0:01
Most people unfortunately are first-order thinkers.
Search this video
0:02
They very rarely take secondary and tertiary effects into formula for thinking about something.
0:10
When you design protocols, if they're successful, you might not be able to change them, so you want to be extremely mindful of designing them in the right way and thinking about, you know, the second-order consequences of doing so.
0:21
Complex problems evolve over time, so you never truly solve them.
0:23
You have to kind of like stay in a loop of constantly paying attention to them and like solving them.
0:29
We basically funded 30-ish people to explore protocols across all of their fields of experience.
0:37
Civilization progresses by increasing the number of actions we can do without thinking about them.
0:42
You have to think with like 100, 200-year timelines, which means that you are thinking about a horizon over which cultural values will change.
0:52
Technologies don't get adopted because the proponents prove right and the critics prove wrong.
0:56
It's because the critics basically die out and the proponents die. That is it.
1:02
It advances one funeral at a time. Well, hello everyone.
1:09
It's Jim O'Shaughnessy with yet another Infinite Loops very special day today.
1:14
I have my principal researcher and writer Ed William joining me as a podcast because I am dealing with two exceptionally smart people and I just didn't think I would have the brainpower to do it alone.
1:30
My guests today are Tim Beiko and Venkatesh Rao, who have been doing something really, really cool over the summer.
1:38
Tim is uh the head of the core protocol meetings for Ethereum and Venkatesh is a writer extraordinaire who writes for the Ribbon Farm blog and consults.
1:51
You guys might remember our chat with Venkatesh about his uh gig economy series.
2:00
But now we're going to be talking about their summer of protocols.
2:03
Is that like the summer of George, guys?
2:11
We're trying to like keep it a cross between summer of George and summer of code. I love it. I love it.
2:17
So, I think this is super cool. I read all about it.
2:24
You funded a bunch of researchers to be able to connect the conversations happening around Ethereum protocols with other areas.
2:31
So, gents, tell us, what what is a summer of protocols?
2:38
What are protocols for our audience who haven't read the background material and and why should we care?
2:47
Um yeah, maybe I can I can get started here and and Banket you can add some color after, but uh the the idea for this idea for this came a while back where um I as you mentioned, I work on the Ethereum protocol and it's kind of a weird uh it's it's a weird thing to work on, right?
3:05
It's not quite a product, um it's not quite like a technology platform, um it's kind of this new um new protocol for lack of a better word that we've created and um about a year year and a half ago, I was wondering, you know, how should we go about stewarding this thing?
3:22
How should we think about, you know, what the evolution over the next 10 years looks like and I reached out to Banket because I saw from his newsletter he was starting to spend more and more time thinking about blockchains.
3:34
So, I asked him, you know, do you want to help me think through this and and and kind of the questions here.
3:39
So, we had a couple calls about it and and then my my takeaway from that was, you know, this is already valuable, but like me and Banket are like way too narrow to do this.
3:49
So, I asked Banket, do you want to come and talk with some other people at the Ethereum Foundation and help us, you know, more as an organization think through this.
3:56
And we spent about 6 months doing that.
3:58
And our takeaway after those 6 months was was kind of the same thing where it was still too narrow looking at just the problems of Ethereum.
4:05
We you keep running into these analogies where, you know, you'll be saying, "Oh, Ethereum is just like a company or it's just like a country or it's just like a tech platform."
4:15
Um and that helps you get, you know, a bit of understanding, but these analogies fall apart pretty quickly as you start to get into the weeds.
4:22
And so we we came up with this idea that you know, Ethereum is not the only the only context in which protocols are used.
4:30
There's obviously protocols a bunch of at across a ton of other domains.
4:33
And so maybe the way to approach this was not to just look at Ethereum specifically, but to try to understand what protocols are.
4:39
Are there recurring problems?
4:41
Are there recurring questions, you know, that show up in blockchains?
4:44
Yes, but that maybe show up in architecture, that maybe show up in healthcare, that maybe show up in you know, workplace safety.
4:50
And the idea was that if we find some problems that resonated across all the all those domains or, you know, some better frameworks or some better insights, then they're probably sound enough that we can go and apply them back.
5:01
Obviously, in our case there's things like Ethereum, but hopefully in a bunch of other people's cases, um in their respective domains.
5:08
And that's kind of where the the the seed started for for cyber protocols.
5:12
Um and then from there, uh we basically funded 30-ish people to yeah, explore protocols across all of all of their sort of fields of experience and and put put all of their findings together.
5:28
Um and and Venkat was the person who effectively led that research program uh over the past uh several months.
5:36
Yeah, yeah, I think that covers most of the ground.
5:38
I I like to add that since you asked it, like, what is protocols and why should we care?
5:41
Actually, when I came on um this podcast last, and we talked about the gig economy, it's a very good example of what protocols are and why we should care.
5:51
Like, you know, we talk about corporations and organizations and being employees of corporations, those are kind of like industrial age platforms for organizing the labor economy, but it's very hard to define what the gig economy is, but if you think in terms of protocols, it becomes slightly clearer.
6:07
The gig economy is a set of interacting protocols that allows somebody like me to exist kind of like in outer space without being inside any organization.
6:15
I use Zoom for talking with clients, I use FreshBooks for billing clients.
6:19
I have like, you know, if I were to drive Uber, I would use an API and an app to like make money without being an employee.
6:28
So, a set of interacting protocols organizes the labor economy, and between like 1980 to today, grew from like barely a few few percentage points like to double-digit percentage of the labor economy is organized in this nebulous way, and that's the power of protocolization.
6:45
And to connect to something Tim said, this is why we, over the summer, became convinced that protocols are like a first-class citizen for thinking about the world.
6:53
You should not be analogizing them to nations or corporations or other like constructs that we're used to thinking about with.
7:00
You should think in terms of like protocols are this first-class citizen that allows human civilization to be organized in seemingly nebulous ways, but with very hard edges that allow very interesting things to happen.
7:12
And the gig economy is a good example that's like, you know, not a blockchain, but it's a set of protocols.
7:19
Yeah, and that's what I got excited about when I was reading all the material.
7:23
The, you know, what would be the best analogy?
7:25
Would it be to the old idea of narratives?
7:28
Cuz you covered that, and you covered the five different narratives that protocols can can help with.
7:34
We're going to get to that later.
7:37
Are are we looking at deeper deeper structure here in terms of are like are these earth patterns are what what's the best analogy an axiom?
7:50
So I think this is where the metaphors to existing concepts kind of like draws a straight.
7:58
So there are parts of protocols that seem like narratives and I personally have been very interested in that aspect and I've been writing some essays on like protocol narratives.
8:08
And then sort of more technical example of that this is going back to the Ethereum world there's this distributed blockchain game called Dark Forest which is inspired by the Dark Forest trilogy science fiction novels but it's set up on the blockchain as a little set of worlds that you know are kind of like planets and they attack each other and so forth and the whole
8:30
thing is it's not a video game it's not a novel it's not even world building in the sense of the Marvel Cinematic Universe it's just set of like technological elements that talk to each other through like messaging rules and allow like a grand narrative to unfold and we were talking before you started recording about we both just came from
8:52
protocol town hall that we hosted by um some bleeding edge artist writer Simon We're he um experiments with like experimental art music storytelling and stuff and he was telling us about um this weird a sort of like augmented reality game type thing called SCP which I think it's called secure uh secure and protect or something but it's this
9:20
bunch of people who go around and look at like normal stuff like you might be in a bus or looking at a vending machine and they imagine that you know weird aliens are inhabiting those machines and they're part of a secret secret organization that is there to protect the world from these aliens and prevent them from bursting out of containing these aliens, right? So, this
9:38
So, this is like it's like Pokémon Go, it's like virtual reality, but it doesn't require those elements.
9:44
All it requires is for a bunch of people to come to like a consensus on here are the rough rules we are playing by how to tell stories.
9:52
We are not going to use like proprietary materials of photographs, but we are going to like just projectively uh project a story imagined by us, improvised by us onto just ordinary reality around us and share it as a wiki website.
10:05
And this is wonderful little metaverse that has come up.
10:07
Novels and video games and um video little movies have come out of it.
10:12
So, that's like it's even more fundamental than world building, I would say.
10:16
So, they've set up like a set of rules on which a world can grow, and on top of that world narratives can unfold.
10:21
So, it's like couple of layers of below the stack that we are used to thinking of.
10:28
Yeah, and the uh it made me kind of think uh I've been reading recently about uh Conway's Game of Life.
10:37
And uh I thought uh that two might connect.
10:41
Uh yours is a little more sophisticated.
10:44
Uh but yours is also seemingly another thing that I very much subscribe to.
10:49
Complex adaptive systems, emergence comes from below, not from above.
10:53
And that is another thing in which the game you've just described uh seems to uh bear out.
10:59
Uh I started asking the question uh Tim before.
11:02
I think you were going to jump in, so I'll switch it over to you.
11:08
Yeah, so I guess this all this all show like the the different background uh that I come from from from Red Cap, you know?
11:13
Like I I always think of these things from like a very just uh dry perspective.
11:20
Like, you know, what's like the simplest like instantiation of of like a protocol.
11:23
protocol. And I think part of the reasons why it's so hard to define it is there's like three um three like preconditions or or like combinations that like I think when you get together you get the a proper protocol but that you end up focusing on one at a time and so the first is obviously this concept of rules right like and you need some set of constraints and but then when you when
11:45
you think only through that lens you know you can think of something like an API right like what's the difference between a protocol and an API or even a language grammar whether it's like a programming language or a or you know written English like language like English there's this concept of rules that's super strong but it's not it's not necessarily sufficient. I think the other thing that
12:01
I think the other thing that came up with protocols is the idea of ordering really matters right?
12:05
And this is where people tend to analogize them to more like social structures like rituals or you know traditions where like something has to happen to trigger something else in an order and if you if you lose that the protocols tend to not work and you know to take an example it's like if you wash your hands before they're dirty you know the health washing or hand washing health protocol doesn't quite work right?
12:28
doesn't quite work right? So so so this concept of like the ordering in which a step happens more than just the steps themselves is really important and then the last the last bit which is what Venkat was getting at with world building as well is this idea that the protocol is like a substrate on which
12:45
more complex things can be built and this is why I think you you end up flip-flopping your thinking that protocols as like simple APIs and then you know something like tech platforms or nation-states and you can kind of alternate between those two views so easily even though they're really really far from each other is the idea that
13:04
like once you have these rules in the specific order that you assign to them you can build stuff on top of that and that quickly becomes really powerful and and I'm not sure this is like exhaustive but it feels like at least a minimal set of things you need to to have a proper protocol and and this is what makes makes like the
13:22
the subject matter so hard is you know each of those is is like a huge rabbit hole where um, you know, people have spent tons of time trying to to develop like that these sub parts um, and trying to like hold them all at like a same level of of of of attention and and look at like the higher level protocol construct. It's it's it's actually
13:42
It's it's it's actually really hard and that's uh yeah, what we tried to do uh with with all these researchers over the summer and um, yeah, so I don't know, hopefully there's like another angle that would be helpful to think about like decomposing protocol into sub parts and and how they fit together.
13:57
I think that the the rules as constraints uh in the beginning is is a good jumping off point.
14:03
I I've been reading The God Problem by Howard Bloom.
14:06
I don't know if you guys have read that, but uh he he he leads in with some really interesting ideas.
14:12
The first being about how uh amazing things can emerge from very simple rules and he gives as his example termites building their beautiful cathedrals.
14:22
I don't know whether you guys have ever seen termite mounds in the wild.
14:26
They are truly stunning and amazing.
14:29
And essentially the simple rule was termite goes along, they are very fastidious, they don't like all over the floor.
14:37
They go along, they see a little pellet of termite scat and they take it, but they go and look at all of the other places they've been put and then they go on top of the tallest one and they put it there.
14:49
And and out of that emerged these beautiful cathedrals that termites have built.
14:54
But then he also connects that to and I want to ask you guys because I'm not sure about this.
15:00
He connects it to the ancient Babylonian priests basically invented writing because of accounting, right?
15:08
And you know, they essentially they were in charge of making sure that all the worshipers made the right offering to the gods of barley corn.
15:17
But counting barley corn like if there's 10,000 of them, counting them one by one it's going to be a very laborious and tedious task.
15:24
So, they got their little clay tablet and they did a Y for a one and they did a less than sign for a 10.
15:33
Is is there a connection?
15:36
Yeah, that I mean you're touching on like several points that I think we've been like obsessed with over the summer.
15:42
So, the first one with termite one I'm not familiar with the bug problem.
15:44
We actually had a researcher Rafa.
15:48
He did a whole project on swarms and what you can learn from swarms in relation to protocols.
15:56
And the interesting thing is swarms have been studied for a long time, flocks of birds and things like that in the wild, but swarms in protocolized environments like, you know, mob marching through the streets protesting something.
16:08
That's a different kind of swarm.
16:10
So, there's an interesting sort of relationship between there are the constraints and rules that define a protocol like they mentioned and then there's the sort of free design space it opens up where you can like have unscripted and even chaotic behaviors.
16:24
And how swarms behave in a constrained set of rules is kind of like a very interesting chaos and order dynamic that kind of creates complexity while emergence.
16:35
So, that's one immediate response point I wanted to make.
16:38
The other is throughout this program and even before the 6 months where we were doing it within the Ethereum Foundation, we were very inspired by an A. N.
16:46
Whitehead quote which is goes something like civilization progresses by increasing the number of actions we can do without thinking about them.
16:56
And this is like a very profound thought that I think has been inspiring all of our researchers.
17:03
It became one of our most cited like quotes.
17:05
We would toss at each other and you can see why that is, right?
17:07
Like termites literally can't think sophisticated thoughts.
17:11
The most complex thought they can think is take a little bit of dirt, go to the highest point and deposit it there, and then you get this fractal structure or whatever that emerges, which means at a at a very hard constraint, any complexity that emerges, 90% of it has to be unconscious.
17:27
And to your example of things like Babylonian priests, we humans tend to have like, I don't know, conceits about like how like uh lofty our philosophical conceptions of our own existence is, but when it comes down to like how we operate in reality, it's literally 90% automatic actions we don't think about.
17:47
Like Tim said some used the example of like, you know, washing your hands or like handshake.
17:54
These are actions we don't normally think about, but still 99% of the time we do right.
17:58
And unless something like the pandemic happens making us rethink, all right, what the hell is a handshake really?
18:03
Should we rethink what it is?
18:05
90% of our actions are automatic, and 90% of the emergent structures of civilization come out through this unconscious process where we might as well be termites, even though we build these lofty religious philosophies on it.
18:19
Civilization is like closer to a termite mound than we might want to admit, right? Yeah.
18:27
Yeah, maybe one quick thing I'll add on that is um this segues pretty well into another theme that was recurrent for us, which was like the idea of like dangerous protocols and you know, that them being entrenched and lasting for a long time.
18:40
Um so, you can see that the termite example is is actually a really good one because it shows you, you know, something with like a very limited reasoning ability being able to build a structure that's like much more at much bigger than it is, with much more complexity.
18:55
Um and it also shows you how hard it would need to change that.
19:00
Like, you know, say that for some reason it was bad that termites built these structures, how would you even get them to coordinate on doing something else given like the level of uh agency they're operating in at that's like, you know, bring this thing to the highest point.
19:14
And this is something that came up over and over in the research.
19:17
Um, and the research on Swarm was an example of that.
19:19
There was somebody literally just researching dangerous protocols.
19:22
Um, but the idea that once you have these this kind of accumulation of of of small simple behavior that's bounded by some constraints, um, it can lead to really, uh, strong outcomes that are either positive or negative.
19:37
Um, but it's it's very hard to to build the coordination around changing those outcomes once the protocols is set in place.
19:44
And obviously like, yeah, the the hand washing and hand shaking example is another great one because, you know, it takes something on the scale of like a global pandemic to have people rethink something as simple as like hand shakes.
19:56
Um, and so this is an another huge sort of area of research like these protocols tend to be very sticky.
20:03
And the Weitzman quote as well actually has you can think about it two different ways, right?
20:07
One is like it's great we get the automate more and more of like our complexity stack as a society as we become more protocolized, but the the other implication is that as these things have like disappeared and society just does them automatically, um, it's very hard to debug them or to change them because they're sort of, uh, taken for granted and, you know, so much other stuff is built on top.
20:29
Um, so yeah, I think there's like a big area of research around like when you design protocols if they're successful you might not be able to change them.
20:38
So you want to be extremely mindful of designing them in the wrong in the right way, um, and and thinking about, you know, the second order consequences of of doing so. Ed.
20:51
Yeah, I I I a couple a couple of questions based on that, Ranil.
20:53
I'll ask them both and you can answer which one you find more interesting.
20:57
So this is is is so the the first one was would a you know, as the world becomes increasingly protocolized does that mean that the world is about to be becoming less agentic?
21:09
Is a protocolized one a one which inherently has less agency?
21:15
Um and the second one tied into the point Tim was just making is are there any behaviors or areas or sectors or however you want to phrase it that shouldn't be protocolized?
21:25
That that that that that do not lend themselves to a protocol-first way of thinking or a protocol-first way of approaching approaching the problem.
21:40
You want to pick one of those to field, Tim?
21:43
So, maybe on the first one, um I feel like I have conflicting thoughts, um but it feels to me like the barrier to entry to create protocols is maybe lowering because you get distribution through the internet, you get better tools like, you know, blockchains and AI.
21:59
Um So, even if I'm not sure if I agree that a more protocolized world leads to people being less agentic, but I think we are moving to a world where more people can start protocols and protocols have like the ability to shape a wide like a huge number of people's behaviors.
22:18
So, maybe that's how I would think about it.
22:20
Um and this comes up a ton, you know, like people talk about this is like AI like, you know, is ChatGPT agentic or or what not.
22:26
And I think um the way I think about something like that is a bit similar to like a blockchain where like under Ethereum obviously Ethereum's not like self-aware or anything, but you can trigger like a huge chain reaction with a relatively small input, you know, you can start a transaction and then have it execute any type of code anywhere in the world, replicated across, you know, tens of thousands of nodes.
22:50
And AI is a bit similar where like obviously the AI today is not like, you know, coming out of the chatbox and doing stuff.
22:55
There's like a huge amount of like energy conservation in a way where like with a small prompt you can get, you know, chat GPT to do some huge uh some huge output.
23:06
And I think it's like as you get these more and more powerful tools, people can build protocols with them.
23:12
And yeah, I'm not I'm not sure if the outcome of that is like a less agentic world or not.
23:16
Um it definitely feels like we'll we'll see more powerful protocols emerging uh soon. Yeah.
23:25
I think I have a more um I don't know uh a strongly opinionated take on agency because I don't think it's like the word agentic is interesting because I think it was popularized by recent AI discourses, but this dynamic you're pointing out um goes back like to the earliest technology, right?
23:44
Like there's always this dynamic of it increases the overall agency pie, but it kind of might cut back agency for particular groups of people, right?
23:53
Like you have um like some machinery that automates something.
23:58
Whoever was doing the artisan work, these other rights or whatever, their agency goes down, but other people who are designing machinery goes up.
24:02
So it's the whole standard argument in economics of like, you know, there's a lump of labor fallacy.
24:08
So the lump of labor fallacy I think is actually equivalent to like a lump of agency fallacy, right?
24:15
So the lump of labor fallacy for those of you who are not familiar is the idea that there's a fixed amount of work to be done, and in a growing economy that's obviously not true.
24:22
Like, you know, textile um artisans are you know, farmers using their muscle power, they might go out of um business or like, you know, lose work, but overall the amount of work to be done simply increases, and people come up with more and more uh uh things they want to do.
24:39
And I think this is where it's like I go back to Neal Stephenson's idea of like Eloi and Morlocks in his essay In the Beginning Was the Command Line, which riffs off like a H. G. Wells story.
24:53
But the idea that when technology progresses, it creates this um breed of superhumans who are a minority and have like vastly increased superhuman levels of agency and on that particular technological dimension, everybody else may become like much less agentic, right? So, that does happen.
25:09
Like, you know, let take auto automobiles for example.
25:11
In the beginning to drive a car, you had to be a little bit of an auto mechanic and you had to know something about like repairing engines, replacing spark plugs, things like that.
25:19
And as cars became more and more sophisticated, a small minority of people are now superhuman car fixer type people who know huge amount about like very sophisticated machines.
25:30
The rest of us are like like car won't turn on.
25:32
Do I call AAA or whatever, right?
25:34
So, it's this split that happens.
25:37
So, my belief is the overall agency pie is always expanding, but because in the short term it gets rearranged in a way, people on the wrong so caught on the wrong side, which is usually a large numbers of very vocal and concentrated people in special interest groups, they tend to have like very vocal reactions as you know, zero-sum or negative-sum way.
26:02
Like and since you brought up AI, I do want to mention that AI is a very interesting going through an interesting inflection point now.
26:07
But, you have these big centralized models and just earlier this week, OpenAI announced its uh a big new product line that anybody can code up their own GPT assistant, right?
26:18
And this to my machine is like shows the importance of protocols because we've moved to the protocolization stage of AI itself because now we're going to talk about what private data are you uploading to OpenAI?
26:28
What are the security and privacy guarantees around your little GPT trained on your data?
26:32
How's that going to interact if you put it on the marketplace and like it interacts with other GPTs?
26:37
So, again, suddenly a lot of people like artists might be losing their job because AI can now produce artwork.
26:43
Like, I we should mention in our publishing of the this research material, we are using AI-generated cover images, which are turning out beautifully, by the way.
26:51
So, yes, there are people who are going to be losing some agency, but other people are gaining the power to like just program up their own little assistant bots, right?
27:01
So, it's This is like a 200-year story, and it's going to like keep repeating, but agency I think keeps increasing.
27:08
On the 200-year story, you actually go into that at some level of detail, looking at pre-industrial protocols, uh all the way up through the industrial era, and now to our current era in which uh you you find memes are inserting themselves into the process as well.
27:30
Now, I I know you don't mean memes in the Dawkins sense of memes, you mean memes more in the internet sense of memes, but talk a bit about about that, about the the as you say, we're looking at a long sweep of history here, and protocols and and world-building and stories and narratives have always been around uh to a greater or lesser extent.
27:53
I I think now with the world completely connected, they move a lot faster than they used to.
27:58
Uh but either of you, please, if you would jump in on on that idea.
28:08
I think Tim's favorite research project is appropriate here. Tim, Yeah.
28:13
Yeah, so it This is a bit more narrow than like all the protocols over 200 years, but I think that the one that we have that uh looks in a very narrow domain and and and evaluates it over a long period of five is is Tim's project, which is around workplace safety.
28:26
Um and to me, this is this project validated that like some of protocols was worth it because as I said, I I work on blockchain, never, you know, been on a commercial construction site in my life.
28:39
Uh and as I was reading uh the essay, I was like, "Wow, he's describing exactly what I work on."
28:44
Even though he's describing coal mining in like the 1870s or something.
28:48
Um And and and the overall idea is that, you know, um workplace safety protocols were invented early on because workplace uh safety was just a huge problem.
28:59
You know, people would go into these mines and they would explode and they would all die.
29:04
Um so then they came up with some ways of figuring out, you know, is there gas in the mine?
29:07
Um and you know, protocols to doing this uh through trial and error while minimizing casualties.
29:13
You know, I can imagine the most naive approach is one person goes there, lights a match, and explodes.
29:16
Uh you still lost that person.
29:19
So over time, you know, people came came up with like they they were able to identify different types of gas and whatnot and be a bit smarter about it.
29:26
And then, you know, you you you fast forward a bit.
29:28
Um the mines aren't blowing up anymore, but you have things like long-term health consequences, right?
29:33
Like, okay, you're you're not dying, you know, today, but like, oh, you realize 10 years in you have all these issues.
29:38
Um so there's more and more and more workplace safety uh standards that happen, how do you protect workers and whatnot.
29:45
Um And then you reach a point where you've pretty much eliminated most of the risks.
29:49
You know, it's it's pretty safe to to to work in in these industrial environments.
29:53
Um where you start to get these weird accidents, you know, it's not that the mine explodes, but it's, you know, some person is backing up the truck at night and there's no light and, you know, somebody walks in front of it and, you know, gets gets uh run over or something like that.
30:07
Um And so you so the the you go from a world where you're like observing with pretty high frequency these pretty uh catastrophic events to one where safety, you know, you most of the time things are going well, you're not observing anything, and you get this weird outlier event.
30:24
And people will then continue to add these workplace safety protocols because not only uh you you you've sort of created this meta protocol at this point where it's like, okay, when bad things happen, we should like, you know, do a retrospective and and set up some new measures and deal with that.
30:41
Um but the benefits you get at this point are much lower because you're not stopping the mine from blowing up.
30:46
You're, you know, trying to overfit to this one weird edge case and adding all these measures.
30:50
Um, and potentially, you know, even if you're not just slowing down productivity, maybe you're creating new edge cases by adding all of this complexity.
30:58
Um, and so this this one for me was really interesting because it's it shows how it's really hard to get to the sweet spot of like, what's the right level of protocol complexity?
31:08
Like, where you know, where should workplace safety standards end up at?
31:12
Um, and both from like a just objective point of view, you know, what are we willing to tolerate, but also from a social point of view where if you've set this culture and this, you know, sort of social protocol that when something happens, we respond to it, then, you know, you get these weirder and weirder edge cases that happen because you sort of blocked out every other potential failure mode.
31:32
Um, but you're still you're sort of spot where like you feel like you have to add something.
31:37
Um, and I won't, you know, bore you with like the the the corollary in like smart contract and blockchain development, but you basically see the exact same thing, right?
31:46
Where, you know, there's maybe speed running like 7 years instead of 200.
31:50
Um, but we're trying to figure out the right level of like safety and and and and, you know, security concerns.
31:57
Um, and trying even just coming up with like good heuristics for, okay, is this actually a failure mode or is this just, you know, a random edge case that we should just accept because, you know, it happens once every uh, so often.
32:10
Um, it's a really hard problem.
32:13
Um, so yeah, I think that was that was something that was kind of interesting to see evolve over one such a long period of time, and two, where yeah, a domain that had nothing to do with tech or the internet that and where, you know, 95% of the actual content comes from before the internet, um, was directly applicable to what I do.
32:35
And so the hope is like if we can find more and more of those connections and start building, you know, theories around them, and the theory works as well for workplace safety as like Ethereum development, um then we know it's it's robust, right?
32:49
The the uh thing that jumps out at me there, though, is you were talking about uh safety protocols.
32:54
I just kind of envisioned the the the basic ones, right?
32:57
We put the canary down, the canary doesn't die, it means it's everything's okay down there. Very, very simple stuff.
33:03
But then I kind of see Ptolemaic circles being drawn, you know, we're all familiar with Ptolemy's uh astronomy that kept They needed to keep adding all of the various circles to to make uh to to take advantage of new observations.
33:20
Is Do you think that's one of the reasons why these things can either become unwieldy or perhaps inadvertently cause us to over-index on safety over other things?
33:36
Yeah, and I think it And the reason why is it is a hard problem because obviously, you know, the canary say the canary is one circle and like, I don't know what the full workplace safety protocols are today, but that's, you know, a million.
33:46
Um where do you put something like the risk of developing lung cancer after 10 years of working in a mine, right?
33:52
It's clearly something where you know, you can argue that it's, you know, from the perspective of of the earliest protocols, like it's this huge addition in complexity, um but it's probably, you know, a a pretty a pretty frequent failure mode, um and and and something like, you know, that that type of work.
34:09
So, I Yeah, there is this tendency to add more and more, and it's it's very hard, not only is it hard when you're close to it to figure out if like the marginal circle is is beneficial or not, but argue even if you have like the benefit of hindsight and you're detached, like it figuring out exactly which circles to remove is this is super hard. Um And again, oh, sorry.
34:33
Yeah, I'll pause here if you want to cut me off if you want to say something.
34:36
Yeah, I just wanted to inject an interesting example that I think makes a particular point, which is um the evolution of safety in nuclear submarines.
34:44
So, one of the most compelling AB test examples of the importance of evolving complexity in safety in safety protocols is that the Russian Navy nuclear navy had I think something like 17 major accidents through the Cold War era. The US Navy had none.
34:59
And this was because um entirely due to Admiral Hyman G.
35:04
Rickover, who's like this famous guy who built the American nuclear submarine fleet as well as America's nuclear program.
35:11
And he basically one of the things was like he focused on like engineering excellence, operational excellence, safety, things like that.
35:20
And you got you can see why this difference was so stark.
35:22
It's because nuclear submarine technology is a technology with a very high ceiling of complexity.
35:30
You might have started with like, you know, Fermi doing like an atomic pile experiment and starting the first reaction and then the Manhattan Project, but you can see that fundamentally you're dealing with nature at its most complex and subtle.
35:41
And any technology you build on it, especially if you make it more powerful, more capable, you know, smaller, more compact reactors that can run for longer, it's going to get more and more complex and your safety protocols have to keep up.
35:56
Now, this is an I think a research area and maybe different industries have ceilings of complexity.
36:01
So, I suspect part of what's going on here is there's of course, like, you know, sociological effects that when a field is very young and immature more sort of like people with high risk appetite go into that.
36:13
Like, you know, crypto in its early decade attracted risk-taking types.
36:16
Gold rushes attract risk-taking types.
36:19
As technologies mature, it attracts like more conservative people, but also those more conservative people build the technology in like you know more patient detailed complex ways.
36:28
Like it takes a different personality to come up with one brilliant idea for a blockchain and 10 years later to build on that like some of the things um Tim might be able to mention like some of the things on the Ethereum roadmap currently, they don't look like the wild west anymore.
36:46
Like 2009 to 2014 when Bitcoin and early Ethereum were building up, what's going on today with higher levels of sophisticated Ethereum, they are really much more complex.
36:56
They look much more like nuclear submarines than primitive coal mines to me.
36:59
So, there's like uh sociological selection effects of who selects into working on a technology, what the actual leading edge problems are, how complex it's getting, and therefore what kinds of weird accidents can happen and what you need to do about them.
37:14
And to add to Tim's point about like outliers and weirdness becoming more common, the consequences also become more complex. Right?
37:20
If you have a small early nuclear reactor, it melting down is not that big a deal.
37:24
If you have like a bleeding edge nuclear reactor that's powering an entire city, that melting down is a big deal.
37:30
If you have an early airliner with like 10 people flying in it, the airliner crashing is 10 people lost.
37:38
In 2013 and 2014, you had those two Malaysia Air like like you know with very weird accidents.
37:42
One getting shot down over Ukraine, the other getting like hijacked in a very weird way and getting lost over the Indian Ocean.
37:47
That was 300 plus people dying because of a very weird outlier thing despite like 50 layers of airline safety protocols to prevent that happening, right?
37:58
Consequences increase, the weirdness of outliers increase, the complexity increases, and the staffing or people involved in the technology churns through from like highly sort of like risk accepting to highly risk averse, and they're working on different problems.
38:12
So, there's like a whole life cycle here and I think uh Tim Harford's coal mine paper, one of its interesting contributions is it comes up with a speculative theory of the evolution of um protocols across time in for precisely these reasons.
38:26
So yeah, I wanted to throw the nuclear submarine example in this um pool of examples to think about. Yeah, sorry.
38:33
I didn't mean to like uh hijack Tim driver.
38:36
No, no, no, no, that was actually uh very instructive uh and it leads to the the the question uh you know, at what point do we uh like what's the best way to allow protocols to build and if we borrow from uh development open source close source What do you guys think?
39:00
I yeah, I think this is really hard.
39:00
I mean, this is why I started this whole thing and reached out to Vac Capital in the first place because um I I think that yeah, that the challenge almost is like yeah, if these things are successful, then you have very little opportunity to change it.
39:16
Um and and the you here as well as is an interesting one because um if it's successful, these protocols have now these very long lifespans.
39:24
So if you know, you probably won't be around when it has to be changed and you have to think about, you know, the like second order implications.
39:31
Um So frankly, I I don't know.
39:33
I think one useful framework we came up with though during the the summer is um defining bad protocols is much easier than defining good ones.
39:43
Um and that's maybe one tool that you could use if you're thinking about that where uh the canonical example we came up with was like the TSA, right?
39:50
If I say just like the words TSA, you know, you you just smiled there.
39:54
You like immediately get all these like visceral reactions of white poorly designed at so many levels and you know, she spent 5 minutes thinking about all of the design constraints of the TSA, um you can you can get a good picture of like, you know, why is this like a bad protocol.
40:11
Um and this is this is like much more illuminating than say if I tell you HTTP.
40:15
You don't have like this emotional reaction.
40:17
You know, you sort of assume it's just working in the background even though it's one of the most used protocols.
40:22
Or saying, you know, if I tell you like washing your hands, you know, um the the the gap that that it takes someone to go from like looking at hand washing to like, okay, how many lives have we saved through disease?
40:32
It's like a much less uh visceral sort of thought process than, yeah, why is the TSA bad or, you know, why is it another broken protocol not working?
40:39
Um So, I think that's one of the most like optimistic uh tools we've come up with that that try and think through that where if you're able to actually think of your protocol, you know, what would be the worst version of it and what would lend it to the the go there?
40:56
Um that would be at least a good first step in trying to minimize those negative consequences.
41:01
Um and especially in terms of like second-order impacts.
41:04
Uh so, another another uh example that came up during your uh summer was that the idea of speed bumps in front of schools.
41:10
Um and this was contested research but shows you how hard it is.
41:14
Um so, obviously, you know, why would you want speed bumps in front of schools?
41:17
Because you don't want people to drive quick and and, you know, hit kids with their cars.
41:20
And then there's some research around uh well, actually, the speed bumps slow down uh ambulances in these areas to an extent where like people died to ambulance more often than people than kids were were being uh hit in an accident.
41:35
Um And this shows you like a extremely simple protocol like, you know, that anyone can understand.
41:40
People are going too quickly on this road.
41:42
You just add something and you solve like the visible cause.
41:46
Um but the second-order effects are are really hard.
41:49
And then they're also much harder to build uh an intuition for.
41:53
So, you know, the the the ratio of like how many school kid accidents you need to prevent to make it worth it to, you know, probabilistically delay ambulances by this much in in this town.
42:07
It becomes a much harder thing to undo than it was to do in the first place because you can get people to rally around, you know, protecting the the kids, but then, you know, taking you off the the the speed bumps is just becomes really, really hard.
42:18
So, I Yeah, so I guess those would be my my one takeaway so far is just trying to think through the negatives and the second-order impacts.
42:25
Um and like the dystopian version of your protocol succeeding can hopefully help you build a better version that, you know, take some of those uh issues into account. Yeah.
42:37
I want to jump I want to just really quick jump on that with a question and an observation.
42:42
Like, so uh the reality is most people, unfortunately, are first-order thinkers.
42:50
They they very rarely take secondary and tertiary effects into uh the formula, you know, for thinking about something, right?
42:58
And then you also mentioned another aspect of this, which is seems very important to me, and that is the visceral reaction of other humans, right?
43:09
Like, it's very, very easy to get a bunch of parents irate over the fact that uh it might be a good idea to take the speed bumps out because the ambulances, I want my kid to be safe.
43:24
I mean, I just respond if both of you wouldn't mind.
43:29
Well, I I can take that one.
43:30
The the previous question you sort of posed is closed better than open, that kind of thing.
43:37
And one of the things that I think we all discovered was developing what we call protocol literacy requires sort of like backing away from like the obvious of like, I don't know, sacred cows that you might otherwise rally around.
43:51
So, the challenge to becoming a second-order thinker is, when faced with such a thing, typically people will reach for their most favorite sacred thing and say something like, if they go second order at all, it's like, what we need is a strong centralized mayor who will like uh look at this holistically and solve the speed bumps versus ambulances a problem, right?
44:09
So, we use So, we reach for that.
44:12
And invariably, I think what happens is when you develop a fetish for, you know, centralized versus decentralized, open versus closed, you're not entering sort of the dialectical relationship with the problem domain.
44:27
So, that's like a complicated phrase, but you know, problems, complex problems evolve over time.
44:33
They you never truly solve them.
44:34
You have to kind of like stay in a loop of constantly paying attention to them and like uh solving them.
44:38
And you cannot afford to get locked onto like any sacred cow like in a centralized or open or whatever.
44:45
Um so, uh oh, and uh the kind of like exercise Tim described of like think of the worst protocol you can uh we actually um did that at the researcher retreat over the summer where I think people were calling them Kafka-esque protocols.
45:00
So, think of the most Kafka-esque one.
45:02
And this is an element of what as protocol literacy, which is stop your first-order reaction of centralized bad, decentralized good, and just stop that train of thought, and instead pick up one of these disciplines like asking yourself the sort of like provocative question, what's the most Kafka-esque version of this that can happen?
45:24
And then play that thought experiment through, and maybe you'll discover that the most Kafka-esque thing you can think of is actually a completely open-source and decentralized version of the thing, right?
45:34
It's like the examples we've been talking about.
45:37
TSA is not super centralized.
45:37
Yes, there's some centralized directives, but it does involve like all local airports.
45:43
The other example we talked about speed bumps and ambulances, that's a bunch of interacting like local government things with like highway state laws and stuff.
45:52
So, it's like all your obvious first-order sacred cows get challenged if you actually force yourself to uh literally think in a literate way about the particular problem.
46:01
So, I would say that's probably the big takeaway from the program, like how do you get the protocol literacy number step one?
46:11
Throw away your sacred cows, and step two, build a few disciplines that allow you to like, I don't know, enter into an open-end open-ended conversation with the problem.
46:19
It's like, all right, that didn't work.
46:21
Let's come at it another way.
46:23
This is the problem with externalities with speed bumps.
46:25
Maybe we can come at it in a different way and give ambulances a different lane to go through, right?
46:30
So, if you keep the problem in an open-ended sort of questioning mode, that itself sort of bootstraps you into the second order way of thinking, and I think that's a big part of like being good at protocols.
46:42
And I think maybe two things I'll I'll add to that.
46:44
Uh, one is when we when we were doing this exercise, another thing we tried to think about was, are there protocols that like have never changed, right?
46:52
Like, once you get them, they're done.
46:54
And one thing we we came up with was nuclear codes for the president, right?
46:59
And you're like, okay, clearly something that's not a rocket science.
47:01
Well, it's it's like very high stakes, but you know, there seems to be a protocol today, and it seems to have been stable.
47:07
And literally during our workshop, there was an article on Bloomberg around how they changed those protocols over the summer.
47:13
So, that that that was like a one example, and it didn't work out.
47:18
Um, and I think the the other thing that I found really helpful is, um, actually like engaging with the protocols, and I I need a better word than engaging, but there's some aspect where when you start like thinking through thought experiments with it, or trying to break something, or almost thinking of like the protocols as like physical Legos, and, you know, trying to rearrange parts of it.
47:43
That seems to be a much more fruitful approach than say, like the approach like the brain space you get when you're trying to like solve a math problem, where you're I don't know, trying to go like linearly through some step of of of sequences, or, you know, writing code, or something around
47:57
like thinking you know, like these protocols as like objects in in space that you can rearrange um that seems that you have to be more fruitful than trying to think of it as more theory or or or a set of rules that you're writing um in in like a a a I don't know, one-dimensional or or linear way. Um and this is I think
48:16
Um and this is I think something that we've we're going to try we we've managed I think to convey this to people who are like full-time in the program and you know, we hate to just think about that.
48:25
But I think this is one of the big challenges.
48:27
How do you scale this, you know, 10, 100, 1,000 X?
48:30
Like how do you get somebody say they want to spend 2 hours thinking about protocols to have them think about it through that lens rather than have them trying to say read the equivalent of a Wikipedia article about HTTP or or workplace safety and and and not quite get the feeling that like these are these are malleable constructs that, you know, can have really different outcomes it depending on the way you arrange them.
48:54
Well, I was just going to ask that.
48:54
I I I as protocols are so malleable and so interop- interoperable, so you know, you're saying how a protocol a building protocol would form how you think about Ethereum and that sort of thing.
49:06
And also with their such sort of fundamental building blocks of of of of of of of of of of of of of of of of of of our behaviors, why do you think more people aren't talking about protocols?
49:17
I mean, I I I I I I outside of the Ethereum Ethereum community, outside of what are you what you guys are doing, I I wasn't particularly familiar with them until I read your stuff.
49:26
It it doesn't bother that why I think yeah, so I I guess the way I would frame it there's probably more people than you'd expect think about protocols, but they think about it in like a tangential way.
49:36
So like say again, you know, you were working on like a construction site, like you think about workplace safety protocols, but like you think of your your domain as like construction and like the workplace safety protocol is one subset.
49:49
Um and I think it's just because it's something that's embedded in every single domain, there just hasn't been I don't know, maybe blockchains are just like the first time where we have enough people just focusing on that and there's so little physical domain around it that we get to spend more of our time thinking about protocols.
50:07
Maybe web protocols are the exception, but then yeah, it it feels like in every other domain except like maybe web and and blockchain stuff, protocols are just like a second or third order concern and people do end up interacting with it and but it's just not their main their main focus and and they think about something else more. Yeah.
50:29
And I like that this is not unique to protocols as a sort of targeted field of study.
50:34
Like if you think about economics for example, economics wasn't a thing until Adam Smith.
50:41
Like if you read like medieval history books, like I was reading during the pandemic, I was reading about the Black Death and the Plague and like the wow, there's like lots of chapters on economic organizations in the years leading up to the plague and how artisan guilds organized the economy and it was quite quite clear that they didn't think of themselves as doing economics.
51:00
They were literally part of like social and cultural sort of like ritual space.
51:07
Like they had lots of like laws about what you couldn't and couldn't could and couldn't do and they were all driven by concerns about maintaining the social hierarchy, making sure apprentices apprentices didn't get above their station.
51:19
You could tell that their entire frame for thinking about economic activity, which is what the artisans of the 13th century were involved in, it was not economic at all.
51:27
It was like cultural programming, societal stability maintenance.
51:30
They didn't think of economics, though they used like token currencies to transact and stuff.
51:34
And then you have Adam Smith, then you have like a few generations of economics.
51:38
Now you have entire like um schools of thought of economics and there's kind of like an economics way of thinking about the world that didn't exist in the 13th century.
51:49
So like to make an analogy to what we are trying to do here, there was something it was like to be economics literate in like, you know, 1900 that kind of didn't really exist before Adam Smith.
52:03
Like before Adam Smith, you might be like socially literate, politically or socially literate or like politically sophisticated or like be good at maintaining the social order in other ways, but there was no such thing as being like economics literate.
52:15
And then Adam Smith gave people like generalized frameworks for doing it.
52:19
And now we all intuitively approach certain problems with an economics mindset.
52:24
It's like we ask, "What are the incentives?
52:26
What are the uh price and supply demand dynamics?"
52:30
You know, we have like a set of like literacy tools to think about problems in an economics way.
52:34
So even though economics as sort of like a domain of stuff has existed forever, like millennia, economics as a way of thinking about that regime of activities like only 200 years old.
52:46
So protocols too, like handshaking and all the, you know, ceremonial ritual protocols, they've existed since the Bronze Age, but like economics, we're going through a threshold of yeah, this stuff has gotten sophisticated enough and we now have like the beginnings of like a universal way of thinking about it that we can actually explicitly talk about it with like, you know, protocol uh science as it just like with economics.
53:07
And so I think we're going through such an inflection point right now.
53:12
And and you uh make that point rather clearly that you you make the case that uh training people to be protocols-first type thinkers is going to hopefully replace uh you know, uh the nation-state thinker or the uh the the the the various ways in which, as you point out, right?
53:33
We didn't have the field of economics.
53:34
And it was just like we didn't have science until Aristotle called it science and basically uh invented that brand new way of thinking.
53:45
How how how are you going to go about other than the obvious what you did this summer which I think is wonderful.
53:50
But how do you how do you go about making the taking the next level up to to help people become more of protocols first type thinkers as opposed to to the other methods that were far more familiar with and comfortable with.
54:08
Yeah, maybe um I I think the thing that economics and science example, one thing that's really worth highlighting there is economics and science existed in people's lives before and they didn't have like a label for it, right?
54:21
And this is the thing that you know, Adam Smith gives you the first set of like really good labels and then if you're just you know, some random shopkeeper and you you're introduced to these concepts, you can then sort of bucket these intuitions you had about how you're running your business in in these labels and potentially build on them and build from there.
54:41
So I think this is a part that's like really valuable where it's it's not that like you're inventing new you're inventing new labels more than you're inventing new behaviors or new realities.
54:51
So the hope is that if the labels are actually good, people recognize recognize that then and then they can they can like apply them.
54:58
And this is the the motivation for for having people from different domains was this again.
55:04
It's you know, if the label is good enough for blockchain and workplace safety and you know, architecture then it's probably generalizable enough.
55:12
Um so I think that we're at this spot where after this first summer, this was this was basically our hypothesis that we wanted to to validate or not.
55:19
This you know, is there actually like a a threat across different domains that that is protocols?
55:25
I think we've we've confirmed that there is.
55:29
And now I think that the next step is figuring out are there some big questions or some big problems that recur in all of these and can we get people to start thinking about solutions or you know, answers to those?
55:40
And then I think once we have those or at least like a framework for answering them, you get closer to something like you know, Adam Smith's like you know, say invisible hand or you know, like supply and demand.
55:51
Like if you're able to have these um yeah, these these these mental models or these frame of references, um then it's very easy for people to map on their local problems to them and and hopefully make some progress.
56:03
And same thing with science, you know, it's like you know, you think of like the scientific method is like whether you're a baker or you're a doctor or you're an astronomer, you can use the scientific method and you know, lay out some hypothesis, try to record observations and whatnot.
56:20
And then maybe you then tweak it, right?
56:22
Like the way that the scientific method works in astronomy is different than it would work you know, in in like high-end gastronomy today.
56:28
Like but at least yeah, laying down these these these mental models, these concepts, these theoretical frameworks I I don't know if that's our next step.
56:37
It's maybe two steps ahead.
56:37
We're at the spot where I think we have a good understanding of like the general areas of concern or in or or like interesting inquiry.
56:45
And then next step is like yeah, can we get some some common frameworks out of that? Um Yeah.
56:51
I I I think that uh you you your point there is very well taken about uh lacking a term or a label, right?
57:00
Because you think about uh what why did Darwin succeed with evolutionary theory where his grandfather Erasmus had similar theories, but he didn't have a label like natural selection which his grand his grandson did.
57:16
So when the grandson had the label, all of a sudden it caught on because it allowed us to allow it into our thought process.
57:23
Uh whereas Erasmus just uh you know, he had the ideas, but he didn't have the label.
57:30
But then the other thing that I think and I go back to the hand washing example.
57:33
I'm sure you're both very familiar with the Semmelweis uh experience when he took over the um the ward for pregnant women uh in Vienna and this is back during a time when male uh doctors felt that it was unmanly to wash your hands.
57:52
So, they would go from dissecting a cadaver to giving birth to a a child without washing their hands and Semmelweis who was given the command of that unit wondered, "Gee, I I wonder why the women who are attended by the male doctors are dying at a rate that's 4x the rate of the ones that are attended by the midwives?"
58:14
And so, they came up with a bunch of different theories and ultimately he disproved all of their silly ones and he said, "I think it's because you're not washing your hands."
58:25
And so, he instituted the protocol of the doctors cleaning their hands in a solution that was a lot tougher than soap. But here's the problem.
58:35
So, Semmelweis was political and he got involved in a political demonstration that the powers that be didn't very much like and that wouldn't didn't work out for Dr. Semmelweis. He was kicked out.
58:48
He was ultimately institutionalized.
58:52
But the person who took over from him with all of the data right there.
58:54
In other words, Semmelweis starts forcing all male doctors to wash their hands.
59:02
Death rate among women attended by male doctors plummets.
59:06
And he reversed it because he's like, "This is What is all this effeminate washing of hands for men?"
59:17
I How do we deal with that?
59:21
This is actually the particular example I've thought about in one particular angle it brings up is very important.
59:26
Um the sheer amount of time it takes for protocols to evolve actually puts them in a different class of um social political phenomena than regular technologies.
59:40
Like, you know, um there's this great article, I just looked it up again um by Atul Gawande, the surgeon who writes in The New Yorker.
59:47
It's called uh Slow Ideas, and he talks about like uh anesthesia and antiseptics, which are down the same story.
59:54
And these are the opposite of things like the iPhones, which, you know, comes out in a dramatic keynote, same as like OpenAI stuff yesterday. Everybody's like, "Wow!"
1:00:04
And like next year, millions of people are like eagerly rushing to try them, right? To the extent they can.
1:00:09
These ideas, they seem mundane, and it takes like a hundred years to like uh sink in.
1:00:14
And I think protocols tend to have this character of This is not a five-year big bang marketing thing.
1:00:24
It's not even a 10-year venture capital sort of investment cycle thing.
1:00:26
It's not even a 30-year government R&D NSF grant making or space program type thing.
1:00:34
You have to think with like 100 to 200 year timelines, which means that you are thinking about a horizon over which cultural values will change, prejudices will change, entire compositions of like the people involved in a picture may change, right?
1:00:48
So, in your example, yeah, Semmelweis started out strong with like good scientific reactions to what he was seeing, then his old prejudices kicked in, but then the whole phenomenon of like, you know, what they say about like, "Technologies don't get adopted because the proponents prove right and the critics prove wrong.
1:01:05
It's because the critics basically die out."
1:01:07
And the proponents are right, they have like the advances one funeral at a time, right? Exactly.
1:01:14
Yeah, and this is also it goes with like the Whitehead quote about like things going into where you don't have to think about them.
1:01:22
The automat- automaticity of things, uh, we we've talked in the last decades about like the idea of like, you know, digital migrants versus digital natives.
1:01:30
Like, people like my dad's age, he's in his 80s, still fumble with computers and phones.
1:01:35
People my age, we started out in, um, like, you know, uh, pre-internet era.
1:01:41
So, I grew up in the 80s with like early PCs.
1:01:43
Um, and then there's people who were born with the internet, then people who were born with, uh, mobile phones.
1:01:47
And each generation for a very long evolving technology or protocol, more things are kind of in their unconscious right when they're born.
1:01:56
And they're like, like, you know, the young babies today, they're like, "Why can't I touch my TV and it, uh, should do stuff, right?"
1:02:03
It's like they expect all screens to be touch screens.
1:02:05
So, that's a behavior that's gotten automatic.
1:02:06
Same thing with like, you know, it's not that men today are necessarily less chauvinistic than in Samuel Weiss's era, but you are born into an era where it's much more common to treat women as equal in like, uh, at least some workplaces like health care more as equals than it was in like the 19th century or whatever.
1:02:26
And even if you might have like the same like psychological dispositions, you internalize very different sets of unconscious behaviors.
1:02:35
And you're kind of like born unconsciously protocol literate in a bunch of areas.
1:02:38
Others that are being developed as you're, uh, coming of age, you might have more trouble acquiring.
1:02:43
And then there's a whole Douglas Adams, um, joke about this, which is like, "Anything invented before you're 35 is like natural.
1:02:51
Anything after 35 is against the nature of reality."
1:02:53
I think we're all like that.
1:02:55
Like, you know, all of us are going to go through that phase where we look at something that comes up in like 2040 and say, "This is wrong.
1:03:02
This is not the way the world should work."
1:03:04
And then we will be the crotchety old people who are like, you know, the part of the problem rather than part of the solution. So, yeah.
1:03:13
And, uh, one of the motivations for this problem that, um, Tim can talk about more is uh given the long timeline and the fact that you want these things to enter like a stable mature old age um which we talk of as the ossification problem in Ethereum. It's hard.
1:03:28
Like if you want like a blockchain protocol to like get a stable maturity like after you're dead, how do you plan for like the cultural environment then, right?
1:03:37
So, it's it's fundamentally a different level of technology problem.
1:03:45
The uh the thing that springs to mind is the idea of cultural evolution, right?
1:03:48
Uh and that speaks to what you just said.
1:03:54
Uh we we do culturally evolve.
1:03:54
And um you know, science advancing one funeral at a time is said but true.
1:04:02
Uh but another aspect that you spend time on in in your work is the fictional treatment of protocols.
1:04:09
I I was really intrigued by that uh because fiction can can be very transformative if it's done right.
1:04:19
And and you you give some great examples with some of my favorite authors.
1:04:23
So, if you wouldn't mind sharing a couple for our audience, that would be that would be great.
1:04:29
Yeah, I was thinking more in the context of the program, but yes.
1:04:32
I I it's kind of interesting that um while this program was going on at a personal level, I was very into like um narrative protocols in particular.
1:04:41
And my favorite right now has to be J. G. Ballard.
1:04:44
He was kind of like this um really towering figure in science fiction, but not as famous as like Asimov or Heinlein or others because he's like he's the science fiction writer other science fiction writers read to like get their big ideas.
1:04:58
And now that I'm like in protocol headspace, every single thing I'm reading by J. G.
1:05:02
Ballard is like striking me as this is like perfect protocol fiction.
1:05:06
Like I'll give you a couple of examples connecting to our program.
1:05:09
Um his novel High-Rise is about a bunch of people who live in like a high-rise and basically stratify into lower, middle, and upper classes and like turn the high-rise building into a war zone where they're fighting over control of the elevators and like throwing furniture down the stairwells at their enemy classes.
1:05:28
And it immediately struck me that this is like a perfect lens with which to look at one of our research projects.
1:05:34
Shinohara, she's an architect.
1:05:36
She did a research project on what she calls addressable spaces, which is how we organize the built environment with like, you know, numbers on doors, addresses, elevators that stop on floors, how you program elevators, like, you know, floor number 13 is missing in many American elevators.
1:05:51
Chinese elevators have like other numbers like weirdly implanted in.
1:05:57
And this novel is about like what happens when that kind of like friction gets overlaid with like sociology.
1:06:04
Um, but to connect to a couple of other points, again with J. G.
1:06:08
Ballard, two of his The Drowned World and Drought.
1:06:11
Both of them deal with deep time.
1:06:14
So, in Drowned World, the world gets flooded with like something that looks like climate change, but it's a different premise.
1:06:21
But as more and more water submerges the world, ancient prehistoric memories of like, um, you know, dinosaur eras crop up in the genome and people start having dreams of like the Triassic era.
1:06:33
And this was a very interesting connection point for me.
1:06:38
One of our researchers, Kia Kreutzer, she studied the relationship between memory and protocols and technology.
1:06:45
So, she goes deep into like, in the Western world, how memory technologies have played a role in like the evolving civilization and how like unconscious things get sort of bubbled up.
1:06:55
So, that's like, you know, the relationship between a fictional speculative fictional exploration of like a conceit, basically, that, you know, we have like deep archaic memories of like 100 million years ago in our genes, with like something quite like that happens with how technology becomes a prosthetic memory.
1:07:12
So, all of us, the way we are talking, we are actually in some unconscious way accessing technologically mediated memories that go back thousands of years. That's one.
1:07:23
Um and um drought is another similar one where like time freezes and there's like a temporality is a theme in like protocols in urban environments where it's like you're you never know what time and space you're in because it all looks the same, right?
1:07:34
Uh but the other connection point I wanted to make again with fiction um is um since you brought up the quote about like science progresses one funeral at a time, we had a project by Sarah Friend about the protocols of death.
1:07:47
So, she looked around I looked at like the cultures around death, funerary protocols, but not just like literal death of people dying, but worlds dying.
1:07:55
So, online worlds, digital worlds, what happens when an um online distributed game is shut down and an avatar you've become very attached to has to die.
1:08:04
Like this has happened many times and apparently there's like a huge lore and culture around managing these kinds of death experiences and uh she also uh one of the other subprojects that she inspired was about Orkut, the social network, which became really popular in Brazil.
1:08:22
And the Brazilian researcher we had in the program, she did a whole study on what happens when a social network dies and your entire social identity dies with it.
1:08:31
So, it's like a fiction you've created along the way.
1:08:33
So, this this spectrum between like reality and fiction doesn't it's not binary.
1:08:37
It's not that you know, there are novels that novelists write and then there's reality we live in.
1:08:42
One of the things we've learned through like this protocols research is things like memory, death, identities, and personas as expressed through protocols, they're like varying degrees of fiction that can go all the way to like completely like psychotic breaks from reality and you still won't notice if it's so long as it fits into the protocol, right?
1:09:01
It's like you could be completely insane, but so long as you're like following the rules of the protocol, it won't look insane.
1:09:08
So, yeah, there's lots of deep connections here and that way that we've been glad to explore.
1:09:13
There have been there's one lovely comic book that's like speculative fiction comic book about a woman exploring a hard disk memory that Na Hee Kim, one of our Korean researchers, wrote.
1:09:27
So, that's one of my favorites.
1:09:29
And another Canadian researcher, Ritika, she wrote a series of speculative fiction set in 100 years in Canada somewhere on like what would the world look like if it were completely protocolized?
1:09:39
So, yeah, we we've been able to plant a few seeds of like can you use fiction and design fiction in powerful ways to explore research themes and we're very happy and excited about the few seeds we've planted.
1:09:49
And hopefully as we continue this program, hey, maybe we'll discover the next J. G.
1:09:55
Ballard or, you know, powerful writer who can like explore this for us.
1:10:00
I'm thinking again about the game of life and and the what emerged from it from the very simple rules.
1:10:04
Is this something that you could kind of go protocol hunting in silico?
1:10:08
Could you maybe like set up a series of simple rules and then let them you know, do their thing over a million 100 million iterations?
1:10:22
And and would would we see emergence from kind of these in silico type things or do you think that these are things as you mentioned earlier in our conversation where these are these are being generated like matched from from nature, what we observe in nature?
1:10:39
You know, the termite example's a good one.
1:10:41
But by the way, there's there's all sorts of examples for human behavior that we find in nature like bees and ants and all of all of the above, right? They all have protocols.
1:10:54
And some of them are pretty for very unsophisticated creatures pretty sophisticated protocols.
1:10:58
For example, the dance of a being when they discover a new nectar or water source uh you have to use advanced calculus to be able to describe the the bees dance that points the way which like a GPS to the other bees here it is.
1:11:16
So, do you do you think trying uh a more randomized in silico 100 million generations would be useful or stick with looking at the natural world?
1:11:32
I think one one reason why it would be hard is protocols tend to operate almost on like a slower cadence than the main thing you're looking at.
1:11:39
Um you know, so if you use the example of like a bee reproducing it's like this reproduction is like on a on a very slow time scale relative to like what the bee is trying to do right now even though parts of the protocols sort of happen you know, like dancing around the flower.
1:11:56
Um and you can think of like maybe the human version of that is and that came up with that was like the diplomacy, right?
1:12:02
You know, you go to like the UN there's all these things that's like, you know, your country does and there are all these like micro actions that you know, diplomat does and what not and it but the the entire like diplomatic protocol plays on a much slower time scale than like the agent in it.
1:12:17
So, I I don't know if there's a way to to to observe or to see them without it being like emergent behavior at some point.
1:12:27
Um and and especially where like the even if you if you define a protocol, you know, blockchains are a great example.
1:12:35
We literally, you know, anyone can just write all the rules and you know, you define the protocol as is and you launch it.
1:12:42
But then the interesting stuff all happens as like a consequence of the emergent behavior.
1:12:45
Um you know, whether you look at Bitcoin, whether you look at Ethereum, um none None these things were just um None of like the interesting insights you got out of these things were there when the whole protocol was created.
1:12:56
Um so I Yeah, I I think it's Whatever Yeah, whatever whatever a test environment or or or or sort of lab environment you want for this, you need to allow enough time and and enough like interactions that to get the meta level uh emerging.
1:13:13
And that's what will show you whether like the the protocol is good or bad. Um at every step, yeah. Yeah, too.
1:13:21
I I actually want to provoke you to you go deeper in crypto on this because I think it's illuminating.
1:13:27
What's the lessons from like the testnet infrastructure that Ethereum runs?
1:13:30
Like before you do major forks and upgrades, you run a bunch of testnets.
1:13:34
Do you learn like Game of Life type emergent lessons from those?
1:13:40
Um short answer is actually it's really hard.
1:13:42
So, we I wish we would learn more.
1:13:45
But, you know, the Yeah, this is maybe a good example where like, you know, on Ethereum, you want to do a change and so you you have like a testnet that's the equivalent of a staging environment in like normal tech.
1:13:57
Um and so we can test it like things work in the happy case and in some simulated unhappy cases.
1:14:01
Uh but blockchains are like super adversarial environments when they're live, you know, so there's no way we can ever replicate the amount of just yeah, perturbations that we see on the main Ethereum blockchain on any testnet.
1:14:16
Uh even though, you know, we build tools to like spam them and break them and we'll like shut it down and and and and and whatnot.
1:14:21
Um there's a degree to which just the yeah, the adversarial environment and the emergent behaviors out of that uh are really hard to replicate.
1:14:30
So, yeah, I guess this convinces me that like observing in like a a sterilized environment, um you know, like maybe the most valuable thing you get out of that though is just simplifying, right?
1:14:42
So, when we when we launch something and it breaks on a testnet, obviously it's not ready to go on mainnet, and there's usually a root cause we can identify and fix, you know, and and then once we get to the point though where we think it's robust, um there's always still a an amount of uncertainty between going on on sort of the real network because you just can't predict that.
1:15:02
Um and I expect a lot of situations are similar like that, you know, think about like pandemic planning or or or emergency response, right?
1:15:09
There's been some of something like that where you can do drills, right?
1:15:12
Like you can, you know, think, "Okay, what would happen?"
1:15:16
Or, you know, war war gaming is like another example.
1:15:18
Like even though you have all the people in the room thinking about all the bad stuff that could happen and all the edge cases, there's just some level of like I guess it's not the reference piece like the swarm behavior that can emerge where it you really can't predict that super well from the start.
1:15:33
Or maybe you can, and again, this is maybe some something where, you know, there's some insight of like protocols that could help us predict this emergent behavior better that we we just don't have today.
1:15:45
Um but it's yeah, it's really hard.
1:15:47
I want to build a little bit connecting this back to what you said about the game of life.
1:15:51
Like I'm Now that I think of it, the game of life, it's a cellular automaton, but it's basically deterministic.
1:15:57
It's a grid, and the rules apply cell by cell, and there's no way for even though lots of lovely complexity and interesting dynamics happen, and people spend their lifetime like, you know, exploring and trying to create particular things on the game game of life and grid, it's fundamentally a closed algorithmic environment.
1:16:16
And if you try to like turn the knob towards what does this look like when you add real-world messiness and randomness to it, you get to like a You guys may be familiar with von Neumann's universal constructor, which is basically like a Turing machine on a you know, 2D automaton grid.
1:16:32
But one of the insights from that is that is Turing equivalent and it can compute everything, but if you allow for randomness, which is like, you know, mutations in genetic evolution, it's like cells in the grid can randomly be flipped by like noise from the outside.
1:16:48
So, you start making the cellular automaton simulation more random and capable of mutation, it starts to mimic open-ended evolution.
1:16:56
So, my sort of philosophical takeaway from that pair of examples of like, you know, James Conway's Game of Life versus, you know, universal constructor with like mutation possibilities is open-ended evolution is like categor- categorically a higher level of phenomenon than, you know, pure simulation.
1:17:15
And the more open an environment is, the more careful you have to be about what lessons you draw from more closed environment.
1:17:23
So, even the termite example, like, think of like termites building their hills in a relatively stable ecological niche, where ecosystem rarely gets disturbed, they may be in some sort of equilibrium with their predators, but like mutations from cosmic rays that turns the termites into weird new kinds of termites are pretty rare.
1:17:42
So, for most periods, if you're trying to draw inspiration from termites building hills, you're actually drawing inspiration from a fairly closed automaton evolutionary environment, kind of like, you know, Game of Life.
1:17:52
Um, but with adversarial environments, more complex open environments where randomness can get injected in from all over the place, things get a lot messier.
1:18:02
And you see this again, like, you know, in the difference between regular gaming of the sort you might do on Xbox or PlayStation with like uh god-like game designer who's designed the game and all its rules versus something like Dark Forest, which is the blockchain game I mentioned earlier. It's much weirder.
1:18:19
Weirder things can happen when you open up uh open-ended network of adversarial phenomena.
1:18:24
So, I think that's why Tim is being a little cautious and like jumping on your thumb there.
1:18:31
But no, and that's a great point.
1:18:31
Like, another again, coming back to the TSA example, you can think of this as something where because it's so high stakes, there's a ton of energy being put in reducing the variance in the environment.
1:18:42
So, like, you know, we can have simple TSA protocols that can be enforced anywhere because we spent so much time making sure that airports are like the most sterile environment.
1:18:49
But, if you compare that with, say, you know, like actual war zone and and and military strategy, you know, that has to allow for like a much wider range of outcomes.
1:19:01
Um but, you need like a much higher agency and and context person to like actually put those into practice.
1:19:09
Whereas, you know, scanning, taking the shoes off, and whatever scanning the bed is like simplified, but it's it's the fact that it happens in an airport that allows you to have this that simple of a protocol work at scale. Yeah.
1:19:20
One more to last build on this, like, if you can see what happens when um closed protocolized swarm environments encounter things that they weren't designed for.
1:19:29
Like, have you seen those videos of um army ants marching in a dead spiral?
1:19:34
That's what happens when the environment is more complex than the protocol is capable of doing.
1:19:38
These ants are marching in circles till they die.
1:19:42
And that's what happens with TSA level sophistication in a war zone.
1:19:46
I had one last on on Rafa's example of swarms.
1:19:49
This is also why you see these traditional companies like stumble on Twitter all the time, right?
1:19:53
So, if you think of like a big traditional company, they think of their comms environment as this sterile controlled thing.
1:20:00
And then, you know, something happens on Twitter and and and it's the equivalent of like, you know, the airport turning into a war zone where you just don't have any uh any notion of like how do you deal in an adversarial environment?
1:20:14
And yeah, um Yeah, so that's a big distinction. Yeah.
1:20:20
Agi- agility being one of the characteristics of a uh new company that we have found, at least, uh being a good indicator for how long it might succeed.
1:20:30
Um if they if it isn't agile, it it's a that's a that's a big red flag for it.
1:20:39
But I also wonder given the examples that you just gave like TSA.
1:20:42
So so TSA is obviously designed for other things, right?
1:20:49
It's it's security theater rather than uh I mean like if we think about it that way and we think about okay, what are the protocols to make people feel like they're safe, then then some of those TSA protocols might make sense.
1:21:05
But the the it seems to me that they were looking for for that as opposed to like hey, how do we make sure that lunatics with bombs and knives don't get on here?
1:21:18
I think that's the most charitable interpretation.
1:21:20
Um and and I I'm trying to be charitable.
1:21:25
Yeah, yeah, but and there's like an interesting Bitcoin parallel there after but if if you if you assume, you know, these were done post 9/11, you know, you can assume that like actual security was also a concern um and they tried to create a protocol
1:21:40
that was actually, you know, meant to improve actual security and maybe it's not that great and then you sort of you you sort of rewrite history a bit and you're like oh well it was it was always about security theater and it's actually not that bad. Um and I think
1:21:52
Um and I think this is something why second order impacts are hard because Bitcoin is is is sort of a a version of that story where, you know, you have the you have the white paper that says it's peer-to-peer electronic cash um and, you know, Satoshi's saying like we're going to use this, we're going to buy pizza from each other and whatnot uh using these Bitcoins.
1:22:11
Um but then we realize, you know, 5 10 years later it's actually never going to scale for transaction.
1:22:15
So you can sort of rewrite a history and say well it wasn't actually cash, it was more like digital gold and and kind of go along with that.
1:22:22
Um so I don't know if enough about the TSA to to know if it's like if it's actually if it was security theater from day one, but there's definitely this aspect of like the protocol gets hard to change.
1:22:32
So like, you know, maybe rewriting the social contract around it is simpler than say, "Oh, if we if we actually wanted security and we're not getting security today, what would we have to do to to change, you know, the TSA?"
1:22:44
Or if Bitcoin actually wanted digital cash and we don't have digital cash today, how would we have to change Bitcoin to you know, get that?
1:22:49
Um But it's a that's a really really hard problem with a protocol that's adoption.
1:22:55
But but that also brings up the question in my mind, which is you know, I you read about like military, for example, heavily protocol oriented.
1:23:05
And and yet you needed a void to rewrite a lot of those protocols, right?
1:23:14
So if you're not familiar with Boyd, he he he basically they're still using his aviation and fighter pilot techniques, but the problem was they were all against current protocols when he developed them.
1:23:28
And on top of that, he was a in the eyes of his superiors a very unpleasant person to have to deal with.
1:23:36
So jumping off from the Boyd or improving existing protocols, is there a path that that leaves room for that actually for the Boyd's of the world?
1:23:50
Y'all actually jump in there with a a similar follow-up question there.
1:23:51
Is that is there leverage in breaking protocols in tactically breaking certain protocols? Yeah.
1:23:59
I'll quickly just connect the points to a couple of research projects that touched on these.
1:24:06
So one of our researchers, David Lang, he was one of the pioneers behind this project called the Open ROV, which is an open source underwater submersible.
1:24:14
And that's evolved now, and one of the things he's working on now is an underwater connector.
1:24:19
So think of it as like a USB cable connector for making underwater connections, and it's a revolutionary kind of standard that will create open-source underwater hardware.
1:24:27
And he wrote a whole research project on the history of standards making and protocols making, and he identifies like how it went over multiple historical eras to where we are now in this age of like disruptive standards setting.
1:24:42
But, literally you take like a Boyden approach to the This is an old entrenched industry with its old protocols that are like completely controlled by incumbents.
1:24:52
And since if you want to revolutionize things, you need a new protocol, how do you do that?
1:24:57
Well, you kind of hack it in a different way than you would expect.
1:25:00
You don't like make up a proposal and go to the standards committee, blah blah blah, like, you know, go through the front door.
1:25:05
You come through the back door, you like demonstrate the technology, maybe a small startup uses it, and on the margins you make it work, and suddenly customers are coming to your new cheaper standard, and eventually you pressure and then attack the incumbent.
1:25:18
So, yes, Boyden strategy does apply directly to like things like standards making and protocols making.
1:25:23
That's one connection point I wanted to make.
1:25:25
The other connection point to our program was The one of the first guest talks we did we have a guest talk track going on, which was weekly during the summer, now it's bi-weekly on the off-season.
1:25:35
But, one of the first ones uh was Jeff Manaugh, who is the author of a book called The Burglar's Guide to the City.
1:25:44
And he talks about how it's a lovely book, by the way.
1:25:47
He wrote an essay earlier called Nakatomi Spaces, which is about the movie Die Hard and how Bruce Willis never seems to walk through corridors or through doors.
1:25:57
He's always going through like air ducts and ventilation shafts and things like that.
1:26:01
And that's He called that Nakatomi Space after Nakatomi Towers, and that led to this book called The Burglar's Guide to the City.
1:26:06
Guide to the City. And of course, the idea there is that you approach a protocol environment with the eye of a burglar of how can I hack it, how can I sneak in through like the spaces the designer didn't expect and connecting
1:26:19
this more recently Jeff Manel also wrote connecting this to the Israel armed forces official military policy which is when moving through a built environment never go through the front door like you should burst up through the floor or down through the ceiling. Now, this was
1:26:35
Now, this was like fun exciting theory to read about like 10 years ago when Jeff wrote that essay and now it's like a dark thought to entertain in the context of the war that's ongoing.
1:26:46
that's ongoing. But the point is urban environments are another example just like you know connector standards where because on the one hand you have this hard extremely slow changing expensive to change urban built structure environment that's like if you wanted to change rewrite the city it's very hard to do so at a hardware level
1:27:05
and then you've got the software level of like you know traffic flowing sewage being flowing or like garbage being collected and one of our researchers Drew Austin who's an urbanist who used to work for Uber and is now an independent guy he kind of like took this line of thought in Jeff Manel's talk and developed it further in his
1:27:23
essays about how the city is a protocolized environment where there's the hardware of the city and then the software of the city and in between there's the soft protocol there of the city and the way to like advance like urban planning agendas and like improve the quality of life in cities and so forth is to take this
1:27:41
burglar's guide approach but from like you know white hat hacker way and ask can you actually protocols of the city to work better like can you like change traffic laws or traffic signal blinking patterns to make traffic flow better when you know driverless cars are on the streets like this is an actual problem with the cruise cars right? So yes it's
1:27:58
So yes it's very germane and we did talk a lot about Boyd and like OODA loops and stuff throughout the program it was one of the others attractors we were circling.
1:28:06
So yes very definitely because of their inertial long term how they develop a lot of unconscious weight and mass.
1:28:17
I would even argue that this is literally the only way you can advance protocols once they're like um entrenched and set it like beyond the point, you can't come straight at a protocol.
1:28:26
You have to hack it with like a boarding approach.
1:28:28
That's the only way to advance um the sophistication of a protocol. Yeah.
1:28:35
To bring it back Jim, you mentioned you know, when you think of companies you know, agility is like a you know, something that that that you you see is like a sign of success.
1:28:44
I think for protocols it's a bit different because in a way it's almost like this two-edged double-edged sword where um if the protocol is too easy to change, there's also risk for that.
1:28:54
Like you gave the example of the the the the hospital where you know, people start washing their hands, things go well, and then some guy's able to come and and revert that change.
1:29:05
Um this it makes sort of the protocol worse, right?
1:29:08
Um And obviously, you know, in in Bitcoin we have the famous like 21 million coins.
1:29:14
If anyone can come and change this, sort of breaks the protocol.
1:29:15
And I I think there the two things there is like one like, you know, if you think these protocols are doing good things, bike shedding your being able to always revert things sort of chips away at like the the the the coordination savings that you get out of the protocols.
1:29:33
And even beyond that, the the idea that things might change in the future reduces assurances you have in in the protocol.
1:29:41
Um and you know, whether this is you know, you look at politics, uh you look at technology, something like that.
1:29:47
Um you know, if if for example, um you know, the email protocol of my company will change in a year, then you might be much less willing to write like an email client because you know, all of your work goes to waste.
1:29:58
Um And so there you know, there's this is a really hard problem to deal with though because of all these these sort of second order impact of protocols.
1:30:08
Um The and you know, there's there's different approaches.
1:30:11
I think as you look in in crypto where I'm I'm most familiar, Bitcoin has this extremely strong stance of, you know, ossification where they've agreed basically to never change their protocol and you sort of, you know, push everything elsewhere and and just sort of pray that the second order consequences are actually going to work out in a positive way.
1:30:32
Um on Ethereum we have this concept uh Vitalik wrote this this blog post a few years ago about called functional escape velocity.
1:30:39
So, the idea is that maybe you keep changing your Ethereum until you see enough stuff built on it um that that second ecosystem has a life of its own and then you can slowly potentially stop changing it and and sort of, you know, trust that instead of that you're trusting the second order consequences directly, you're maybe trusting the second order like stewards, right?
1:31:02
Saying like, "Okay, if all these people are like happy on the protocol, maybe we can like, you know, stop changing this thing and and trust them that they they keep going forward."
1:31:10
Um And then I think at what you were saying around like the idea of like a backdoor and and having to come like obliquely with changes.
1:31:17
Um I agree this has been true like pretty much in every case and this might be one of like the most underrated innovations in in in blockchains and in in tech where the concept of forking, right?
1:31:28
Where uh with even with normal software, so normal software uh for listeners who aren't familiar, if you have code on GitHub um and I'm I'm maintaining your software project there and it's open source, anybody can take that code and modify it and we we call this a fork.
1:31:43
Um and this is great because it means, you know, I don't have to rewrite the whole thing uh from scratch and I can benefit from your changes and maybe I I I send uh I send some changes in as well.
1:31:53
Um but the thing you don't get out of that is like the network effects.
1:31:56
So, for example, imagine you have something like Twitter, um even if you gave me all of the Twitter code base and I launch, you know, timtwitter.
1:32:04
com, it's not sufficient to get people to come to twitter. com.
1:32:10
Um so the the one innovation we have in blockchains that's actually very neat is the idea of like this shared state.
1:32:13
So if you own Bitcoin or Ethereum, all the nodes on the network know that, you know, I have one Bitcoin, you have two Bitcoin and and and all of that.
1:32:24
And we can fork the actual state where if you want to make a protocol change, say that, you know, Venkat, you wanted to double the amount of Bitcoin that there will be because you think that's the way forward, not only can you change the software, but you can bring and copy everybody's balance over.
1:32:37
So you can say, you know, on Venkat Bitcoin, Jim can come and his coins are still there and the rules of the protocols have still changed.
1:32:46
And using that, I think might be like that idea that you can fork not only like the software, but also the current state of things and bring the network effect and kickstart it.
1:32:56
it. That might be a way where in certain domains, you can get these like different flavors or different, you know, ideas of a protocol competing against each other in a way where you're when you're competing and and coming in as a as as a newer player, you don't
1:33:12
have to recreate the network effect from scratch, which is I think one of the biggest challenges when you look at things either to have like technological network effects, you know, through distribution or even that have sort of regulatory modes, right? Like this is
1:33:22
Like this is the whole thing around, you know, if you just look at something like the TSA and whatnot.
1:33:27
The reason you can't go in and change it is all these layers of of bureaucracy and you can't you can't just create another airport side by side with the same airport and, you know, have everybody pass through both and and and and sort of run the experiment.
1:33:38
Um well, yeah, with with technology, you can.
1:33:41
So I think this is something where hopefully, you know, blockchains can become a useful case study of like what does it look like to make changes to a protocol where you get the network effects and you get to actually see the sort of user preference and and um yeah, and build on top of that.
1:34:01
Ed Ed, I think I think you had a question there. I had one question.
1:34:04
Switching gears slightly, but let's go back to something we mentioned earlier.
1:34:09
Um and you sort of you mentioned it in the context of stories and storytelling, but one thing I really noticed when we were at the perspective was that several different projects seem to be focused around memory.
1:34:23
Um there are there are seven or eight.
1:34:23
I can't remember exactly how many, but it was a sizable portion of them.
1:34:28
What about memory makes it such a sort of ripe topic to to think about in the in the context of practicals?
1:34:35
That's a really good question and I feel like we should make that a challenge question for people next year.
1:34:41
But I'll give you kind of like a bad temporary answer.
1:34:43
Um it actually connects to what Tim had just said, which is a lot of the intelligence in any technological system or built environment is actually in the state, not the structure.
1:34:55
So, the history of the state, which is what memory that's one definition of memory, is actually the most valuable asset in the civilization.
1:35:03
Like, you know, the people often play that fun game of like, if civilization were destroyed tomorrow and you could like save one book, what would you save?
1:35:11
That tells you that we sort of intuitively know what the valuable thing is.
1:35:16
And I think this current generation of technology is more explicitly recognizing that everywhere.
1:35:24
Like this is happening in AI as well.
1:35:25
Like, if you remember uh 10 years ago there was this Google paper which kicked off the big data revolution.
1:35:35
The main premise was like, more data and simpler algorithms beats less data and more complex algorithms.
1:35:41
And that was the big data revolution.
1:35:43
And you extrapolate that and that's how you get modern deep learning, which is almost all the intelligence is in the data.
1:35:48
You look at like PyTorch code for machine learning, it's like a few hundred lines, it's like some of the simplest code imaginable.
1:35:53
It does, like you know, the simple like back propagation and other thing compared to like really complex old code, all the intelligence is literally in the data that then gets digested into the parameter weights and so forth.
1:36:05
And I think something similar is true of blockchains as well.
1:36:10
The actual like code for blockchains, yes, it's complex, but the real value is like just the sheer amount of state history, and that's actually a big problem with all blockchains.
1:36:20
Like, how do you deal with the snowballing state?
1:36:21
And I think one of the reasons memory has been such a big topic for us is um these large time contact constants of how technologies change, how maintaining state and like propagating state not just through forks, but through like very radical paradigm shifts in technology is important.
1:36:43
Like, we talked earlier about like, you know, things like corporations and organizations, they are our current generation of organizing technology, but they carry streams of data that go back to the Roman Empire, for example.
1:36:53
I I think this is apocryphal, but the idea that, you know, modern railway gauges and road dimensions go back to like cartwheel wheelbase lengths to the Roman Empire.
1:37:04
So, you can see that uh the phenomenology, the data, it's fundamentally much more important than the superstructures we built to like exploit it or do things with it, which means that if you had a choice at I I'm to Oh, yeah, the code I'm thinking of is um Brooks Frederick Brooks on the Mythical Man-Month.
1:37:26
He said something in like in the 1960s with like PDP operating system, something like if you show me your code and hide your tables, I will understand nothing, but if you show me your tables and hide your code, I will understand everything about your program.
1:37:38
So, even back in the 1960s, it was apparent that the intelligence and the value in any sort of like technological assemblage of things lies in its data and memory.
1:37:48
So, I think it's not an accident that several of our researchers gravitated on like memory and in related more time as kind of like the important things to poke at in understanding protocols.
1:38:00
So, evolutionary dynamics, history, how is state perpetuated, all that is like a very critical thing.
1:38:06
And uh yeah, thanks for asking that question because this now makes me think in this just program design we should like provoke people to think more about this angle.
1:38:17
And it's it's one where it flows so well between like the social and human side and the technological side.
1:38:22
I think this is maybe for all of the other domains it gets really hard to reconcile those two except maybe the built environment which is something where we had a ton of projects as well.
1:38:32
These domains were like you can feel both the like social and human side and the technological side tend to resonate um more than say I you know like say our work on blockchains and I get the feel for like the social side of blockchains cuz I'm so far down that rabbit hole.
1:38:48
For most people it's like just a technological thing.
1:38:50
But I think like yeah, memory and and like built environment are these things where anyone can think of both sides and and have like a strong intuition for them and and it is sort of hints that there's something there between those two those two rounds that that you can you can you can touch. Yeah.
1:39:11
Well, this has been an absolutely fascinating conversation.
1:39:14
I I feel like we got maybe through half of my notes uh and Ed uh how did we do?
1:39:23
You think we we maybe have to do a round two here?
1:39:28
The half is optimistic, but yeah, maybe that.
1:39:31
So So, uh this is absolutely fascinating.
1:39:33
I I think that this is a great project because I think you're right uh and that explicitly looking at this and thinking about it and having these challenge questions are only going to improve the process of protocol and I also agree that we are moving into a new era where I think the open discussion and propagation of uh, the ideas around protocols uh, is uh, hitting exactly at the right time.
1:40:06
So, um, where can everyone find all of the materials that hopefully this uh, podcast will uh, will get them to run off to and and click on?
1:40:18
Yeah, um, so the main place to go is summerofprotocols. com.
1:40:22
Uh, so all of the research that was done throughout the summer is going to be published uh, openly and for anyone to access.
1:40:28
Um, so we're going to we've also after reviewing everything it felt like a lot to drop on people at once.
1:40:34
So, we decided we're going to we're going to scatter the the release of it over the next, you know, 6 to 8 months.
1:40:40
Uh, so summerofprotocols.
1:40:42
com is the main place you guys can go there.
1:40:44
Everything uh, will be available and we're also putting together a limited set of physical kits which contain all of the research and a bunch of other sort of tools and and interesting uh, collaterals to get people thinking and and working with protocols.
1:40:58
Um, the kits won't be for sale, they'll all be available for free um, and we're trying to get them to people who want to actually, you know, build study groups or sort of try to advance protocol studies one way or or another.
1:41:11
Uh, so if if your listeners would like to have a kit, um, we can give 10 away to people who email us to hello@summerofprotocols.
1:41:20
com um, and tell us how they how they're going to use sort of the physical version of the kit to kickstart whether it's a meetup, a study group or you know, put it somewhere where interesting people will get the the engage with the stuff.
1:41:33
Um, so if you just want to get access to the research, summerofprotocols.
1:41:36
com and if you'd like uh, to get one of the physical kits, send us a note to hello@summerprotocols.
1:41:42
com and explain to us what you do with it and we'll be giving those out in the in the next couple of months as well. Terrific.
1:41:50
Well, guys, this has been absolutely fascinating.
1:41:52
I think the work is fantastic.
1:41:55
Thank you for joining us and for helping our listeners and viewers understand protocols a little bit better.
1:42:02
I think everyone needs Thank you so much for having us. Cheers.