grant Sanderson (1:26:01.200)

I mean, the honest problem is like a lot of the educational content is posted by people

Grant Sanderson (1:26:04.680)

who like, we're just starting to research it two weeks ago and are on a certain schedule

Lex Fridman (1:26:09.680)

and who maybe should think like, who am I to explain and choose your favorite topic,

Lex Fridman (1:26:15.960)

quantum mechanics or something.

Grant Sanderson (1:26:17.720)

Um, and the people who have the self awareness, uh, to not post are probably the people also

Lex Fridman (1:26:23.760)

best positioned to give a good, honest explanation of it.

Grant Sanderson (1:26:27.400)

That's why there's a lot of value in a channel like numberphile where they basically trap

Grant Sanderson (1:26:32.480)

a really smart person and force them to explain stuff on a bronze sheet of paper.

Grant Sanderson (1:26:38.240)

So, but of course that's not scalable as a single channel.

Grant Sanderson (1:26:41.760)

If they, if there's anything beautiful that it could be done as people take it in their

Grant Sanderson (1:26:45.640)

own hands, uh, educators, which is again, circling back, I do think the pandemic will

Lex Fridman (1:26:51.220)

serve to force a lot of people's hands.

Grant Sanderson (1:26:54.560)

You're going to be making online content anyway.

Lex Fridman (1:26:56.940)

It's happening, right?

Grant Sanderson (1:26:58.940)

Just hit that publish button and see how it goes.

Lex Fridman (1:27:01.520)

Yeah.

Grant Sanderson (1:27:02.720)

See how it goes.

Lex Fridman (1:27:03.720)

The cool thing about YouTube is it might not go for a while, but like 10 years later, right?

Grant Sanderson (1:27:10.040)

Yeah.

Grant Sanderson (1:27:11.040)

It'll be like, this, the thing this, what people don't understand with YouTube, at least

Grant Sanderson (1:27:14.440)

for now, at least that's my hope with it is, uh, it's a leg.

Lex Fridman (1:27:20.000)

It's a, it's literally better than publishing a book in terms of the legacy.

Grant Sanderson (1:27:24.800)

It's it will live for a long, long time.

Grant Sanderson (1:27:27.800)

Of course it's, um, one of the things I mentioned Joe Rogan before, it's kinda, there's a sad

Grant Sanderson (1:27:34.480)

thing cause I'm a fan.

Lex Fridman (1:27:36.640)

He's moving to Spotify.

Grant Sanderson (1:27:38.240)

Yeah.

Lex Fridman (1:27:39.240)

Yeah.

Grant Sanderson (1:27:40.240)

Nine digit numbers will do that to you.

Lex Fridman (1:27:41.800)

Yeah.

Lex Fridman (1:27:42.800)

But he doesn't really that he was one of the person that doesn't actually care that much

Lex Fridman (1:27:46.040)

about money.

Grant Sanderson (1:27:47.040)

Like having talked to him here, it wasn't because of money.

Lex Fridman (1:27:50.900)

It's because he legitimately thinks that they're going to do like a better job.

Grant Sanderson (1:27:58.880)

Like, so they're, so from his perspective, YouTube, you have to understand where they're

Lex Fridman (1:28:03.040)

coming from.

Grant Sanderson (1:28:04.040)

YouTube has been cracking down on people who they, you know, Joe Rogan talks to Alex Jones

Lex Fridman (1:28:10.880)

and conspiracy theories and stuff.

Lex Fridman (1:28:13.120)

And YouTube is really like careful that kind of stuff.

Lex Fridman (1:28:16.480)

And that's not a good feeling.

Grant Sanderson (1:28:18.400)

Like, and Joe didn't, doesn't feel like YouTube was on his side.

Grant Sanderson (1:28:22.760)

You know, he's often has videos that they don't put in trending that like are obviously

Grant Sanderson (1:28:28.680)

should be in trending because they're nervous about like, you know, if this concert is this,

Grant Sanderson (1:28:34.720)

is this content going to, you know, upset people that all that kind of stuff have misinformation.

Lex Fridman (1:28:41.520)

And that's not a good place for a person to be in.

Lex Fridman (1:28:44.320)

And Spotify is giving them a, we're never going to censor you.

Grant Sanderson (1:28:48.480)

We're never going to do that.

Lex Fridman (1:28:50.360)

But the reason I bring that up, whatever you think about that, I personally think as bullshit

Grant Sanderson (1:28:55.120)

because podcasting should be free and not constrained to a platform.

Lex Fridman (1:28:59.640)

It's pirate radio.

Lex Fridman (1:29:00.640)

What the hell?

Grant Sanderson (1:29:01.640)

You can't, as much as I love Spotify, you can't just, you can't put fences around it.

Lex Fridman (1:29:08.360)

But anyway, the reason I bring that up is Joe's going to remove his entire library from

Lex Fridman (1:29:13.480)

YouTube.

Lex Fridman (1:29:14.480)

Whoa, really?

Lex Fridman (1:29:15.480)

I didn't know that.

Grant Sanderson (1:29:16.480)

His full length, the clips are going to stay, but the full length videos are all, I mean,

Lex Fridman (1:29:20.640)

made private or deleted.

Grant Sanderson (1:29:22.320)

That's part of the deal.

Lex Fridman (1:29:23.660)

And like, that's the first time where I was like, Oh, YouTube videos might not live forever.

Grant Sanderson (1:29:29.200)

Like things you find like, okay, I'm sorry.

Grant Sanderson (1:29:32.640)

This is why you need an IPFS or something where it's like, if there's a content link,

Lex Fridman (1:29:36.960)

are you familiar with this system at all?

Lex Fridman (1:29:39.300)

Like right now, if you have a URL, it points to a server.

Grant Sanderson (1:29:41.920)

There's like a system where the address points to content and then it's like distributed.

Lex Fridman (1:29:46.240)

So you, you can't actually delete what's at an address because it's, it's content addressed.

Lex Fridman (1:29:50.920)

And as long as there's someone on the network who hosts it, it's always accessible at the

Lex Fridman (1:29:54.500)

address that it once was.

Lex Fridman (1:29:56.860)

But I mean, that raises a question.

Lex Fridman (1:29:58.600)

I'm not going to put you on the spot, but like somebody like Vsauce, right?

Grant Sanderson (1:30:03.420)

Spotify comes along and gives him, let's say $100 billion.

Lex Fridman (1:30:07.800)

Okay.

Lex Fridman (1:30:08.800)

Let's say some crazy number and then removes it from YouTube, right?

Lex Fridman (1:30:13.160)

It's made me, I don't know, for some reason I thought YouTube is forever.

Grant Sanderson (1:30:19.840)

I don't think it will be.

Grant Sanderson (1:30:20.840)

I mean, you know, another variant that this might take is like, uh, that, you know, um,

Grant Sanderson (1:30:25.560)

you fast forward 50 years and, uh, you know, Google or Alphabet isn't the company that

Lex Fridman (1:30:30.360)

it once was.

Lex Fridman (1:30:31.360)

And it's kind of struggling to make ends meet.

Lex Fridman (1:30:33.240)

And you know, it's been supplanted by the whoever wins on the AR game or whatever it

Grant Sanderson (1:30:38.360)

might be.

Lex Fridman (1:30:39.640)

And then they're like, you know, all of these videos that we're hosting are pretty costly.

Lex Fridman (1:30:43.880)

So we're just, we're going to start deleting the ones that aren't watched that much and

Lex Fridman (1:30:47.680)

tell people to like try to back them up on their own or whatever it is.

Grant Sanderson (1:30:51.560)

Um, or even if it does exist in some form forever, it's like if people are, um, not

Grant Sanderson (1:30:56.520)

habituated to watching YouTube in 50 years, they're watching something else, which seems

Grant Sanderson (1:30:59.600)

pretty likely.

Grant Sanderson (1:31:00.600)

Like it would be shocking if YouTube remained as popular as it is now indefinitely into

Grant Sanderson (1:31:06.040)

the future.

Lex Fridman (1:31:07.520)

So, uh, it won't be forever.

Grant Sanderson (1:31:10.160)

Makes me sad still, but cause it's such a nice, it's just like you said of the canonical

Lex Fridman (1:31:16.160)

videos.

Grant Sanderson (1:31:17.160)

Sorry.

Lex Fridman (1:31:18.160)

I didn't mean to interrupt.

Grant Sanderson (1:31:19.160)

You know, you should get Juan Bennett on the, uh, on the thing and then talk to him about

Lex Fridman (1:31:21.800)

permanence.

Grant Sanderson (1:31:22.800)

I think you would have a good conversation.

Lex Fridman (1:31:24.400)

Who's that?

Lex Fridman (1:31:25.400)

So he's the one that founded this thing called IPFS that I'm talking about.

Lex Fridman (1:31:28.880)

And if you have him talk about basically what you're describing, like, Oh, it's sad that

Grant Sanderson (1:31:32.560)

this isn't forever.

Grant Sanderson (1:31:33.560)

Then you'll get some articulate pontification around it that's like been pretty well thought

Grant Sanderson (1:31:38.160)

through.

Grant Sanderson (1:31:39.160)

Uh, but yeah, I do see YouTube, just like you said, as a, as a place, like what your

Grant Sanderson (1:31:44.040)

channel creates, which is like a set of canonical videos on a topic.

Grant Sanderson (1:31:47.880)

Now others could create videos on that topic as well, but as a collection, it creates a

Grant Sanderson (1:31:54.560)

nice set of places to go.

Grant Sanderson (1:31:56.480)

Uh, if you're curious about a particular topic and it seems like coronavirus is a nice opportunity

Grant Sanderson (1:32:02.320)

to, uh, put that knowledge out there in the world at, uh, MIT and beyond, I have to talk

Lex Fridman (1:32:10.680)

to you a little bit about machine learning, deep learning and so on.

Grant Sanderson (1:32:13.440)

Again, we talked about last time you have a set of beautiful videos on neural networks.

Grant Sanderson (1:32:19.560)

Uh, let me ask you first, what is the most beautiful aspect of neural networks and machine

Lex Fridman (1:32:28.160)

learning to you, like for making those videos from watching how the field is evolving?

Lex Fridman (1:32:35.320)

Is there something mathematically or in applied sense, just beautiful to you about them?

Grant Sanderson (1:32:42.560)

Well, I think what I would go to is the layered structure and how, um, you can have what feel

Grant Sanderson (1:32:48.320)

like qualitatively distinct things happening, going from one layer to another, but that

Grant Sanderson (1:32:52.960)

are, um, following the same mathematical rule because you look at it as a piece of math.

Lex Fridman (1:32:56.960)

It's like you got a non linearity and then you've got a matrix multiplication.

Grant Sanderson (1:33:00.720)

That's what's happening on all the layers.

Grant Sanderson (1:33:02.400)

Um, but especially if you look at like some of the visualizations that, uh, like Chris

Grant Sanderson (1:33:06.880)

Ola has done with respect to, um, like convolutional nets that have been trained on image net trying

Lex Fridman (1:33:12.640)

to say, what does this neuron do?

Lex Fridman (1:33:14.120)

What do this, uh, does this family of neurons do?

Lex Fridman (1:33:17.440)

What you can see is that, um, the ones closer to the input side are picking up on very low

Lex Fridman (1:33:22.260)

level ideas like the texture, right?

Lex Fridman (1:33:24.720)

And then as you get further back, you have higher level ideas.

Lex Fridman (1:33:26.920)

Like what is the, where are the eyes in this picture?

Lex Fridman (1:33:29.200)

And then how do the eyes form like an animal is this animal, a cat or a dog or a deer.

Grant Sanderson (1:33:33.760)

You have this series of qualitatively different things happening, even though it's the same

Lex Fridman (1:33:37.800)

piece of math on each one.

Lex Fridman (1:33:39.680)

So that's a pretty beautiful idea that you can have like a generalizable object that,

Grant Sanderson (1:33:44.760)

um, runs through the layers of abstraction, which in some sense constitute intelligence

Grant Sanderson (1:33:50.220)

is having, um, those many different layers of an understanding to something form abstractions

Lex Fridman (1:33:55.720)

in a automated way.

Grant Sanderson (1:33:57.640)

Exactly.

Lex Fridman (1:33:58.640)

It's automated abstracting, which, I mean, that just feels very powerful.

Grant Sanderson (1:34:02.160)

Um, and the idea that it can be so simply mathematically represented.

Grant Sanderson (1:34:06.200)

I mean, a ton of like modern ML research seems a little bit like you do a bunch of ad hoc

Grant Sanderson (1:34:10.360)

things, then you decide which one worked and then you retrospectively come up with the

Lex Fridman (1:34:14.000)

mathematical reason that it always had to work.

Grant Sanderson (1:34:16.240)

Um, but you know, who cares how you came to it when you have like that elegant piece of

Lex Fridman (1:34:19.820)

math?

Grant Sanderson (1:34:20.820)

Uh, it's hard not to just smile seeing it work in action.

Grant Sanderson (1:34:24.440)

Well, and when you talked about topology before, one of the really interesting things is, is

Grant Sanderson (1:34:30.880)

beginning to be investigated under kind of the field of like science and deep learning,

Grant Sanderson (1:34:34.560)

which is like the craziness of the surface that, uh, is trying to be optimized, uh, in

Grant Sanderson (1:34:42.920)

neural networks.

Grant Sanderson (1:34:43.920)

I mean, the, the amount of local minima, local optima there is in these surfaces and somehow

Grant Sanderson (1:34:51.520)

a dumb gradient descent algorithm was able to find really good solutions.

Lex Fridman (1:34:55.640)

That's like, that's really surprising.

Grant Sanderson (1:34:58.280)

Well, so on the one hand it is, but also it's like not, it's not terribly surprising that

Grant Sanderson (1:35:04.200)

you have these interesting points that exist when you make your space so high dimensional,

Lex Fridman (1:35:08.880)

like GPT three, what did it have?

Lex Fridman (1:35:10.880)

175 billion parameters.

Lex Fridman (1:35:12.640)

So it doesn't feel as mesmerizing to think about, Oh, there's some surface of intelligent

Lex Fridman (1:35:19.900)

behavior in this crazy high dimensional space.

Grant Sanderson (1:35:21.760)

It's like, there's so many parameters that of course, but what's more interesting is

Grant Sanderson (1:35:24.840)

like, how, how is it that you're able to efficiently get there, which is maybe what you're describing

Grant Sanderson (1:35:28.940)

that something as dumb as gradient descent does it, but like the re the reason that gradient

Grant Sanderson (1:35:35.120)

descent works well with neural networks and not just, you know, choose however you want

Grant Sanderson (1:35:38.920)

to parameterize this space and then like apply gradient descent to it is that that layered

Grant Sanderson (1:35:42.640)

structure lets you decompose the derivative in a way that makes it computationally feasible.

Grant Sanderson (1:35:47.160)

Um, yeah, it's just that, that there's so many good solutions, probably infinitely infinitely

Lex Fridman (1:35:54.160)

many good solutions, not best solutions, but good solutions.

Grant Sanderson (1:35:58.780)

That's that's what's interesting.

Grant Sanderson (1:36:00.840)

It's similar to, uh, Steven Wolfram has this idea of like the, if you just look at all

Grant Sanderson (1:36:07.120)

space of computations of all space of basically algorithms that you'd be surprised how many

Lex Fridman (1:36:13.040)

of them are actually intelligent.

Grant Sanderson (1:36:15.760)

Like if you just randomly pick from the bucket, uh, that's surprising.

Grant Sanderson (1:36:19.600)

We tend to think like a tiny, tiny minority of them would be intelligent, but his sense

Grant Sanderson (1:36:26.400)

is like, it seems weirdly easy to find computations that do something interesting.

Grant Sanderson (1:36:32.760)

Well, okay, so that from like a calm agor, calm agor of complexity standpoint, almost

Grant Sanderson (1:36:38.680)

everything will be interesting.

Grant Sanderson (1:36:40.080)

What's fascinating is to find the stuff that's describable with low information, but still

Grant Sanderson (1:36:44.160)

does interesting things.

Grant Sanderson (1:36:45.640)

Uh, like one fun example of this, you know, um, Shannon's noisy coding and theorem, uh,

Grant Sanderson (1:36:51.080)

noisy coding theorem and, uh, information theory that basically says if, you know, I

Grant Sanderson (1:36:55.400)

want to send some bits to you, um, maybe, uh, some of them are going to get flipped.

Grant Sanderson (1:36:59.560)

Uh, there's some noise along the channel.

Lex Fridman (1:37:01.720)

I can come up with some way of coding it.

Grant Sanderson (1:37:04.320)

That's resilient to that noise.

Lex Fridman (1:37:06.040)

That's very good.

Grant Sanderson (1:37:07.280)

Um, and then he quantitatively describes what very good is.

Grant Sanderson (1:37:10.400)

What's funny about how he proves the existence of good error correction codes is rather than

Grant Sanderson (1:37:15.400)

saying like, here's how to construct it or even like a sensible nonconstructive proof.

Grant Sanderson (1:37:20.040)

The nature of his nonconstructive proof is to say, um, if we chose a random encoding,

Grant Sanderson (1:37:25.260)

it would be almost at the limit, which is weird because then it took decades for people

Lex Fridman (1:37:30.040)

to actually find any that were anywhere close to the limit.

Lex Fridman (1:37:33.000)

And what his proof was saying is choose a random one.

Lex Fridman (1:37:35.880)

And it's like the best kind of encoding you'll ever find.

Lex Fridman (1:37:39.160)

But what's what that tells us is that sometimes when you choose a random element from this

Grant Sanderson (1:37:44.600)

ungodly huge set, that's a very different task from finding an efficient way to actively

Grant Sanderson (1:37:49.120)

describe it.

Grant Sanderson (1:37:50.120)

Cause in that case, the random element to actually implement it as a bit of code, you

Grant Sanderson (1:37:52.920)

would just have this huge table of like, um, telling you how to encode one thing into another.

Lex Fridman (1:37:58.200)

That's totally computationally infeasible.

Lex Fridman (1:38:00.520)

So on the side of like how many possible programs are interesting in some way, it's like, yeah,

Lex Fridman (1:38:06.200)

tons of them.

Lex Fridman (1:38:07.200)

But the much, much more delicate question is when you can have a low information description

Lex Fridman (1:38:11.920)

of something that still becomes interesting.

Lex Fridman (1:38:14.720)

And thereby this kind of gives you a blueprint for how to engineer that kind of thing.

Lex Fridman (1:38:18.400)

Right.

Grant Sanderson (1:38:19.400)

Yeah.

Grant Sanderson (1:38:20.400)

Chaos theory is another good instance there where it's like, yeah, a ton of things are

Grant Sanderson (1:38:22.800)

hard to describe, but how do you have ones that have a simple set of governing equations

Lex Fridman (1:38:27.040)

that remain like arbitrarily hard to describe?

Grant Sanderson (1:38:30.160)

Well, let me ask you, uh, you mentioned GPT three.

Grant Sanderson (1:38:33.680)

It's interesting to ask, uh, what are your thoughts about the recently released open

Grant Sanderson (1:38:40.000)

AI GPT three model that I believe is already trying to learn how to communicate like Grant

Lex Fridman (1:38:46.640)

Sanderson?

Grant Sanderson (1:38:47.640)

You know, I think I got an email a day or two ago about someone who wanted to, um, try

Grant Sanderson (1:38:51.480)

to use GPT three with manum where you would like give it a high level description of something

Lex Fridman (1:38:57.280)

and then it'll like automatically create the mathematical animation, like trying to put

Lex Fridman (1:39:01.120)

me out of a job here.

Grant Sanderson (1:39:03.120)

I mean, it probably won't put you out of a job, but it'll create something visually beautiful

Lex Fridman (1:39:08.920)

for sure.

Grant Sanderson (1:39:09.920)

I would be surprised if that worked as stated, but maybe there's like variants of it like

Lex Fridman (1:39:15.040)

that you can get to.

Grant Sanderson (1:39:16.040)

Um, I mean like a lot of those demos, it's interesting.

Grant Sanderson (1:39:18.600)

I think, uh, there's a lot of failed experiments, like depending on how you prime the thing,

Grant Sanderson (1:39:26.360)

you're going to have a lot of failed, I'm certainly with code and program synthesis.

Grant Sanderson (1:39:30.280)

Most of it won't even run, but eventually I think if you, if you're, if you pick the

Grant Sanderson (1:39:35.480)

right examples, you'll be able to generate something cool.

Lex Fridman (1:39:38.840)

And I think that even that's good enough, even though if it's, if it's, if you're being

Grant Sanderson (1:39:42.280)

very selective, it's still cool that something can be generated.

Lex Fridman (1:39:46.240)

Yeah.

Grant Sanderson (1:39:47.240)

That's a huge value.

Lex Fridman (1:39:48.720)

Um, I mean, think of the writing process.

Grant Sanderson (1:39:50.360)

Sometimes a big part of it is just getting a bunch of stuff on the page and then you

Lex Fridman (1:39:52.960)

can decide what to whittle down to.

Lex Fridman (1:39:54.920)

So if it can be used in like a man machine symbiosis where it's just giving you a spew

Grant Sanderson (1:39:59.640)

of potential ideas that then you can refine down, um, like it's serving as the generator

Lex Fridman (1:40:05.080)

and then the human serves as the refiner.

Lex Fridman (1:40:07.320)

That seems like a pretty powerful dynamic.

Grant Sanderson (1:40:09.280)

Yeah.

Lex Fridman (1:40:10.280)

Have you, uh, have you gotten a chance to see any of the demos like on Twitter?

Lex Fridman (1:40:14.120)

Is there a favorite you've seen or?

Lex Fridman (1:40:15.800)

Oh, my absolute favorite.

Grant Sanderson (1:40:17.440)

Yeah.

Grant Sanderson (1:40:18.440)

Uh, so Tim Blay who runs a channel called acapella science, he was like tweeting a bunch

Grant Sanderson (1:40:23.000)

about playing with it.

Lex Fridman (1:40:24.760)

Um, and so he, so GPT three was trained on the internet from before COVID.

Lex Fridman (1:40:30.680)

So in a sense it doesn't know about the Corona virus.

Lex Fridman (1:40:33.660)

So what he seeded it with was just a short description about like, um, a novel virus,

Grant Sanderson (1:40:37.800)

uh, emerges in Wuhan, China and starts to spread around the globe.

Lex Fridman (1:40:41.580)

What follows is a month by month description of what happens, January, colon, right?

Grant Sanderson (1:40:46.500)

That's what he sees it with.

Lex Fridman (1:40:47.500)

So then what GPT three generates is like January, then a paragraph of description, February

Lex Fridman (1:40:51.440)

and such.

Lex Fridman (1:40:52.640)

And it's the funniest thing you'll ever read because, um, it predicts a zombie apocalypse,

Grant Sanderson (1:40:58.240)

which of course it would because it's trained on like the internet, the stories, but what

Lex Fridman (1:41:02.960)

you see unfolding is a description of COVID 19 if it were a zombie apocalypse.

Lex Fridman (1:41:08.240)

And like the early aspects of it are kind of shockingly in line with what's reasonable

Lex Fridman (1:41:12.680)

and then it gets out of hand so quickly.

Lex Fridman (1:41:14.880)

And the other flip side of that is, uh, I wouldn't be surprised if it's onto something

Grant Sanderson (1:41:19.600)

at some point here when, you know, 2020 has been full of surprises, who knows, like we

Grant Sanderson (1:41:25.360)

might all be in like this crazy militarized zone as it predicts just a couple of months

Lex Fridman (1:41:30.880)

off.

Grant Sanderson (1:41:31.880)

Yeah.

Lex Fridman (1:41:32.880)

I think there's definitely an interesting tool of storytelling.

Grant Sanderson (1:41:36.220)

It has struggled with mathematics, which is interesting, or in just even numbers, it's

Grant Sanderson (1:41:40.800)

able to, it's not able to generate like patterns, you know, like you give it, um, in like five

Grant Sanderson (1:41:49.280)

digit numbers and it's not able to figure out the sequence, you know, or like, um, I

Grant Sanderson (1:41:55.560)

didn't look in too much, but I'm talking about like sequences, like the Fibonacci numbers

Lex Fridman (1:42:00.000)

and to see how far it can go because obviously it's leveraging stuff from the internet and

Grant Sanderson (1:42:04.480)

it starts to lose it, but it is also cool that I've seen it able to generate some interesting

Grant Sanderson (1:42:09.400)

patterns, um, that are mathematically correct.

Lex Fridman (1:42:12.120)

Yeah.

Grant Sanderson (1:42:13.120)

I honestly haven't dug into like what's going on within it, uh, in a way that I can speak

Grant Sanderson (1:42:18.440)

intelligently to, I guess it doesn't surprise me that it's bad at numerical patterns because

Grant Sanderson (1:42:24.040)

I mean, maybe I should be more impressed with it, but like that requires having, um, a weird

Lex Fridman (1:42:30.960)

combination of intuitive and, uh, and formulaic worldview.

Lex Fridman (1:42:35.560)

So you're not just going off of intuition.

Grant Sanderson (1:42:37.320)

When you see Fibonacci numbers, you're not saying like intuitively, what do I think will

Lex Fridman (1:42:39.940)

follow the 13?

Lex Fridman (1:42:40.940)

Like I've seen patterns a lot where like 13s are followed by 21s instead.

Grant Sanderson (1:42:45.640)

It's the, like the way you're starting to see a shape of things is by knowing what hypotheses

Grant Sanderson (1:42:50.480)

to test where you're saying, oh, maybe it's generated based on the previous terms or maybe

Grant Sanderson (1:42:54.480)

it's generated based on like multiplying by a constant or whatever it is you like have

Grant Sanderson (1:42:58.000)

a bunch of different hypotheses and your intuitions are around those hypotheses, but you still

Grant Sanderson (1:43:01.760)

need to actively test it.

Grant Sanderson (1:43:04.240)

Um, and it seems like GPT three is extremely good at, um, like that sort of pattern matching

Grant Sanderson (1:43:10.100)

recognition that usually is very hard for computers.

Grant Sanderson (1:43:13.360)

That is what humans get good at through expertise and exposure to lots of things.

Grant Sanderson (1:43:17.520)

It's why it's good to learn from as many examples as you can rather than just from the definitions

Grant Sanderson (1:43:21.920)

it's to get that level of intuition, but to actually concretize it into a piece of math,

Grant Sanderson (1:43:27.640)

you do need to, um, like test your hypotheses and if not prove it, um, like have an actual

Lex Fridman (1:43:33.600)

explanation for what's going on, not just a, uh, a pattern that you've seen.

Grant Sanderson (1:43:37.800)

Yeah.

Grant Sanderson (1:43:38.800)

And, but then the flip side to play devil's advocate, that's a very kind of probably correct

Grant Sanderson (1:43:43.920)

intuitive understanding of just like we said, a few, a few layers creating abstractions,

Lex Fridman (1:43:49.400)

but it's been able to form something that looks like, uh, a compression of the data

Grant Sanderson (1:43:58.080)

that it's seen that looks awfully a lot like it understands what the heck it's talking

Lex Fridman (1:44:02.240)

about.

Grant Sanderson (1:44:03.240)

Well, I think a lot of understanding is like, I don't mean to denigrate pattern recognition.

Grant Sanderson (1:44:08.080)

Pattern recognition is most of understanding and it's super important and it's super hard.

Grant Sanderson (1:44:12.320)

Um, and so like when it's demonstrating this kind of real understanding, compressing down

Lex Fridman (1:44:16.640)

some data, like that, that might be pattern recognition at its finest.

Grant Sanderson (1:44:20.720)

My only point would be that like what differentiates math, I think to a large extent is that, um,

Grant Sanderson (1:44:27.480)

the pattern recognition isn't sufficient and that the kind of patterns that you're recognizing

Grant Sanderson (1:44:32.040)

are not like the end goals, but instead they're, they are the little bits and paths that get

Lex Fridman (1:44:37.400)

you to the end goal.

Grant Sanderson (1:44:39.520)

That's certainly true for mathematics in general.

Grant Sanderson (1:44:41.560)

It's an interesting question if that might, uh, for certain kinds of series of numbers,

Grant Sanderson (1:44:47.240)

it might not be true.

Grant Sanderson (1:44:48.360)

Like you might, um, because that's a basic, you know, like Taylor's like certain kinds

Grant Sanderson (1:44:53.780)

of series, it feels like compressing the internet, uh, is, is enough to figure out because those

Lex Fridman (1:45:01.920)

patterns in some form appear in the text somewhere.

Grant Sanderson (1:45:05.360)

Yeah.

Grant Sanderson (1:45:06.360)

Well, I mean, there's, uh, there's all sorts of wonderful examples of false patterns in

Grant Sanderson (1:45:09.120)

math where, um, one of the earliest videos I put on the channel was talking about the

Lex Fridman (1:45:13.320)

extent of dividing a circle up using these chords.

Lex Fridman (1:45:15.640)

And you see this pattern of one, two, four, eight, 16, I was like, okay, pretty easy to

Lex Fridman (1:45:20.180)

see what that pattern is.

Grant Sanderson (1:45:21.180)

It's powers of two.

Lex Fridman (1:45:22.180)

You've seen it a million times.

Grant Sanderson (1:45:23.680)

Um, but it's not powers of two.

Lex Fridman (1:45:25.800)

The next term is 31.

Lex Fridman (1:45:27.760)

And so it's like almost a power of two, but it's a little bit shy.

Lex Fridman (1:45:30.840)

And there's, there's actually a very good explanation for what's going on.

Grant Sanderson (1:45:33.640)

Um, but I think it's a good test of whether you're thinking clearly about mechanistic

Grant Sanderson (1:45:40.400)

explanations of things, how quickly you jump to thinking it must be powers of two because

Grant Sanderson (1:45:44.520)

the problem itself, there's really no, no good way to, I mean, there can't be a good

Grant Sanderson (1:45:49.560)

way to think about it as like doubling a set because ultimately it doesn't, but even before

Grant Sanderson (1:45:53.920)

it starts to, it's not something that screams out as being a doubling phenomenon.

Lex Fridman (1:45:58.520)

So at best, if it did turn out to be powers of two, it would have only been so very subtly.

Lex Fridman (1:46:03.000)

And I think the difference between like, you know, a math student making the mistake and

Grant Sanderson (1:46:06.240)

a mathematician who's experienced seeing that kind of pattern is that they, they'll have

Grant Sanderson (1:46:10.280)

a sense from what the problem itself is, whether the pattern that they're observing is reasonable

Lex Fridman (1:46:15.440)

and how to test it.

Lex Fridman (1:46:16.940)

And like, uh, I w I would just be very impressed if there was any algorithm that, um, was actively

Lex Fridman (1:46:23.440)

accomplishing that goal.

Grant Sanderson (1:46:24.920)

Yeah.

Lex Fridman (1:46:25.920)

Like a learning base algorithm.

Grant Sanderson (1:46:26.920)

Yeah.

Lex Fridman (1:46:27.920)

Like a little scientist, I guess.

Grant Sanderson (1:46:29.920)

Basically.

Lex Fridman (1:46:30.920)

Yeah.

Grant Sanderson (1:46:31.920)

That's a fascinating thought because GPT three, these language models are already accomplishing

Lex Fridman (1:46:36.840)

way more than I've expected.

Lex Fridman (1:46:38.600)

So I'm learning not to doubt, but we'll get there.

Lex Fridman (1:46:42.840)

Yeah.

Grant Sanderson (1:46:43.840)

I, I, I'm not saying I'd be impressed, but like surprised, like I'll be impressed, but

Lex Fridman (1:46:46.840)

I think we'll get there on, um, algorithms doing math like that.

Lex Fridman (1:46:52.680)

So one of the amazing things you've done for the world is to some degree, open sourcing

Grant Sanderson (1:47:00.520)

the tooling that you use to make your videos with Madam, uh, this Python library.

Grant Sanderson (1:47:08.000)

Now it's quickly evolving because I think you're inventing new things every time you

Lex Fridman (1:47:11.480)

make a video.

Grant Sanderson (1:47:12.480)

In fact, I wanted, um, I've been working on playing around with something.

Lex Fridman (1:47:17.600)

I wanted to do like an ode to three blue on Brown.

Grant Sanderson (1:47:20.160)

Like I love playing Hendrix.

Grant Sanderson (1:47:22.200)

I wanted to do like a cover, you know, of a concept I wanted to visualize and use Madam.

Lex Fridman (1:47:27.440)

And I saw that you had like a little piece of code on like Mobia strip and I tried to

Grant Sanderson (1:47:31.920)

do some cool things with spinning a Mobia strip, like continue, um, twisting it, I guess

Grant Sanderson (1:47:39.240)

is the term, uh, and it was easier to, uh, it was tough.

Lex Fridman (1:47:44.640)

So I haven't figured it out yet.

Grant Sanderson (1:47:45.640)

Well, so I guess the question I want to ask is so many people love it, uh, that you've

Lex Fridman (1:47:50.760)

put that out there.

Grant Sanderson (1:47:51.760)

They want to, uh, do the same thing as I do with Hendrix and want to cover it.

Lex Fridman (1:47:54.920)

They want to explain an idea using the tool, including Russ.

Lex Fridman (1:47:58.360)

How would you recommend they try to, I'm very sorry.

Grant Sanderson (1:48:02.420)

They try to go, they try to go by, uh, about it and what kind of choices should they choose

Lex Fridman (1:48:11.280)

to be most effective?

Lex Fridman (1:48:13.360)

That I can answer.

Lex Fridman (1:48:14.360)

So I always feel guilty if this comes up because, um, I think of it like this scrappy tool.

Lex Fridman (1:48:19.540)

It's like a math teacher who put together some code.

Grant Sanderson (1:48:22.320)

People asked what it was, so they made it open source and they kept scrapping it together.

Lex Fridman (1:48:26.240)

And there's a lot, like a lot of things about it that make it harder to work with than it

Grant Sanderson (1:48:29.400)

needs to be that are a function of like me not being a software engineer.

Grant Sanderson (1:48:33.040)

Um, I, I've, I've put some work this year trying to like make it better and more flexible.

Grant Sanderson (1:48:39.160)

Um, that is still just kind of like a work in process.

Grant Sanderson (1:48:43.360)

Um, one thing I would love to do is just get my act together about properly integrating

Grant Sanderson (1:48:48.280)

with what like the community wants to work with and like what stuff I work on and making

Grant Sanderson (1:48:53.420)

that, um, not like deviate, uh, and just like actually fostering that community in a way

Grant Sanderson (1:48:58.520)

that I've, I've been like shamefully neglectful of.

Lex Fridman (1:49:01.160)

So I'm just always guilty if it comes up.

Lex Fridman (1:49:03.220)

So let's put that guilt aside, just kind of Zen, like I'll pretend like it isn't terrible

Lex Fridman (1:49:08.080)

for someone like Russ.

Grant Sanderson (1:49:09.400)

Um, I think step one is like, make sure that what you're animating should be done so programmatically

Lex Fridman (1:49:14.240)

because a lot of things maybe shouldn't.

Grant Sanderson (1:49:16.320)

Um, like if you're just making a quick graph of something, uh, if it's a graphical intuition

Grant Sanderson (1:49:20.960)

that maybe has a little motion to it, use Desmos, use grapher, use GeoGebra, use Mathematica,

Grant Sanderson (1:49:26.840)

certain things that are like really oriented around graph.

Lex Fridman (1:49:28.640)

GeoGebra is kind of cool.

Grant Sanderson (1:49:29.640)

I did super amazing.

Lex Fridman (1:49:31.240)

You can get very, very far with it.

Grant Sanderson (1:49:33.280)

Um, and in a lot of ways, like it would make more sense for STEM stuff that I do to just

Grant Sanderson (1:49:37.680)

do in GeoGebra, but I kind of have this cycle of liking to try to improve man and by doing

Grant Sanderson (1:49:42.080)

videos and such.

Lex Fridman (1:49:43.080)

So, uh, do as I say, not as I do.

Grant Sanderson (1:49:45.760)

The original like thought I had in making manum was that there's so many different ways

Grant Sanderson (1:49:49.640)

of representing functions other than graphs, um, in particular things like transformations,

Grant Sanderson (1:49:55.000)

like use movement over time to communicate relationships between inputs and outputs instead

Grant Sanderson (1:49:59.680)

of like estimate direction and Y direction, um, or like vector fields or things like that.

Lex Fridman (1:50:04.520)

So I wanted something that was flexible enough that you didn't feel constrained into a graphical

Lex Fridman (1:50:08.280)

environment.

Grant Sanderson (1:50:09.280)

Um, by graphical, I mean like graphs with like X coordinate, Y coordinate kind of stuff,

Lex Fridman (1:50:15.520)

but also make sure that, um, you're taking advantage of the fact that it's programmatic.

Grant Sanderson (1:50:20.720)

You have loops, you have conditionals, you have abstraction.

Grant Sanderson (1:50:23.600)

If any of those are like well fit for what you want to teach to, you know, have a scene

Grant Sanderson (1:50:27.560)

type that you tweak a little bit based on parameters or to have conditional so that

Grant Sanderson (1:50:31.500)

things can go one way or another or loops so that you can create these things of like

Grant Sanderson (1:50:34.880)

arbitrarily increasing complexity.

Lex Fridman (1:50:37.000)

That's the stuff that's like meant to be animated programmatically.

Grant Sanderson (1:50:39.680)

If it's just like writing some text on the screen or shifting around objects or something

Lex Fridman (1:50:43.720)

like that, um, things like that, you should probably just use keynote, right?

Grant Sanderson (1:50:48.280)

Um, you'd be a lot simpler.

Grant Sanderson (1:50:50.120)

So, uh, try to find a workflow that distills down that which should be programmatic into

Grant Sanderson (1:50:55.100)

manum and that which doesn't need to be into like other domains.

Lex Fridman (1:50:58.280)

Again, do as I say, not as I do.

Grant Sanderson (1:51:01.240)

I mean, Python is an integral part of it.

Grant Sanderson (1:51:03.840)

Just for the fun of it, let me ask, uh, what, uh, what's your most and least favorite aspects

Lex Fridman (1:51:09.480)

of Python?

Lex Fridman (1:51:10.480)

Ooh, most and least.

Grant Sanderson (1:51:12.600)

I mean, I love that it's like object oriented and functional, I guess that you can kind

Lex Fridman (1:51:18.040)

of like get both of those, um, uh, benefits for how you structure things.

Lex Fridman (1:51:23.960)

So if you would just want to quickly whip something together, the functional aspects

Lex Fridman (1:51:26.760)

are nice.

Grant Sanderson (1:51:27.760)

It's your primary language, like for programmatically generating stuff.

Lex Fridman (1:51:31.160)

Yeah.

Grant Sanderson (1:51:32.160)

It's home for me.

Lex Fridman (1:51:33.160)

It's home.

Grant Sanderson (1:51:34.160)

Yeah.

Lex Fridman (1:51:35.160)

Sometimes you travel, but it's home.

Grant Sanderson (1:51:36.160)

Got it.

Lex Fridman (1:51:37.160)

It's home.

Grant Sanderson (1:51:38.160)

Uh, I mean, the biggest disadvantage is that it's slow.

Lex Fridman (1:51:39.360)

So when you're doing computationally intensive things, either you have to like think about

Grant Sanderson (1:51:42.840)

it more than you should how to make it efficient or it just like takes long.

Lex Fridman (1:51:47.400)

Do you run into that at all?

Lex Fridman (1:51:48.720)

Like with your work?

Grant Sanderson (1:51:49.720)

Well, so, uh, certainly old man is like way slower than it needs to be because of, uh,

Lex Fridman (1:51:54.840)

how it renders things on the backend is like kind of absurd.

Grant Sanderson (1:51:58.640)

I've rewritten things such that it's all done with like shaders in such a way that it should

Grant Sanderson (1:52:02.760)

be just like live and actually like interactive while you're coding it.

Grant Sanderson (1:52:06.680)

If you want to, to have like a 3d scene, you can move around, you can, um, have, um, elements

Grant Sanderson (1:52:12.200)

respond to where your mouse is or things.

Grant Sanderson (1:52:14.240)

That's not something that user of a video is going to get to experience cause there's

Grant Sanderson (1:52:17.860)

just a play button and a pause button.

Lex Fridman (1:52:19.360)

But while you're developing, that can be nice.

Grant Sanderson (1:52:21.280)

Um, so it's gotten better in speed in that sense, but that's basically because the hard

Lex Fridman (1:52:25.480)

work is being done in the language that's not Python, but GLSL, right?

Grant Sanderson (1:52:29.840)

Um, but yeah, there are some times when it's like a, um, there's just a lot of data that

Grant Sanderson (1:52:35.120)

goes into the object that I want to animate that then it just like Python is slow.

Grant Sanderson (1:52:40.000)

Well, let me ask, quickly ask, what do you think about the walrus operator, if you're

Lex Fridman (1:52:44.640)

familiar with it at all?

Grant Sanderson (1:52:46.040)

The reason it's interesting, there's a new operator in Python 3.8.

Grant Sanderson (1:52:49.680)

I find it psychologically interesting cause it, the toxicity over it led Guido to resign

Grant Sanderson (1:52:54.760)

the step down from this.

Lex Fridman (1:52:55.760)

Is that actually true?

Grant Sanderson (1:52:56.760)

Or was it like, there's a bunch of surrounding things that also, was it actually the walrus

Lex Fridman (1:53:00.700)

operator that, that.

Grant Sanderson (1:53:02.200)

Well, it was, it was a text, it was an accumulation of toxicity, but that was the, the most, that

Lex Fridman (1:53:08.320)

was the most toxic one, like the discussion.

Grant Sanderson (1:53:11.640)

That's the most number of Python core developers that were opposed to Guido's decision.

Lex Fridman (1:53:16.480)

Um, he didn't particularly, I don't think cared about it either way.

Grant Sanderson (1:53:20.500)

He just thought it was a good idea.

Lex Fridman (1:53:21.680)

This is where you approve it.

Lex Fridman (1:53:23.580)

And like the structure of the idea of a BDFL is like you listen to everybody, hear everybody

Lex Fridman (1:53:30.000)

out.

Grant Sanderson (1:53:31.000)

You make a decision and you move forward.

Lex Fridman (1:53:33.400)

And he didn't like the negativity that burdened him after that.

Grant Sanderson (1:53:37.600)

People like some parts of the benevolent dictator for life mantra, but once the dictator does

Lex Fridman (1:53:41.520)

things different than you want, suddenly dictatorship doesn't seem so great.

Grant Sanderson (1:53:44.840)

Yeah.

Lex Fridman (1:53:45.840)

I mean, they still liked it.

Grant Sanderson (1:53:46.840)

He just couldn't because he truly is the bee in the benevolent.

Lex Fridman (1:53:50.680)

He's really, he really is a nice guy.

Grant Sanderson (1:53:52.520)

He, I mean, and I think he can't, it's a lot of toxicity.

Lex Fridman (1:53:56.880)

It's difficult.

Grant Sanderson (1:53:57.880)

It's a difficult job.

Lex Fridman (1:53:58.880)

And that's why Linus Torvalds is perhaps the way he is.

Grant Sanderson (1:54:01.480)

You have to have a thick skin to fight off, fight off the warring masses.

Grant Sanderson (1:54:06.920)

It's kind of surprising to me how many people can like threatened to murder each other over

Grant Sanderson (1:54:11.380)

whether we should have braces or not, or like it's incredible.

Lex Fridman (1:54:15.360)

Yeah.

Grant Sanderson (1:54:16.360)

I mean, that's my knee jerk reaction to the walrus operators.

Lex Fridman (1:54:18.160)

Like I don't actually care that much either way.

Grant Sanderson (1:54:20.080)

I'm not going to get personally passionate.

Grant Sanderson (1:54:22.280)

My initial reaction was like, yeah, this seems to make things more confusing to read.

Lex Fridman (1:54:26.440)

But then again, so does list comprehension until you're used to it.

Lex Fridman (1:54:29.300)

So like if there's a use for it, great, if not great, but like, let's just all calm down

Grant Sanderson (1:54:33.880)

about our spaces versus tabs debates here and like, be chill.

Lex Fridman (1:54:37.840)

Yeah.

Grant Sanderson (1:54:38.840)

To me, it just represents the value of great leadership, even in open source communities.

Lex Fridman (1:54:44.240)

Does it represent that if he stepped down as a leader?

Grant Sanderson (1:54:46.640)

Well, he fought for it.

Lex Fridman (1:54:48.120)

No, he got it passed.

Grant Sanderson (1:54:49.600)

I guess, but I guess, I could represent multiple things too.

Grant Sanderson (1:54:54.000)

It can represent like failed dictatorships or it can, it can represent a lot of things,

Lex Fridman (1:54:59.320)

but to me, great leaders take risks.

Lex Fridman (1:55:03.120)

Even if it, even if it's a mistake at the end, like you have to make decisions.

Grant Sanderson (1:55:09.140)

The thing is this world won't go anywhere.

Grant Sanderson (1:55:11.760)

If you constantly, if whenever there's a divisive thing, you wait until the division is no longer

Grant Sanderson (1:55:17.320)

there.

Grant Sanderson (1:55:18.320)

Like that's the paralysis we experienced with like Congress and political systems.

Grant Sanderson (1:55:22.560)

It's good to be slow when there's indecision, when there's people disagree, it's good to

Lex Fridman (1:55:28.800)

take your time.

Lex Fridman (1:55:29.880)

But like at a certain point it results in paralysis and you just have to make a decision.

Grant Sanderson (1:55:34.020)

The background of the site, whether it's yellow, blue, or red can cause people to like go to

Grant Sanderson (1:55:40.040)

war over each other, which I've seen this with design.

Grant Sanderson (1:55:43.080)

People are very touch on color, color choices at the end of the day, just make a decision

Lex Fridman (1:55:49.520)

and go with it.

Lex Fridman (1:55:50.520)

And that, that's what the Walrus operator represents to me is it represents the fighter

Grant Sanderson (1:55:55.400)

pilot instinct of like quick action is more important than, than just like hearing everybody

Lex Fridman (1:56:01.720)

out and really think it through it because that's going to lead to paralysis.

Grant Sanderson (1:56:05.600)

Yeah.

Grant Sanderson (1:56:06.600)

Like if that's the actual case that, you know, it's something where he's consciously hearing

Grant Sanderson (1:56:10.800)

people's disagreement, disagreeing with that disagreement and saying he wants to move forward

Lex Fridman (1:56:16.280)

anyway, that's an admirable aspect of leadership.

Lex Fridman (1:56:21.160)

So we don't have much time, but I want to ask just cause it's some beautiful mathematics

Lex Fridman (1:56:26.520)

involved.

Grant Sanderson (1:56:27.520)

2020 brought us a couple of in the physics world theories of everything, Eric Weinstein

Grant Sanderson (1:56:35.360)

kind of, I mean, it's been working for probably decades, but he put out this idea of geometric

Grant Sanderson (1:56:41.120)

unity or started sort of publicly thinking and talking about it more, Steven Wolfram

Grant Sanderson (1:56:46.720)

put out his physics project, which is kind of this hypergraph view of a theory of everything.

Lex Fridman (1:56:53.560)

Do you find interesting, beautiful things to these theories of everything?

Lex Fridman (1:56:58.920)

What do you think about the physics world and sort of the beautiful, interesting, insightful

Grant Sanderson (1:57:04.520)

mathematics in that world, whether we're talking about quantum mechanics, which you touched

Grant Sanderson (1:57:09.600)

on in a bunch of your videos a little bit, quaternions, like just the mathematics involved

Grant Sanderson (1:57:13.840)

or the general relativity, which is more about surfaces and topology, all that stuff.

Grant Sanderson (1:57:19.360)

Well, I think, um, as far as like popularized science is concerned, people are more interested

Grant Sanderson (1:57:25.160)

in theories of everything than they should be like, cause the problem is whether we're

Grant Sanderson (1:57:29.640)

talking about trying to make sense of Weinstein's lectures or Wolfram's project, or let's just

Grant Sanderson (1:57:34.520)

say like listening to, uh, Witten talk about string theory, whatever proposed path to a

Lex Fridman (1:57:40.000)

theory of everything, um, you're not actually going to understand it.

Grant Sanderson (1:57:44.120)

Some physicists will, but like, you're just not actually going to understand the substance

Lex Fridman (1:57:48.120)

of what they're saying.

Lex Fridman (1:57:49.240)

What I think is way, way more productive is, um, to let yourself get really interested

Grant Sanderson (1:57:53.520)

in the phenomena that are still deep, but which you have a chance of understanding because

Grant Sanderson (1:57:58.320)

the path to getting to like even understanding what questions these theories of everything

Grant Sanderson (1:58:02.300)

are trying to answer involves like walking down that, um, I mean, I was watching a video

Grant Sanderson (1:58:06.680)

before I came here about from Steve mold talking about, um, why sugar polarizes light in a

Lex Fridman (1:58:11.480)

certain way.

Lex Fridman (1:58:12.740)

So fascinating, like really, really interesting.

Grant Sanderson (1:58:15.400)

It's not like this novel theory of everything type thing, but to understand what's going

Grant Sanderson (1:58:19.800)

on there really requires digging in in depth to certain ideas.

Lex Fridman (1:58:23.480)

And if you let yourself think past what the video tells you about what does circularly

Grant Sanderson (1:58:27.440)

polarized light mean and things like that, it actually would get you to a pretty good

Grant Sanderson (1:58:31.160)

appreciation of like two state states and quantum systems, um, in a way that just trying

Grant Sanderson (1:58:36.280)

to read about like, Oh, what's the, um, what are the hard parts about resolving quantum

Lex Fridman (1:58:40.920)

field theories with general relativity is never going to get you.

Lex Fridman (1:58:44.400)

So as far as popularizing science is concerned, like the audience should be less interested

Lex Fridman (1:58:50.340)

than they are in theories of everything.

Grant Sanderson (1:58:52.720)

Um, the popularizers should be less emphatic than they are about that for like actual practicing

Lex Fridman (1:58:59.360)

physicists.

Lex Fridman (1:59:00.360)

And that might be the case.

Grant Sanderson (1:59:01.360)

Maybe more people should think about fundamental questions, but it's difficult to create, uh,

Grant Sanderson (1:59:06.040)

like a three blue, one brown video on the theory of everything.

Lex Fridman (1:59:09.600)

So basically we should really try to find the beauty in mathematics or physics by looking

Grant Sanderson (1:59:16.040)

at concepts that are like within reach.

Lex Fridman (1:59:18.440)

Yeah, I think that's super important.

Grant Sanderson (1:59:20.320)

I mean, so you see this in math too with, um, the big unsolved problems.

Lex Fridman (1:59:25.360)

So like the clay millennium problems, Riemann hypothesis, um, have you ever done a video

Lex Fridman (1:59:29.400)

on Fermat's last theorem?

Lex Fridman (1:59:30.400)

No, I have not yet.

Grant Sanderson (1:59:31.720)

No.

Lex Fridman (1:59:32.720)

But if I did, do you know what I would do?

Grant Sanderson (1:59:34.120)

I would talk about, um, proving Fermat's last theorem in the specific case of N equals

Lex Fridman (1:59:38.720)

three.

Grant Sanderson (1:59:39.720)

Okay.

Lex Fridman (1:59:40.720)

Is that still accessible though?

Grant Sanderson (1:59:41.720)

Yes.

Lex Fridman (1:59:42.720)

Actually barely.

Grant Sanderson (1:59:43.720)

Um, Mathologer might be able to do like a great job on this.

Grant Sanderson (1:59:46.160)

He does a good job of taking stuff that's barely accessible and making it, but the,

Grant Sanderson (1:59:50.480)

the core ideas of proving it for N equals three are hard, but they do get you real ideas

Lex Fridman (1:59:55.280)

about algebraic number theory.

Lex Fridman (1:59:56.960)

And it involves looking at a number field that's, uh, it lives in the complex plane.

Grant Sanderson (20:02.160)

it such that it's all above the plane, but it's boundary sits exactly on the plane.

Grant Sanderson (20:08.120)

I don't think I can do that without crossing itself, but that feels really vague.

Lex Fridman (20:11.320)

How do I formalize it?

Lex Fridman (20:12.320)

And as you're starting to formalize that, that's what's going to get you to try to come

Lex Fridman (20:16.480)

up with a definition for what it means to be orientable or non orientable.

Lex Fridman (20:19.760)

And like once you have that motivation, a lot of the otherwise arbitrary things that

Grant Sanderson (20:22.920)

are sitting at the very beginning of a topology stack textbook start to make a little more

Grant Sanderson (20:26.560)

sense.

Lex Fridman (20:27.560)

Yeah.

Lex Fridman (20:28.560)

And I mean that, that whole video beautifully was a motivation for topology school.

Grant Sanderson (20:32.400)

That was my, well, my hope with that is I feel like topology is, um, I don't want to

Grant Sanderson (20:36.400)

say it's taught wrong, but I do think sometimes it's popularized in the wrong way where, uh,

Grant Sanderson (20:41.920)

you know, you'll hear these things that people saying, Oh, topologists, they're very interested

Grant Sanderson (20:44.920)

in surfaces that you can bend and stretch, but you can't cut or glue.

Lex Fridman (20:50.180)

Are they?

Lex Fridman (20:51.180)

Why?

Lex Fridman (20:52.180)

Yeah.

Grant Sanderson (20:53.180)

There's all sorts of things you can be interested in with random, like imaginative manipulations

Lex Fridman (20:56.780)

of things.

Lex Fridman (20:57.780)

Is that really what like mathematicians are into?

Lex Fridman (21:00.200)

And the short answer is not, not really.

Grant Sanderson (21:03.120)

That's uh, it's not as if someone was sitting there thinking like, I wonder what the properties

Grant Sanderson (21:07.580)

of clay are by add some arbitrary rules about what, when I can't cut it and when I can't

Grant Sanderson (21:12.120)

glue it instead, it's, there's a ton of pieces of math that, um, can actually be equivalent

Grant Sanderson (21:18.680)

to, uh, like these very general structures that's like geometry, except you don't have

Grant Sanderson (21:23.200)

exact distances.

Lex Fridman (21:24.200)

You just want to maintain a notion of closeness.

Lex Fridman (21:26.400)

And once you get it to those general structures, constructing mappings between them translate

Grant Sanderson (21:31.400)

into non trivial facts about other parts of math and that I just, I don't think that's

Grant Sanderson (21:36.360)

actually like popularized.

Grant Sanderson (21:38.040)

Um, I don't even think it's emphasized well enough when you're starting to take a topology

Grant Sanderson (21:41.640)

class because you kind of have these two problems.

Lex Fridman (21:43.560)

It's like either it's too squishy.

Grant Sanderson (21:45.320)

You're just talking about coffee mugs and donuts, or it's a little bit too rigor first.

Lex Fridman (21:49.720)

And you're talking about, um, the axiom systems with open sets and an open set is not the

Grant Sanderson (21:55.280)

opposite of closed set.

Lex Fridman (21:56.600)

So sorry about that.

Grant Sanderson (21:57.600)

Everyone, we have a notion of clopin sets for ones that are both at the same time.

Lex Fridman (22:01.800)

Yeah.

Grant Sanderson (22:02.800)

It's just, it's not, it's not an intuitive axiom system in comparison to other fields

Lex Fridman (22:06.860)

of math.

Lex Fridman (22:07.860)

So you as the student like really have to walk through mud to get there and you're constantly

Grant Sanderson (22:11.140)

confused about how this relates to the beautiful things about coffee mugs and Mobius strips

Lex Fridman (22:15.240)

and such.

Lex Fridman (22:16.240)

And it takes a really long time to actually see like see topology in the way that mathematicians

Grant Sanderson (22:21.140)

see topology.

Lex Fridman (22:22.400)

But I don't think it needs to take that time.

Grant Sanderson (22:23.840)

I think there's, um, this is making me feel like I need to make more videos on the topic

Lex Fridman (22:27.800)

because I think you do, but you know, I've also seen it in my narrow view.

Grant Sanderson (22:32.760)

Uh, like, um, I find game theory very beautiful and I know topology has been used, uh, elegantly

Lex Fridman (22:39.840)

to prove things in game theory.

Grant Sanderson (22:41.320)

Yeah.

Lex Fridman (22:42.320)

You have like facts that seem very strange.

Grant Sanderson (22:43.760)

Like I could tell you, you stir your coffee and um, after you stir it and like, let's

Grant Sanderson (22:47.920)

say all the molecules settled to like not moving again, one of the molecules will be

Grant Sanderson (22:51.040)

basically in the same position it was before.

Lex Fridman (22:53.360)

Um, you have all sorts of fixed point theorems like this, right?

Lex Fridman (22:57.200)

That kind of fixed point theorem directly relevant to Nash equilibriums, right?

Grant Sanderson (23:01.320)

Um, so you can imagine popularizing it by describing the coffee fact, but then you're

Lex Fridman (23:05.680)

left to wonder like who cares about if a molecule of coffee like stays in the same spot?

Lex Fridman (23:09.160)

Is this what we're paying our mathematicians for?

Grant Sanderson (23:11.160)

Um, you have this very elegant mapping onto economics in a way that's very concrete or

Grant Sanderson (23:15.640)

very, I shouldn't say concrete, very, uh, tangible, like actually adds value to people's

Grant Sanderson (23:20.240)

lives through the predictions that it makes.

Grant Sanderson (23:22.320)

Uh, but that line isn't always drawn because like you have to get a little bit technical

Grant Sanderson (23:26.780)

in order to properly draw that line out, um, and often I think popularized forms of media

Lex Fridman (23:34.160)

just shy away from being a little too technical for sure.

Grant Sanderson (23:37.760)

Uh, by the way, for people who are watching the video, I do not condone the message in

Lex Fridman (23:41.840)

this mug.

Grant Sanderson (23:42.840)

It's the only one I have, which is this.

Lex Fridman (23:44.760)

The snuggle is real.

Grant Sanderson (23:47.360)

By the way, for anyone watching, I do condone the message of that mug.

Lex Fridman (23:50.240)

The snuggle is real.

Grant Sanderson (23:51.240)

The snuggle is real.

Lex Fridman (23:52.240)

Okay, so you mentioned the SIR model.

Grant Sanderson (23:57.160)

I think, uh, there are certain ideas there of growth of exponential growth.

Lex Fridman (24:03.220)

What maybe have you learned about, um, pandemics from, from making that video?

Grant Sanderson (24:11.680)

Because it was kind of exploratory.

Grant Sanderson (24:12.680)

You were kind of building up an intuition and it's, again, people should watch the video.

Grant Sanderson (24:17.680)

It's kind of an abstract view.

Lex Fridman (24:19.160)

It's not really modeling in detail.

Grant Sanderson (24:23.480)

The whole field of epidemiology, those, those people, they go really far in terms of modeling,

Lex Fridman (24:31.440)

like how people move about.

Grant Sanderson (24:32.440)

I don't know if you've seen it, but like there is the mobility patterns, like how, like the

Grant Sanderson (24:37.600)

track, like how many people you encounter in a certain situations when you go to a school,

Grant Sanderson (24:42.560)

when you go to a mall, they like model every aspect of that for a particular city.

Lex Fridman (24:46.400)

Like they have maps of actual city streets.

Grant Sanderson (24:49.880)

They model it really well and natural patterns of the people have it's crazy.

Lex Fridman (24:54.280)

So you don't do any of that.

Grant Sanderson (24:55.280)

You're just doing an abstract model to explore different ideas of simple pedigree.

Grant Sanderson (24:59.400)

Well, because I don't want to pretend like an epidemiologist, I'm an epidemiologist.

Grant Sanderson (25:02.680)

Like we have a ton of armchair epidemiologists and the spirit of that was more like, uh,

Lex Fridman (25:08.440)

can we through a little bit of play, uh, draw like reasonable ish conclusions.

Grant Sanderson (25:12.680)

Um, and also just like, uh, get ourselves in a position where we can judge the validity

Lex Fridman (25:17.680)

of a model.

Grant Sanderson (25:18.680)

Like, I think people should look at that and they should criticize it.

Lex Fridman (25:21.160)

They should point to all the ways that it's wrong because it's definitely naive, right?

Lex Fridman (25:24.640)

And the way that it's set up.

Grant Sanderson (25:25.920)

Um, but to say like what, what lessons from that hold, like thinking about the are not

Grant Sanderson (25:30.480)

value and what that represents and what it can imply.

Grant Sanderson (25:33.720)

Um, so are not is if you are infectious and you're in a population which is completely

Grant Sanderson (25:40.440)

susceptible, uh, what's the average number of people that you're going to infect during

Lex Fridman (25:44.760)

your infectiousness?

Grant Sanderson (25:46.080)

Um, so certainly during the beginning of an epidemic, this basically gives you kind of

Lex Fridman (25:51.280)

the, um, the exponential growth rate.

Grant Sanderson (25:53.280)

Like if every person infects two others, you've got that one, two, four, eight, uh, exponential

Lex Fridman (25:57.280)

growth pattern.

Grant Sanderson (25:58.280)

Um, as it goes on and, uh, let's say it's something, um, uh, endemic where you've got

Grant Sanderson (26:04.280)

like a ton of people who have had it, uh, and are recovered, then, uh, you, you would,

Grant Sanderson (26:10.040)

the are not value doesn't tell you that as directly because a lot of the people you interact

Lex Fridman (26:13.960)

with aren't susceptible, but in the early phases it does.

Grant Sanderson (26:16.880)

Um, and this is like the fundamental constant that it seems like epidemiologists look at

Lex Fridman (26:21.360)

and you know, the whole goal is to get that down.

Grant Sanderson (26:23.460)

If you can get it below one, then it's no longer epidemic.

Grant Sanderson (26:26.560)

If it's equal to one, then it's endemic, um, and it's above one, then your epidemic.

Grant Sanderson (26:30.960)

So, uh, like just teaching what that value is and giving some intuitions on how do certain

Grant Sanderson (26:36.120)

changes in behavior change that value and then what does that imply for exponential

Lex Fridman (26:40.160)

growth?

Grant Sanderson (26:41.160)

I think those are, um, general enough lessons and they're like resilient to all of the

Grant Sanderson (26:47.140)

chaoses of the world, um, that it's still like valid to take from the video.

Grant Sanderson (26:52.400)

I mean, one of the interesting aspects of that is just exponential growth and we think

Grant Sanderson (26:56.040)

about growth.

Grant Sanderson (26:57.040)

Is that the one of the first times you've done a video on, on, uh, no, of course not

Grant Sanderson (27:03.480)

the whole, uh, well there's identity.

Lex Fridman (27:07.360)

Okay.

Grant Sanderson (27:08.360)

So

Lex Fridman (27:09.360)

sure.

Grant Sanderson (27:10.360)

I guess I've done a lot of videos about exponential growth in the circular direction, uh, only

Lex Fridman (27:13.360)

minimal in the normal direction.

Grant Sanderson (27:14.880)

I mean, another way to ask, like, do you think we're able to reason intuitively about exponential

Lex Fridman (27:22.280)

growth?

Grant Sanderson (27:23.280)

It's, it's funny.

Grant Sanderson (27:25.360)

I think it's, um, I think it's extremely intuitive to humans and then we train it out of ourselves

Grant Sanderson (27:30.360)

such that it's then really not intuitive and then I think it can become intuitive again

Lex Fridman (27:33.720)

when you study a technical field.

Grant Sanderson (27:35.200)

Uh, so what I mean by that is, um, have you ever heard of these studies where in a, uh,

Grant Sanderson (27:41.240)

like anthropological setting where you're studying a group that has been disassociated

Grant Sanderson (27:46.000)

from a lot of like modern society and you ask what number is between one and nine and

Grant Sanderson (27:51.760)

maybe you would ask you, you've got like one rock and you've got nine rocks, you're like

Lex Fridman (27:54.600)

what pile is halfway in between these and our instinct is usually to say five.

Lex Fridman (27:59.960)

That's the number that sits right between one and nine.

Grant Sanderson (28:02.000)

Um, but sometimes when a numeracy and, uh, the kind of just basic arithmetic that we

Grant Sanderson (28:07.360)

have isn't in a society, the natural instinct is three because it's, uh, in between in an

Grant Sanderson (28:12.840)

exponential sense and a geometric sense that, uh, one is three times bigger and then the

Lex Fridman (28:16.920)

next one is three times bigger than that.

Lex Fridman (28:18.520)

So it's like, what's, you know, if you have one friend versus a hundred friends, what's

Lex Fridman (28:22.240)

in between that?

Grant Sanderson (28:23.240)

Ten friends seems like the social status in between those two states.

Lex Fridman (28:26.840)

So that's like deeply intuitive to us to think logarithmically like that.

Grant Sanderson (28:30.200)

Um, and for some reason we kind of train it out of ourselves to start thinking linearly

Lex Fridman (28:35.120)

about things.

Lex Fridman (28:36.120)

So in the sense, yeah, the early, early basic math is, uh, yeah, forces us to take a step

Lex Fridman (28:42.240)

back.

Grant Sanderson (28:43.240)

It's, it's the same criticism if there's any of science is the lessons of science make

Grant Sanderson (28:51.160)

us like see the world in a slightly narrow sense to where we, we have an over exaggerated

Grant Sanderson (28:57.760)

confidence that we understand everything as opposed to just understanding a small slice

Lex Fridman (29:02.480)

of it.

Lex Fridman (29:03.560)

But I think that probably only really goes for small numbers cause the real counterintuitive

Lex Fridman (29:07.280)

thing about exponential growth is like as the numbers start to get big.

Lex Fridman (29:10.340)

So I bet if you took that same setup and you asked them, oh, if I keep tripling the size

Lex Fridman (29:14.660)

of this rock pile, you know, um, seven times, how big will it be?

Grant Sanderson (29:18.320)

I bet it would be surprisingly big even to like an a society without numeracy.

Lex Fridman (29:23.680)

And that's the side of it that, um, I think is pretty counterintuitive to us, uh, but

Grant Sanderson (29:28.360)

that you can basically train into people like I think computer scientists and physicists

Grant Sanderson (29:33.320)

when they're looking at the early numbers of, um, like COVID were, they were the ones

Grant Sanderson (29:38.520)

thinking like, oh God, this is following an exact exponential curve.

Grant Sanderson (29:42.080)

Um, and I heard that from a number of people, uh, so it's, and, and almost all of them are

Grant Sanderson (29:47.240)

like techies in some capacity, probably just cause I like live in the Bay area, but, but

Grant Sanderson (29:51.800)

for sure they, they're cognizant of this kind of, this kind of growth is present in a lot

Grant Sanderson (29:56.320)

of natural systems and a lot of, in a lot of, in a lot of systems.

Grant Sanderson (2:00:00.760)

It looks like a hexagonal lattice and you start asking questions about factoring numbers

Grant Sanderson (2:00:04.680)

in this hexagonal lattice.

Lex Fridman (2:00:06.540)

So it takes a while, but I've talked about this sort of like lattice arithmetic, um,

Grant Sanderson (2:00:10.280)

in other contexts and you can get to a okay understanding of that.

Lex Fridman (2:00:15.680)

And the things that make Fermat's last theorem hard are actually quite deep.

Grant Sanderson (2:00:18.720)

Um, and so the cases that we can solve it for, it's like you can get these broad sweeps

Lex Fridman (2:00:23.040)

based on some hard, but like accessible, um, bits of number theory.

Lex Fridman (2:00:28.800)

But before you can even understand why the general case is as hard as it is, you have

Lex Fridman (2:00:32.280)

to walk through those.

Lex Fridman (2:00:33.960)

And so any other attempt to describe it would just end up being like shallow and not really

Lex Fridman (2:00:38.040)

productive for the viewer's time.

Grant Sanderson (2:00:39.840)

Um, I think the same goes for, uh, most like unsolved problem type things where I think,

Grant Sanderson (2:00:45.900)

you know, as a kid, I was actually very inspired by the twin prime conjecture, um, that like

Grant Sanderson (2:00:49.860)

totally sucked me in as this thing that was understandable.

Grant Sanderson (2:00:52.400)

I kind of had this dream like, Oh, maybe I'll be the one to prove the twin prime conjecture

Lex Fridman (2:00:55.960)

and new math that I would learn would be like viewed through this lens of like, Oh, maybe

Lex Fridman (2:01:00.000)

I can apply it to that in some way.

Grant Sanderson (2:01:01.920)

But, uh, you sort of mature to a point where you realize that, uh, you should spend your

Grant Sanderson (2:01:08.280)

brain cycles on problems that you will see resolved because then you're going to grow

Grant Sanderson (2:01:12.920)

to see what it feels like for these things to be resolved rather than spending your brain

Lex Fridman (2:01:16.420)

cycles on something where it's not, it's not going to pan out.

Grant Sanderson (2:01:19.920)

Um, and the people who do make progress towards these things like James Maynard, uh, is a

Grant Sanderson (2:01:24.280)

great example here of like young creative mathematician who like pushes in the direction

Grant Sanderson (2:01:28.960)

of things like the twin prime conjecture rather than hitting that head on, just see all the

Grant Sanderson (2:01:33.040)

interesting questions that are hard for similar reasons, but become more tractable and let

Grant Sanderson (2:01:36.520)

themselves really engage with those.

Lex Fridman (2:01:38.480)

Um, so I think people should get in that habit.

Grant Sanderson (2:01:41.280)

I think the popularization of physics should encourage that habit through things like the

Lex Fridman (2:01:46.160)

physics of simple everyday phenomena, because it can get quite deep.

Lex Fridman (2:01:50.000)

And um, yeah, I think I, you know, I've, I've heard a lot of the interest that, you know,

Grant Sanderson (2:01:54.680)

people send me messages asking to explain Weinstein's thing or asking to explain Wolfram's

Grant Sanderson (2:01:58.880)

thing.

Grant Sanderson (2:01:59.880)

One, I don't understand them, but more importantly, um, it's too big a bite to, you shouldn't

Lex Fridman (2:02:06.400)

be interested in those, right?

Lex Fridman (2:02:08.960)

The giant sort of a ball of interesting ideas.

Grant Sanderson (2:02:12.480)

There's probably a million of interesting ideas in there that individually could be

Lex Fridman (2:02:16.600)

explored effectively.

Lex Fridman (2:02:17.600)

And to be clear, you should be interested in fundamental questions.

Grant Sanderson (2:02:20.080)

I think that's a good habit to ask what the fundamentals of things are, but I think it

Grant Sanderson (2:02:25.160)

takes a lot of steps to like, certainly you shouldn't be trying to answer that unless

Grant Sanderson (2:02:29.480)

you actually understand quantum field theory and you actually understand general relativity.

Grant Sanderson (2:02:33.520)

That's the cool thing about like your videos, people who haven't done mathematics, like

Grant Sanderson (2:02:37.200)

if you really give it time, watch it a couple of times and like try to try to reason about

Grant Sanderson (2:02:42.560)

it, you can actually understand the concept that's being explained.

Lex Fridman (2:02:45.400)

And it's not a coincidence that the things I'm describing aren't like the most, um, up

Grant Sanderson (2:02:49.980)

to date, uh, progress on the Riemann hypothesis cousins or, um, like there's context in which

Grant Sanderson (2:02:55.720)

the analog of the Riemann hypothesis has been solved in like more, uh, discrete feeling

Grant Sanderson (2:03:00.200)

finite settings that are more well behaved.

Lex Fridman (2:03:02.360)

I'm not describing that because it just takes a ton to get there.

Lex Fridman (2:03:05.840)

And instead I think it'll be like productive to have an actual understanding of something

Lex Fridman (2:03:10.800)

that can, you can pack into 20 minutes.

Grant Sanderson (2:03:12.940)

I think that's beautifully put ultimately.

Grant Sanderson (2:03:15.160)

That's where like the most satisfying thing is when you really understand, um, yeah, really

Grant Sanderson (2:03:20.320)

understand, build a habit of feeling what it's like to actually come to resolution.

Lex Fridman (2:03:25.080)

Yeah.

Grant Sanderson (2:03:26.080)

Yeah.

Grant Sanderson (2:03:27.080)

As opposed to, which it can also be enjoyable, but just being in awe of the fact that you

Grant Sanderson (2:03:32.120)

don't understand anything.

Lex Fridman (2:03:33.120)

Yeah.

Grant Sanderson (2:03:34.120)

That's not like, I don't know.

Grant Sanderson (2:03:35.120)

Maybe we'll get entertainment out of that, but it's not as fulfilling as understanding

Grant Sanderson (2:03:40.440)

you won't grow.

Lex Fridman (2:03:41.920)

Yeah.

Grant Sanderson (2:03:42.920)

And, but also just the fulfilling, it really does feel good when you first don't understand

Lex Fridman (2:03:47.560)

something and then you do, that's a beautiful feeling.

Grant Sanderson (2:03:51.040)

Hey, let me ask you one, uh, last, last time we got awkward and weird about, uh, a fear

Grant Sanderson (2:03:57.080)

of mortality, which you made fun of me off, but let me ask you on the, the other absurd

Lex Fridman (2:04:01.600)

question is, um, what do you think is, uh, the meaning of our life of meaning of life?

Lex Fridman (2:04:08.000)

I'm sorry if I made fun of you about, no, you didn't.

Grant Sanderson (2:04:11.160)

I'm just joking.

Lex Fridman (2:04:12.160)

It was great.

Grant Sanderson (2:04:13.160)

I don't think life has a meaning.

Lex Fridman (2:04:15.300)

I think like meaning, I don't understand the question.

Grant Sanderson (2:04:18.560)

I think meaning is something that's described to stuff that's created with purpose.

Grant Sanderson (2:04:22.600)

There's a meaning to, uh, like this water bottle label and that someone created it with

Grant Sanderson (2:04:26.440)

a purpose of conveying meaning.

Lex Fridman (2:04:27.960)

And there was like one consciousness that wanted to get its ideas into another consciousness.

Grant Sanderson (2:04:31.560)

Um, most things don't have that property.

Grant Sanderson (2:04:35.640)

It's a little bit like if I asked you, um, like what is the height, all right, so it's

Grant Sanderson (2:04:41.160)

all relative.

Lex Fridman (2:04:42.160)

Yeah.

Grant Sanderson (2:04:43.160)

You'd be like the height of what you can't ask.

Lex Fridman (2:04:44.840)

What is the height without an object?

Grant Sanderson (2:04:46.600)

You can't ask what is the meaning of life without like an intentful consciousness, putting

Grant Sanderson (2:04:50.880)

it like, I guess I'm revealing I'm not very religious, but you know, the mathematics of

Grant Sanderson (2:04:56.400)

everything seems kind of beautiful.

Grant Sanderson (2:04:58.800)

It seems like, it seems like there's some kind of structure relative to which, I mean,

Grant Sanderson (2:05:05.280)

you could calculate the height.

Lex Fridman (2:05:06.880)

Well, so, but what I'm saying is I don't understand the question.

Lex Fridman (2:05:09.840)

What is the meaning of life in that?

Lex Fridman (2:05:10.840)

I think people might be asking something very real.

Grant Sanderson (2:05:13.680)

I don't understand what they're asking.

Lex Fridman (2:05:14.680)

Are they asking like, why does life exist?

Lex Fridman (2:05:16.800)

Like how did it come about?

Lex Fridman (2:05:17.840)

What are the natural laws?

Lex Fridman (2:05:19.320)

Are they asking, um, as I'm making decisions day by day for what should I do?

Lex Fridman (2:05:23.600)

What is the guiding light that inspires like, what should I do?

Grant Sanderson (2:05:26.240)

I think that's what people are kind of asking.

Lex Fridman (2:05:27.960)

But also like why the thing that gives you joy about education, about mathematics, what

Lex Fridman (2:05:36.160)

the hell is that?

Grant Sanderson (2:05:37.680)

Like what interactions with other people, interactions with like minded people, I think

Grant Sanderson (2:05:41.840)

is the meaning of, in that sense, bringing others joy, essentially, like in something

Grant Sanderson (2:05:46.860)

you've created, it connects with others somehow and the same and the vice versa.

Grant Sanderson (2:05:53.560)

I think that that is what, um, when we use the word meaning to mean like you're sort

Grant Sanderson (2:05:57.460)

of filled with a sense of happiness and energy to create more things, like I have so much

Grant Sanderson (2:06:01.600)

meaning taken from this, like that, yeah, that's what fuels, fuels my pump at least.

Lex Fridman (2:06:06.640)

So a life alone on a desert island would be kind of meaningless.

Grant Sanderson (2:06:10.480)

Yeah.

Lex Fridman (2:06:11.480)

You want to be alone together with someone.

Grant Sanderson (2:06:13.960)

I think we're all alone together.

Lex Fridman (2:06:15.680)

I think there's no better way to end it, Grant.

Grant Sanderson (2:06:18.600)

You've been, first time we talked, it was amazing again, it's a huge honor that you

Lex Fridman (2:06:22.200)

make time for me.

Grant Sanderson (2:06:23.200)

I appreciate talking with you.

Lex Fridman (2:06:24.200)

Thanks, man.

Grant Sanderson (2:06:25.200)

Awesome.

Lex Fridman (2:06:26.200)

Thanks for listening.

Grant Sanderson (2:06:27.200)

Thanks for listening to this conversation with Grant Sanderson.

Lex Fridman (2:06:29.640)

And thank you to our sponsors, Dollar Shave Club, DoorDash, and Cash App.

Grant Sanderson (2:06:34.840)

Click the sponsor links in the description to get a discount and to support this podcast.

Grant Sanderson (2:06:39.000)

If you enjoy this thing, subscribe on YouTube, review it with five stars on Apple Podcast,

Grant Sanderson (2:06:44.520)

follow on Spotify, support on Patreon, or connect with me on Twitter at Lex Friedman.

Lex Fridman (2:06:50.800)

And now let me leave you with some words from Richard Feynman.

Grant Sanderson (2:06:54.240)

I have a friend who's an artist and has sometimes taken a view which I don't agree with very

Lex Fridman (2:06:59.440)

well.

Grant Sanderson (2:07:00.440)

He'll hold up a flower and say, look how beautiful it is, and I'll agree.

Grant Sanderson (2:07:04.960)

Then he says, I as an artist can see how beautiful this is, but you as a scientist take this

Grant Sanderson (2:07:10.800)

all apart and it becomes a dull thing.

Lex Fridman (2:07:13.960)

And I think he's kind of nutty.

Grant Sanderson (2:07:16.240)

First of all, the beauty that he sees is available to other people and to me too, I believe.

Grant Sanderson (2:07:21.560)

Although I may not be quite as refined aesthetically as he is, I can appreciate the beauty of a

Grant Sanderson (2:07:26.960)

flower.

Lex Fridman (2:07:27.960)

At the same time, I see much more about the flower than he sees.

Grant Sanderson (2:07:31.160)

I can imagine the cells in there, the complicated actions inside, which also have a beauty.

Grant Sanderson (2:07:37.040)

I mean, it's not just beauty at this dimension at one centimeter, there's also beauty at

Grant Sanderson (2:07:40.900)

smaller dimensions, the inner structure, also the processes.

Grant Sanderson (2:07:46.280)

The fact that the colors in the flower evolved in order to attract insects to pollinate it

Grant Sanderson (2:07:50.480)

is interesting.

Lex Fridman (2:07:52.040)

It means that insects can see the color.

Grant Sanderson (2:07:54.880)

It adds a question.

Lex Fridman (2:07:56.240)

Does this aesthetic sense also exist in the lower forms?

Lex Fridman (2:07:59.760)

Why is it aesthetic?

Grant Sanderson (2:08:01.320)

All kinds of interesting questions which the science knowledge only adds to the excitement,

Grant Sanderson (2:08:05.920)

the mystery and the awe of a flower.

Lex Fridman (2:08:08.700)

It only adds.

Grant Sanderson (2:08:09.700)

I don't understand how it subtracts.

Lex Fridman (2:08:13.280)

Thank you for listening and hope to see you next time.

Grant Sanderson (30:01.240)

Uh, I don't know if you've seen like, I mean, there's a lot of ways to visualize this obviously,

Lex Fridman (30:04.920)

but Raker as well, I think was the one that had this like chess board where, um, every,

Grant Sanderson (30:12.000)

every square on the chess board, you double the number of stones or something in that

Lex Fridman (30:15.720)

chess board.

Grant Sanderson (30:16.720)

I've heard, this is like an old proverb where it's like, you know, someone, the King offered

Grant Sanderson (30:20.800)

him a gift and he said, ah, the only gift I would like very modest, give me a single

Grant Sanderson (30:24.560)

grain of rice for the first chess board and then two grains of rice for the next square.

Lex Fridman (30:29.800)

Then twice that for the next square and just continue on.

Grant Sanderson (30:32.600)

That's my only modest ask your sire and like, it's all, you know, more grains of rice than

Lex Fridman (30:37.120)

there are, uh, anything in the world, um, by the time you get to the end.

Lex Fridman (30:41.680)

And I, I, my intuition falls apart there, like, I would have never predicted that, like

Grant Sanderson (30:47.960)

for some reason, that's a really compelling, uh, illustration, how poorly breaks down.

Grant Sanderson (30:54.080)

Just like you said, maybe we're okay for the first few piles, but after, uh, of rocks,

Lex Fridman (30:58.280)

but after a while it's game over.

Grant Sanderson (31:00.860)

You know, the other classic example for, um, gauging someone's intuitive understanding

Grant Sanderson (31:04.820)

of exponential growth is, uh, I've got like a Lily pad on a, on like really big lake,

Grant Sanderson (31:10.200)

um, like lake Michigan and that Lily pad replicates, it doubles, um, one day and then it doubles

Lex Fridman (31:16.320)

the next day and it doubles the next day.

Grant Sanderson (31:17.960)

Um, and after 50 days, um, it actually is going to cover the entire lake.

Lex Fridman (31:22.320)

Okay.

Lex Fridman (31:23.320)

So after how many days does it cover half the lake?

Lex Fridman (31:25.840)

49.

Lex Fridman (31:26.840)

So you, you have a good instinct for exponential growth.

Lex Fridman (31:30.600)

Right.

Lex Fridman (31:31.600)

So I think a lot of, uh, like the knee jerk reaction is sometimes to think that it's like

Grant Sanderson (31:35.560)

half the amount of time or to at least be like surprised that like after 49 days, you've

Grant Sanderson (31:41.160)

only covered half of it.

Lex Fridman (31:42.400)

Um, yeah.

Grant Sanderson (31:43.400)

I mean, that's the reason you heard a pause for me.

Lex Fridman (31:46.080)

Um, I literally thought that can't be right.

Grant Sanderson (31:48.440)

Right.

Lex Fridman (31:49.440)

Yeah, exactly.

Lex Fridman (31:50.440)

So even when you know the fact and you do the division, it's like, wow.

Lex Fridman (31:53.480)

So you've gotten like that whole time and then day 49, it's only covering half.

Lex Fridman (31:57.000)

And then after that it gets the whole thing.

Lex Fridman (31:58.940)

But I think you can make that even more visceral if rather than going one day before you say

Lex Fridman (32:02.080)

how long until, um, it's covered 1% of the lake, right.

Lex Fridman (32:06.280)

And it's, uh, so what would that be?

Grant Sanderson (32:08.080)

Um, how many times you have to double to get over a hundred, like seven, six and a half

Lex Fridman (32:12.240)

times, something like that.

Grant Sanderson (32:13.240)

Right.

Lex Fridman (32:14.240)

So at that point you're looking at 43, 44 days into it.

Grant Sanderson (32:17.920)

You're not even at 1% of the lake.

Lex Fridman (32:19.880)

So you've, you've experienced, you know, 44 out of 50 days and you're like, ah, that's

Grant Sanderson (32:23.360)

really bad.

Lex Fridman (32:24.360)

It's just 1% of the lake.

Lex Fridman (32:25.580)

But then next thing you know, it's the entire lake.

Lex Fridman (32:28.760)

You're wearing a space X shirt.

Lex Fridman (32:30.040)

So let me ask you, let me ask you one, one person who talks about exponential, you know,

Lex Fridman (32:38.000)

just the miracle of the exponential function in general is Elon Musk.

Lex Fridman (32:42.700)

So he kind of advocates the idea of exponential thinking, you know, realizing that technological

Grant Sanderson (32:51.720)

development can, at least in the short term, follow exponential improvement, which breaks

Grant Sanderson (32:58.080)

apart our intuition, our ability to reason about what is and isn't impossible.

Lex Fridman (33:03.120)

So he's a big one.

Grant Sanderson (33:04.760)

It's a good leadership kind of style of saying like, look, the thing that everyone thinks

Lex Fridman (33:09.080)

is impossible is actually possible because exponentials.

Lex Fridman (33:13.800)

But what's your sense about, um, about that kind of way to see the world?

Grant Sanderson (33:19.240)

Well, so I think it's, um, it can be very inspiring to note when something like Moore's

Grant Sanderson (33:25.140)

law is another great example where you have this exponential pattern that holds shockingly

Lex Fridman (33:29.160)

well.

Grant Sanderson (33:30.160)

Um, and it enables, um, just better lives to be led.

Grant Sanderson (33:33.480)

I think the people who took Moore's law seriously in the sixties, we're seeing that, wow, it's

Grant Sanderson (33:37.980)

not going to be too long before like these giant computers that are either batch processing

Grant Sanderson (33:41.580)

or time shared, you could actually have one small enough to put on your desk on top of

Grant Sanderson (33:45.180)

your desk and you could do things.

Lex Fridman (33:46.780)

And if they took it seriously, like you have people predicting smartphones like a long

Grant Sanderson (33:49.740)

time ago.

Grant Sanderson (33:50.740)

Um, and it's only out of like kind of this, I don't want to say faith in exponentials,

Lex Fridman (33:55.120)

but an understanding that that's what's happening.

Grant Sanderson (33:58.120)

What's more interesting I think is to, um, really understand why exponential growth happens

Lex Fridman (34:03.400)

and that the mechanism behind it is when the rate of change is proportional to the thing

Lex Fridman (34:07.400)

in and of itself.

Lex Fridman (34:08.880)

So the reason that technology would grow exponentially is only going to be if, um, the rate of progress

Lex Fridman (34:14.240)

is proportional to the amount that you have.

Lex Fridman (34:16.360)

So that the software you write enables you to write more software.

Grant Sanderson (34:19.960)

Um, and I think we see this with the internet, like the advent of the internet makes it faster

Grant Sanderson (34:25.080)

to learn things, which makes it faster to, uh, create new things.

Grant Sanderson (34:29.200)

Um, I think this is, uh, oftentimes why like investment will grow exponentially that the

Grant Sanderson (34:35.600)

more resources a company has, if it knows how to use them, well, the more, uh, the more

Lex Fridman (34:40.680)

it can actually grow.

Grant Sanderson (34:41.680)

So, I mean, you know, you referenced Elon Musk.

Lex Fridman (34:43.520)

I think he seems to really be into vertically integrating his companies.

Grant Sanderson (34:47.320)

I think a big part of that is because you have the sense, what you want is to make sure

Grant Sanderson (34:49.920)

that the things that you develop, you have ownership of in the, they enable further development

Lex Fridman (34:54.060)

of the adjacent parts, right?

Lex Fridman (34:55.800)

So it's not just this, you, you see a curve and you're blindly drawing a line through

Grant Sanderson (35:00.320)

it.

Lex Fridman (35:01.320)

What's much more interesting is to ask, when do you have this proportional growth property?

Grant Sanderson (35:04.960)

Um, because then you can also recognize when it breaks down, like in an epidemic, as you

Lex Fridman (35:09.520)

approach saturation, that would break down.

Grant Sanderson (35:11.640)

Um, as you do anything that, uh, skews what that proportionality constant is, um, you

Grant Sanderson (35:16.520)

can make it maybe not break down as being an exponential, but it can seriously slow

Lex Fridman (35:20.020)

what that exponential rate is.

Grant Sanderson (35:21.480)

This is the opposite of a pandemic is you want, in terms of ideas, you want to minimize

Grant Sanderson (35:28.960)

barriers that, um, prevent the spread.

Lex Fridman (35:31.680)

You want to maximize the spread of impact.

Lex Fridman (35:33.620)

So like you want it to, to grow when you're doing technological development is so that

Lex Fridman (35:38.720)

you do hold up that rate holds up.

Lex Fridman (35:42.240)

And that's, that's almost like, uh, like an operational challenge of like how you run

Grant Sanderson (35:47.400)

a company, how you run a group of people is that any one invention has a ripple that's

Grant Sanderson (35:52.640)

unstopped.

Lex Fridman (35:54.440)

And that ripple effect then has its own ripple effects and so on.

Lex Fridman (35:58.000)

And that continues.

Lex Fridman (35:59.000)

Yeah.

Grant Sanderson (36:00.000)

Like Moore's law is fascinating.

Lex Fridman (36:01.000)

And the, like on a psychological level and a human level, cause it's not exponential.

Grant Sanderson (36:06.760)

It's, it's just a consistent set of like what you would call like S curves, which is like,

Lex Fridman (36:13.400)

it's constantly like breakthrough innovations nonstop.

Grant Sanderson (36:18.120)

That's a good point.

Grant Sanderson (36:19.120)

Like it might not actually be an example of exponentials because of something which grows

Grant Sanderson (36:22.680)

in proportion to itself.

Lex Fridman (36:23.920)

But instead it's almost like a benchmark that was set out that everyone's been pressured

Grant Sanderson (36:27.880)

to meet.

Lex Fridman (36:29.000)

And it's like all these innovations and micro inventions along the way, rather than some

Grant Sanderson (36:33.840)

consistent sit back and just let the lily pad grow across the lake phenomenon.

Lex Fridman (36:38.480)

And it's also that there's a human psychological level for sure of like the four minute mile,

Grant Sanderson (36:42.600)

like it's something about it.

Lex Fridman (36:45.080)

Like saying that, look, there is, you know, Moore's law, it's a law.

Lex Fridman (36:51.800)

So like it's a, it's certainly an achievable thing.

Grant Sanderson (36:56.720)

You know, we achieved it for the last decade, for the last two decades, for the last three

Grant Sanderson (36:59.920)

decades, you just keep going and it somehow makes it happen.

Grant Sanderson (37:04.320)

I mean, it makes people, I'm continuously surprised in this world how few people do

Grant Sanderson (37:10.520)

the best work in the world, like in that particular, whatever that field is, like it's very often

Grant Sanderson (37:18.200)

that like the genius, I mean, you couldn't argue that community matters, but it's certain

Grant Sanderson (37:25.480)

like I've been in groups of engineers where like one person is clearly like doing an incredible

Grant Sanderson (37:32.080)

amount of work and just is the genius and it's fascinating to see basically it's kind

Grant Sanderson (37:38.880)

of the Steve Jobs idea is maybe the whole point is to create an atmosphere where the

Grant Sanderson (37:46.840)

genius can discover themselves, like have the opportunity to do the best work of their

Grant Sanderson (37:52.080)

life and yeah, and that the exponential is just milking that.

Grant Sanderson (37:57.920)

It's like rippling the idea that it's possible and that idea that it's possible finds the

Grant Sanderson (38:03.120)

right people for the four minute mile and the idea that it's possible finds the right

Grant Sanderson (38:07.200)

runners to run it and then expose the number of people who can run faster than four minutes.

Grant Sanderson (38:12.520)

It's kind of interesting to, I don't know, basically the positive way to see that is

Grant Sanderson (38:17.140)

most of us are way more intelligent, have way more potential than we ever realized.

Grant Sanderson (38:22.320)

I guess that's kind of depressing, but I mean like the ceiling for most of us is much higher

Lex Fridman (38:26.520)

than we ever realized.

Grant Sanderson (38:28.640)

That is true.

Grant Sanderson (38:29.640)

A good book to read if you want that sense is Peak, which essentially talks about peak

Grant Sanderson (38:34.740)

performance in a lot of different ways, like chess, London cab drivers, how many pushups

Grant Sanderson (38:39.200)

people can do, short term memory tasks, and it's meant to be like a concrete manifesto

Grant Sanderson (38:45.600)

about deliberate practice and such, but the one sensation you come out with is wow, no

Grant Sanderson (38:50.480)

matter how good people are at something, they can get better and like way better than we

Grant Sanderson (38:54.560)

think they could.

Grant Sanderson (38:55.680)

I don't know if that's actually related to exponential growth, but I do think it's a

Grant Sanderson (38:59.480)

true phenomenon that's interesting.

Grant Sanderson (39:01.440)

Yeah, I mean, there's certainly no law of exponential growth in human innovation.

Grant Sanderson (39:06.720)

Well, I don't know.

Lex Fridman (39:09.920)

Well kind of, there is.

Grant Sanderson (39:11.480)

I think it's very interesting to see when innovations in one field allow for innovations

Lex Fridman (39:15.200)

in another.

Grant Sanderson (39:16.200)

Like the advent of computing seems like a prerequisite for the advent of chaos theory.

Lex Fridman (39:20.440)

You have this truth about physics and the world that in theory could be known.

Grant Sanderson (39:25.120)

You could find Lorenz's equations without computers, but in practice, it was just never

Grant Sanderson (39:29.960)

going to be analyzed that way unless you were doing like a bunch of simulations and that

Grant Sanderson (39:34.200)

you could computationally see these models.

Lex Fridman (39:36.520)

So it's like physics allowed for computers, computers allowed for better physics, and

Grant Sanderson (39:40.360)

you know, wash, rinse and repeat.

Lex Fridman (39:42.500)

That self proportionality, that's exponential.

Lex Fridman (39:45.720)

So I think I wouldn't think it's too far to say that that's a law of some kind.

Grant Sanderson (39:50.920)

Yeah, a fundamental law of the universe is that these descendants of apes will exponentially

Grant Sanderson (39:59.760)

improve their technology and one day be taken over by the AGI.

Lex Fridman (40:05.280)

That's built in.

Grant Sanderson (40:06.280)

That'll make the video game fun, whoever created this thing.

Lex Fridman (40:09.640)

So I mean, since you're wearing a space X shirt, let me ask.

Grant Sanderson (40:12.960)

I didn't realize I was wearing a space X shirt.

Lex Fridman (40:15.960)

I apologize.

Grant Sanderson (40:16.960)

It's on point.

Lex Fridman (40:17.960)

So it's on topic.

Grant Sanderson (40:18.960)

I'll take it.

Grant Sanderson (40:19.960)

It's the first crewed mission out into space since the space shuttle and just by first

Grant Sanderson (40:32.600)

time ever by a commercial company, I mean, it's an incredible accomplishment, I think,

Lex Fridman (40:37.880)

but it's also just an incredible, it inspires imagination amongst people that this is the

Grant Sanderson (40:44.560)

first step in a long, vibrant journey of humans into space.

Lex Fridman (40:50.960)

So how do you feel?

Lex Fridman (40:52.800)

Is this exciting to you?

Lex Fridman (40:55.000)

Yeah, it is.

Grant Sanderson (40:56.000)

I think it's great.

Grant Sanderson (40:57.000)

The idea of seeing it basically done by smaller entities instead of by governments.

Grant Sanderson (41:00.360)

I mean, it's a heavy collaboration between space X and NASA in this case, but moving

Grant Sanderson (41:04.720)

in the direction of not necessarily requiring an entire country and its government to make

Grant Sanderson (41:09.340)

it happen, but that you can have something closer to a single company doing it.

Grant Sanderson (41:14.840)

We're not there yet because it's not like they're unilaterally saying like we're just

Grant Sanderson (41:18.320)

shooting people up into space.

Grant Sanderson (41:20.960)

It's just a sign that we're able to do more powerful things with smaller groups of people.

Grant Sanderson (41:25.680)

I find that inspiring.

Lex Fridman (41:26.680)

Innovate quickly.

Grant Sanderson (41:27.680)

I hope we see people land on Mars in my lifetime.

Lex Fridman (41:30.800)

Do you think we will?

Grant Sanderson (41:32.360)

I think so.

Lex Fridman (41:33.360)

I mean, I think there's a ton of challenges there, right?

Grant Sanderson (41:35.600)

Like radiation being kind of the biggest one.

Lex Fridman (41:37.720)

And I think there's a ton of people who look at that and say, why?

Lex Fridman (41:42.160)

Why would you want to do that?

Lex Fridman (41:43.880)

Let's let the robots do the science for us.

Lex Fridman (41:46.100)

But I think there's enough people who are genuinely inspired about broadening the worlds

Grant Sanderson (41:50.520)

that we've touched or people who think about things like backing up the light of consciousness

Grant Sanderson (41:55.800)

with super long term visions of terraforming, like as long as there's a backing up the

Lex Fridman (41:59.800)

light of consciousness.

Grant Sanderson (42:00.800)

Yeah.

Lex Fridman (42:01.800)

I thought that if Earth goes to hell, we've got to have a backup somewhere.

Grant Sanderson (42:07.800)

A lot of people see that as pretty out there and it's like not in the short term future,

Lex Fridman (42:10.800)

but I think that's an inspiring thought.

Grant Sanderson (42:12.680)

I think that's a reason to get up in the morning and I feel like most employees at SpaceX feel

Lex Fridman (42:17.160)

that way too.

Lex Fridman (42:18.160)

Do you think we'll colonize Mars one day?

Lex Fridman (42:21.360)

No idea.

Grant Sanderson (42:22.360)

Like either AGI kills us first or if we're like allowed, I don't know if it'll take us

Lex Fridman (42:26.080)

for allowed.

Grant Sanderson (42:27.080)

Well, like honestly, it would take such a long time.

Lex Fridman (42:29.960)

Like, okay, you might have a small colony, right?

Grant Sanderson (42:32.760)

Something like what you see in the Martian, but not like people living comfortably there.

Lex Fridman (42:39.840)

But if you want to talk about actual like second Earth kind of stuff, that's just like

Grant Sanderson (42:46.200)

way far out there and the future moves so fast that it's hard to predict.

Lex Fridman (42:50.680)

We might just kill ourselves before that even becomes viable.

Grant Sanderson (42:53.520)

Yeah.

Grant Sanderson (42:54.520)

I mean, there's a lot of possibilities where it could be just, it doesn't have to be on

Grant Sanderson (42:58.040)

a planet, we could be floating out in space, have a space faring backup solution that doesn't

Grant Sanderson (43:07.360)

have to deal with the constraints that a planet, I mean, a planet provides a lot of possibilities

Lex Fridman (43:11.440)

and resources, but also has some constraints.

Lex Fridman (43:13.800)

Yeah.

Grant Sanderson (43:14.800)

I mean, for me, for some reason, it's a deeply exciting possibility.

Lex Fridman (43:19.560)

Oh yeah.

Grant Sanderson (43:20.560)

Yeah.

Grant Sanderson (43:21.560)

All of the people who were like skeptical about it are like, why do we care about going

Lex Fridman (43:24.720)

to Mars?

Lex Fridman (43:25.720)

Like, what makes you care about anything that's inspiring?

Grant Sanderson (43:28.640)

It's hard.

Grant Sanderson (43:29.640)

It actually is hard to hear that because exactly as you put it on a philosophical level, it's

Lex Fridman (43:34.760)

hard to say, why do anything?

Lex Fridman (43:37.600)

I don't know.

Grant Sanderson (43:38.740)

It's like the people say like, I've been doing like an insane challenge last 30 something

Lex Fridman (43:45.560)

days.

Lex Fridman (43:46.560)

Your pull ups?

Lex Fridman (43:47.560)

The pull ups and push ups and like, a bunch of people are like, awesome.

Grant Sanderson (43:55.080)

They're insane, but awesome.

Lex Fridman (43:57.040)

And then some people are like, why?

Lex Fridman (43:59.520)

Why do anything?

Lex Fridman (44:00.720)

I don't know.

Grant Sanderson (44:02.720)

There's a calling.

Grant Sanderson (44:03.720)

It's, I'm with JFK a little bit is because we do these things because they're hard.

Grant Sanderson (44:09.720)

There's something in the human spirit that says like, same with like a math problem.

Grant Sanderson (44:14.120)

There's something you fail once and it's like this feeling that, you know what, I'm not

Grant Sanderson (44:20.240)

going to back down from this.

Lex Fridman (44:21.800)

There's something to be discovered in overcoming this thing.

Lex Fridman (44:25.320)

So what I like about it is, and I also like this about the moon missions, sure, it's kind

Lex Fridman (44:29.560)

of arbitrary, but you can't move the target.

Lex Fridman (44:32.020)

So you can't make it easier and say that you've accomplished the goal.

Lex Fridman (44:36.120)

And when that happens, it just demands actual innovation, right?

Grant Sanderson (44:39.640)

Like protecting humans from the radiation in space on the flight there while they're

Lex Fridman (44:44.720)

hard problem demands innovation.

Grant Sanderson (44:46.820)

You can't move the goalpost to make that easier.

Lex Fridman (44:49.500)

But certainly the innovations required for things like that will be relevant in a bunch

Grant Sanderson (44:52.560)

of other domains too.

Lex Fridman (44:54.820)

So like the idea of doing something merely because it's hard, it's like loosely productive.

Grant Sanderson (44:59.280)

Great.

Lex Fridman (45:00.280)

But as long as you can't move the goalposts, there's probably going to be these secondary

Grant Sanderson (45:03.240)

benefits that like we should all strive for.

Lex Fridman (45:07.000)

Yeah.

Grant Sanderson (45:08.000)

I mean, it's hard to formulate the Mars colonization problem as something that has a deadline,

Lex Fridman (45:13.280)

which is the problem.

Lex Fridman (45:15.360)

But if there was a deadline, then the amount of things we would come up with by forcing

Lex Fridman (45:23.360)

ourselves to figure out how to colonize that place would be just incredible.

Grant Sanderson (45:29.100)

This is what people, like the internet didn't get created because people sat down and try

Grant Sanderson (45:34.280)

to figure out how do I, you know, send TikTok videos of myself dancing to people.

Grant Sanderson (45:41.480)

They, you know, it was, there's an application.

Grant Sanderson (45:44.880)

I mean, actually I don't even know what do you think the application for the internet

Grant Sanderson (45:48.120)

was when it was, it must've been very low level basic network communication within DARPA,

Grant Sanderson (45:53.360)

like military based, like how do I send like a networking, how do I send information securely

Lex Fridman (46:00.600)

between two places?

Lex Fridman (46:03.040)

Maybe it was an encryption.

Grant Sanderson (46:04.040)

I'm totally speaking totally outside of my knowledge, but like it was probably intended

Lex Fridman (46:08.160)

for a very narrow, small group of people.

Grant Sanderson (46:10.200)

Well, so I mean, it was, there was like this small community of people who are really interested

Grant Sanderson (46:13.880)

in timesharing computing and like interactive computing in contrast with a batch processing.

Lex Fridman (46:20.000)

And then the idea that as you set up like a timesharing center, basically meaning kind

Grant Sanderson (46:24.600)

of multiple people like logged in and using that like central computer, why not make it

Lex Fridman (46:28.820)

accessible to others?

Lex Fridman (46:30.360)

And this was kind of what I had always thought like, Oh, is this like fringe group that was

Grant Sanderson (46:33.880)

interested in this new kind of computing and they all like got themselves together.

Lex Fridman (46:37.820)

But the thing is like DARPA wouldn't act, you wouldn't have the U S government funding

Lex Fridman (46:41.360)

that just for the funds of it, right?

Grant Sanderson (46:44.020)

In some sense, that's what ARPA was all about was like just really advanced research for

Grant Sanderson (46:48.880)

the sake of having advanced research and it doesn't have to pay out with utility soon.

Lex Fridman (46:53.400)

But the core parts of its development were happening like in the middle of the Vietnam

Grant Sanderson (46:57.480)

war when there was budgetary constraints all over the place.

Grant Sanderson (47:01.360)

I only learned this recently, actually, like if you look at the documents, basically justifying

Grant Sanderson (47:06.400)

the budget for the ARPANET as they were developing it, and not just keeping it where it was,

Lex Fridman (47:12.840)

but actively growing it while all sorts of other departments were having their funding

Grant Sanderson (47:15.880)

cut because of the war, a big part of it was national defense in terms of having like a

Grant Sanderson (47:21.440)

more robust communication system, like the idea of packet switching versus circuit switching.

Grant Sanderson (47:26.500)

You could kind of make this case that in some calamitous circumstance where a central location

Grant Sanderson (47:31.000)

gets nuked, this is a much more resilient way to still have your communication lines

Grant Sanderson (47:36.360)

that like traditional telephone lines weren't as resilient to, which I just found very interesting.

Grant Sanderson (47:45.560)

Even something that we see as so happy go lucky is just a bunch of computer nerds trying

Grant Sanderson (47:48.760)

to get like interactive computing out there.

Grant Sanderson (47:51.060)

The actual thing that made it funded and thing that made it advance when it did was because

Grant Sanderson (47:57.480)

of this direct national security question and concern.

Lex Fridman (48:00.880)

I don't know if you've read it.

Grant Sanderson (48:02.360)

I haven't read it.

Grant Sanderson (48:03.360)

I don't know if I've been meaning to read it, but Neil deGrasse Tyson actually came out

Grant Sanderson (48:06.320)

with a book that talks about like science in the context of the military, like basically

Grant Sanderson (48:11.640)

saying all the great science we've done in the 20th century was like because of the military.

Grant Sanderson (48:18.640)

He paints a positive, it's not like a critical, a lot of people say like military industrial

Lex Fridman (48:23.880)

complex and so on.

Grant Sanderson (48:25.880)

Another way to see the military and national security is like a source of, like you said,

Grant Sanderson (48:30.400)

headlines and like hard things you can't move, like almost like scaring yourself into being

Grant Sanderson (48:37.200)

productive.

Lex Fridman (48:38.200)

It is that.

Grant Sanderson (48:39.200)

I mean, Manhattan Project is a perfect example, probably the quintessential example.

Grant Sanderson (48:43.240)

That one is a little bit more macabre than others because of like what they were building,

Lex Fridman (48:48.440)

but in terms of how many focused, smart hours of human intelligence get pointed towards

Lex Fridman (48:54.800)

a topic per day, you're just maxing it out with that sense of worry.

Grant Sanderson (48:58.580)

In that context, everyone there was saying like, we've got to get the bomb before Hitler

Grant Sanderson (49:01.600)

does and that just lights a fire under you that I, again, like the circumstances macabre,

Lex Fridman (49:08.340)

but I think that's actually pretty healthy, especially for researchers that are otherwise

Grant Sanderson (49:11.540)

going to be really theoretical to take these like theorizers and say, make this real physical

Grant Sanderson (49:17.040)

thing happen.

Grant Sanderson (49:18.640)

Meaning a lot of it is going to be unsexy, a lot of it's going to be like young Feynman

Grant Sanderson (49:22.380)

sitting there kind of inventing a notion of computation in order to like compute what

Grant Sanderson (49:28.120)

they needed to compute more quickly with like the rudimentary automated tools that they

Grant Sanderson (49:31.940)

had available.

Grant Sanderson (49:34.280)

I think you see this with Bell Labs also where you've got otherwise very theorizing minds

Grant Sanderson (49:39.460)

in very pragmatic contexts that I think is like really helpful for the theory as well

Lex Fridman (49:43.600)

as for the applications.

Grant Sanderson (49:46.280)

I think that stuff can be positive for progress.

Lex Fridman (49:50.280)

You mentioned Bell Labs and Manhattan Project.

Grant Sanderson (49:52.760)

This kind of makes me curious for the things you've create, which are quite singular.

Grant Sanderson (49:58.240)

Like if you look at all YouTube or just not YouTube, it doesn't matter what it is.

Grant Sanderson (50:04.040)

It's just teaching content, art, it doesn't matter.

Lex Fridman (50:06.920)

It's like, yep, that's, that's grant, right?

Grant Sanderson (50:11.620)

That's unique.

Lex Fridman (50:12.620)

I know you're teaching style and everything.

Grant Sanderson (50:15.700)

Does it, Manhattan Project and Bell Labs was like famously a lot of brilliant people, but

Lex Fridman (50:22.000)

there's a lot of them.

Grant Sanderson (50:23.480)

They play off of each other.

Lex Fridman (50:25.200)

So like my question for you is that, does it get lonely?

Grant Sanderson (50:28.880)

Honestly, that right there, I think is the biggest part of my life that I would like

Grant Sanderson (50:32.640)

to change in some way that I look at a Bell Labs type situation and I'm like, God damn,

Grant Sanderson (50:39.640)

I love that whole situation and I'm so jealous of it and you're like reading about Hamming

Lex Fridman (50:44.160)

and then you see that he also shared an office with Shannon and you're like, of course he

Grant Sanderson (50:47.480)

did.

Lex Fridman (50:48.480)

Of course they shared an office.

Grant Sanderson (50:49.480)

That's how these ideas get.

Lex Fridman (50:50.480)

And they actually probably very likely worked separately.

Grant Sanderson (50:53.320)

Yeah, totally, totally separate.

Lex Fridman (50:55.360)

But there's a literally, and sorry to interrupt, there's a literally magic that happens when

Grant Sanderson (50:59.720)

you run into each other, like on the way to like getting a snack or something.

Grant Sanderson (51:06.480)

Conversations you overhear, it's other projects you're pulled into, it's like puzzles that

Grant Sanderson (51:09.320)

colleagues are sharing, like all of that.

Grant Sanderson (51:12.240)

I have some extent of it just because I try to stay well connected in communities of people

Grant Sanderson (51:17.960)

who think in similar ways.

Lex Fridman (51:19.720)

But it's not in the day to day in the same way, which I would like to fix somehow.

Grant Sanderson (51:25.120)

That's one of the, I would say one of the biggest, well, one of the many drawbacks,

Grant Sanderson (51:35.040)

negative things about this current pandemic is that whatever the term is, but like chance

Grant Sanderson (51:40.920)

collisions are significantly reduced.

Grant Sanderson (51:44.160)

I saw, I don't know why I saw this, but on my brother's work calendar, he had a scheduled

Grant Sanderson (51:50.440)

slot with someone that he scheduled a meeting and the title of the whole meeting was no

Lex Fridman (51:57.240)

specific agenda.

Grant Sanderson (51:58.240)

I just missed the happenstance serendipitous conversations that we used to have, which

Lex Fridman (52:02.200)

the pandemic and remote work has so cruelly taken away from us.

Grant Sanderson (52:05.080)

Brilliant.

Lex Fridman (52:06.080)

That's brilliant.

Grant Sanderson (52:07.080)

I'm like, that's the way to do it.

Lex Fridman (52:09.720)

You just schedule those things, schedule the serendipitous interaction.

Grant Sanderson (52:13.680)

That's like, I mean, you can't do it in an academic setting, but it's basically like

Grant Sanderson (52:16.400)

going to a bar and sitting there just for the strangers you might meet, just the strangers

Grant Sanderson (52:22.080)

or striking up a conversation with strangers on the train.

Grant Sanderson (52:26.280)

Harder to do when you're deeply like maybe myself or maybe a lot of academic types who

Grant Sanderson (52:33.820)

are like introverted and avoid human contact as much as possible.

Lex Fridman (52:38.440)

So it's nice when it's forced, those chance collisions, but maybe scheduling is a possibility

Lex Fridman (52:43.600)

but for the most part, do you work alone?

Lex Fridman (52:48.000)

I'm sure you struggle a lot.

Grant Sanderson (52:53.660)

You probably hit moments when you look at this and you say like, this is the wrong way

Lex Fridman (53:00.200)

to show it.

Grant Sanderson (53:01.200)

It's a long way to visualize it.

Lex Fridman (53:02.440)

I'm making it too hard for myself.

Grant Sanderson (53:04.160)

I'm going down the wrong direction.

Lex Fridman (53:05.740)

This is too long.

Grant Sanderson (53:06.740)

This is too short.

Lex Fridman (53:07.740)

All those self doubt that could be paralyzing.

Lex Fridman (53:11.360)

What do you do in those moments?

Lex Fridman (53:13.160)

I actually much prefer like work to be a solitary affair for me.

Grant Sanderson (53:18.720)

That's like a personality quirk.

Grant Sanderson (53:19.880)

I would like it to be in an environment with others and like collaborative in the sense

Grant Sanderson (53:23.280)

of ideas exchanged.

Lex Fridman (53:24.700)

But those phenomena you're describing when you say this is too long, this is too short,

Grant Sanderson (53:27.980)

this visualization sucks, it's way easier to say that to yourself than it is to say

Lex Fridman (53:31.920)

to a collaborator.

Lex Fridman (53:34.300)

And I know that's just a thing that I'm not good at.

Lex Fridman (53:36.440)

So in that way, it's very easy to just throw away a script because the script isn't working.

Grant Sanderson (53:41.200)

It's hard to tell someone else they should do the same.

Grant Sanderson (53:43.400)

Actually last time we talked, I think it was like very close to me talking Don Knuth was

Grant Sanderson (53:47.560)

kind of cool.

Lex Fridman (53:48.560)

Like two people that...

Grant Sanderson (53:49.560)

I can't believe you got that interview.

Lex Fridman (53:51.560)

It's the hard...

Lex Fridman (53:52.560)

No, can I brag about something?

Lex Fridman (53:54.960)

Please.

Grant Sanderson (53:55.960)

My favorite thing is Don Knuth, after did the interview, he offered to go out to hot

Lex Fridman (54:01.520)

dogs with me.

Grant Sanderson (54:02.520)

To get hot dogs.

Lex Fridman (54:03.520)

That was never...

Grant Sanderson (54:04.520)

Like people ask me what's the favorite interview you've ever done and that has to be...

Lex Fridman (54:09.480)

But unfortunately I couldn't, I had a thing after.

Lex Fridman (54:12.760)

So I had to turn down Don Knuth.

Lex Fridman (54:14.440)

You missed Knuth dogs?

Grant Sanderson (54:15.440)

Knuth dogs.

Lex Fridman (54:16.440)

Sorry.

Lex Fridman (54:17.440)

So that was a little bragging, but the hot dogs, he's such a sweet.

Lex Fridman (54:22.200)

But the reason I bring that up is he works through problems alone as well.

Grant Sanderson (54:28.000)

He prefers that struggle, the struggle of it.

Grant Sanderson (54:33.360)

Writers like Stephen King often talk about their process of what they do, what they eat

Grant Sanderson (54:40.720)

when they wake up, when they sit down, how they like their desk on a perfectly productive

Lex Fridman (54:49.280)

day.

Lex Fridman (54:51.680)

What they like to do, how long they like to work for, what enables them to think deeply,

Lex Fridman (54:56.600)

all that kind of stuff.

Grant Sanderson (54:57.600)

Hunter S. Thompson did a lot of drugs.

Lex Fridman (55:00.520)

Everybody has their own thing.

Lex Fridman (55:03.000)

Do you have a thing?

Grant Sanderson (55:05.880)

If you were to lay out a perfect productive day, what would that schedule look like do

Lex Fridman (55:10.720)

you think?

Grant Sanderson (55:11.720)

Part of that's hard to answer because like the mode of work I do changes a lot from day

Grant Sanderson (55:19.360)

to day.

Lex Fridman (55:20.360)

Like some days I'm writing.

Grant Sanderson (55:21.400)

The thing I have to do is write a script.

Lex Fridman (55:22.900)

Some days I'm animating.

Grant Sanderson (55:23.900)

The thing I have to do is animate.

Lex Fridman (55:24.900)

Sometimes I'm like working on the animation library.

Grant Sanderson (55:27.160)

The thing I have to do is like a little, I'm not a software engineer, but something in

Lex Fridman (55:30.720)

the direction of software engineering.

Grant Sanderson (55:32.560)

Some days it's like a variant of research.

Lex Fridman (55:34.520)

It's like learn this topic well and try to learn it differently.

Lex Fridman (55:37.620)

So those are like four very different modes.

Grant Sanderson (55:41.520)

Some days it's like get through the email backlog of people I've been, tasks I've been

Grant Sanderson (55:45.360)

putting off.

Grant Sanderson (55:46.360)

It goes research, scripting, like the idea starts with research and then there's scripting

Lex Fridman (55:51.960)

and then there's programming and then there's the showtime.

Lex Fridman (55:56.800)

And the research side, by the way, I think a problematic way to do it is to say I'm starting

Grant Sanderson (56:01.440)

this project and therefore I'm starting the research.

Grant Sanderson (56:03.960)

Instead it should be that you're like ambiently learning a ton of things just in the background

Lex Fridman (56:08.020)

and then once you feel like you have the understanding for one, you put it on the list of things

Lex Fridman (56:11.660)

that there can be a video for.

Grant Sanderson (56:14.040)

Otherwise either you're going to end up roadblocked forever or you're just not going to like have

Lex Fridman (56:18.660)

a good way of talking about it.

Lex Fridman (56:21.920)

But still some of the days it's like the thing to do is learn new things.

Lex Fridman (56:25.120)

So what's the most painful one?

Grant Sanderson (56:26.360)

I think you mentioned scripting.

Lex Fridman (56:29.120)

Scripting is yeah, that's the worst.

Grant Sanderson (56:30.520)

Yeah, writing is the worst.

Lex Fridman (56:31.880)

So what's your, on a perfectly, so let's take the hardest one.

Lex Fridman (56:35.520)

What's a perfectly productive day?

Grant Sanderson (56:37.320)

You wake up and it's like, damn it, this is the day I need to do some scripting.

Lex Fridman (56:41.800)

And like you didn't do anything the last two days so you came up with excuses to procrastinate

Lex Fridman (56:45.600)

so today must be the day.

Grant Sanderson (56:47.480)

Yeah, I wake up early, I guess I exercise and then I turn the internet off.

Grant Sanderson (56:57.100)

If we're writing, yeah, that's what's required is having the internet off and then maybe

Grant Sanderson (57:01.520)

you keep notes on the things that you want to Google when you're allowed to have the

Lex Fridman (57:04.200)

internet again.

Grant Sanderson (57:05.200)

I'm not great about doing that, but when I do, that makes it happen.

Lex Fridman (57:08.820)

And then when I hit writer's block, like the solution to writer's block is to read.

Grant Sanderson (57:13.200)

Doesn't even have to be related.

Grant Sanderson (57:14.360)

Just read something different just for like 15 minutes, half an hour and then go back

Grant Sanderson (57:18.340)

to writing.

Lex Fridman (57:20.160)

That when it's a nice cycle, I think can work very well.

Lex Fridman (57:22.520)

And when you're writing the script, you don't know where it ends, right?

Lex Fridman (57:26.200)

Like you have a problem solving videos.

Grant Sanderson (57:28.820)

I know where it ends, expositional videos.

Grant Sanderson (57:30.840)

I don't know where it ends coming up with a, with the magical thing that makes this

Lex Fridman (57:36.760)

whole story, like ties this whole story together that when does that happen?

Grant Sanderson (57:41.980)

That's that's the thing that makes it such that a topic gets put on the list of like

Grant Sanderson (57:45.120)

videos.

Lex Fridman (57:46.120)

Oh, that's an issue.

Grant Sanderson (57:47.120)

You shouldn't start the project unless there's one of those and you have, you have so many

Grant Sanderson (57:50.760)

nice bags that you haven't such a big bag of aha moments already that you could just

Grant Sanderson (57:55.040)

pull at it.

Lex Fridman (57:56.880)

That's one of the things.

Lex Fridman (57:57.960)

And one of the sad things about time and that nothing lasts forever and that we're all mortal.

Grant Sanderson (58:05.400)

Let's not get into that discussion is, you know, if I see like, even when I ask for people

Grant Sanderson (58:14.660)

to ask, like ask, I did a call for questions and people want to ask you questions and so

Lex Fridman (58:19.600)

many requests from people about like certain videos they would love you to do.

Grant Sanderson (58:23.520)

It's such a pile and I think that's a, that's a sign of like admiration from people for

Lex Fridman (58:30.480)

sure.

Lex Fridman (58:31.480)

But it's like, it makes me sad cause like whenever I see them, people give ideas, they're

Lex Fridman (58:35.200)

all like very often really good ideas.

Lex Fridman (58:38.640)

And it's like, it's such a, it makes me sad in the same kind of way when I go through

Grant Sanderson (58:44.960)

a library or through a bookstore, you see all these amazing books that you'll never

Grant Sanderson (58:49.160)

get to open.

Lex Fridman (58:52.720)

So yeah.

Lex Fridman (58:53.720)

So you gotta enjoy the ones that you have, enjoy the books that are open and don't let

Lex Fridman (58:59.200)

yourself lament the ones that stay closed.

Lex Fridman (59:02.960)

What else?

Lex Fridman (59:03.960)

Is there any other magic to that day?

Lex Fridman (59:05.120)

So do you try to dedicate like a certain number of hours?

Lex Fridman (59:08.720)

Do you, Cal Newport has this deep work kind of idea.

Grant Sanderson (59:13.160)

There's systematic people who like get really on top of, you know, they checklist of what

Lex Fridman (59:17.640)

they're going to do in the day and they like count their hours.

Lex Fridman (59:20.280)

And I am not a systematic person in that way.

Lex Fridman (59:23.280)

Which is probably a problem.

Grant Sanderson (59:24.560)

I very likely would get more done if I was systematic in that way, but that doesn't happen.

Lex Fridman (59:30.800)

So you talk to me later in life and maybe I'll have like changed my ways and give you

Grant Sanderson (59:35.920)

a very different answer.

Grant Sanderson (59:37.360)

I think Benjamin Franklin like later in life figured out the rigor is these like very rigorous

Grant Sanderson (59:42.760)

schedules and how to be productive.

Lex Fridman (59:45.000)

I think those schedules are much more fun to write.

Grant Sanderson (59:47.640)

Like it's very fun to like write a schedule and make a blog post about like the perfect

Lex Fridman (59:50.840)

productive day that like might work for one person.

Lex Fridman (59:54.360)

But I don't know how much people get out of like reading them or trying to adopt someone

Lex Fridman (59:57.820)

else's style.

Lex Fridman (59:59.160)

And I'm not even sure that they've ever followed.