Daniel Schmidt argues that mazes are a minimal formal instantiation of protocols, not just a metaphor, using his own maze-drawing history and games to illustrate concepts like hard vs. soft protocols (physical walls vs. crossable paper lines), scale-entangled vs. scale-lined navigation, and stigmergic memory. He demonstrates an ant colony pheromone simulation, a multiplayer infinite maze with shareable 'pebble' markers, a maze-design platform separating protocol-author from protocol-navigator with heatmaps of friction points, and an art-gallery maze as medium. Venkatesh Rao joins to explain his 'clockless clock' maze, a topological map of thought-space (rat/goat/crow vertices) designed by Schmidt to represent arrested development and spiraling toward nirvana.
0:02 Um, next up we have um Daniel Schmidt to talk about mazes. Take it away, Dan.
0:19 You are on mute at the moment, but we see your slides.
0:23 Here I am. Yeah. Hey everyone. Um, yeah, thank you so much, Timber. Um, and for anybody that is uh just listening, I would love for you to watch the slides if you can because it’s all about uh showing you images of of mazes. So would love for you to follow along. And in general, this talk is going to be about um the strong connection that I see between mazes and protocols. And a lot of that was thanks to the work that this institute is doing and the research happening that really opened my my eyes to this connection and led to me being even more excited about mazes than I was before.
1:00 Yeah. So, when I was a kid, um, I started drawing mazes and I had no clear idea like why I was doing it. When I was seven or eight, I would sit on the couch and I would draw these endless maze forms. Uh, people would tell me they were like brains or intestines. I was always disturbed when people told me they looked like in intestines. I much more preferred the brain metaphor.
1:24 Um, I think most people would stop drawing mazes after they were a kid. But when I was in college, I was sitting in an art history class and I was somehow inspired to draw them again. But now I was more interested in the visual form of the mazes. For example, in this one, it was inspired by a philosophy course. And in the image, there’s no clean boundary between the sky um and the mountain. And while maybe it’s telling a story to some degree, uh the maze itself is not telling a story. It was like I was filling up empty space with maze forms.
2:00 Then my mazes would devolve even more. Um here is like the maze form, but there’s actually no starter end to this maze and they fragmented more and more.
2:12 Um, then I would take maze templates and I would fill them with different kind of maze forms inside of them and I would be interested in seeing the same template repeated again and again and again.
2:30 But I would say um the first really major breakthrough I had with mazes is when I started using the topology of mazes to tell stories. For example, this maze tell us the story of how the brain operates. Um, this is not entirely uh don’t read into this into too much detail, but it’s a kind of crude model of if you start in the left visual field, you find a direct path to the end of this maze, which is a verbal expression by going straight from like the left side of the brain through the corpus colossum to an expression. where if you would try to solve the maze by starting on the in the right side of the brain, you wouldn’t find a path through to an expression. Sort of like how it’s hard to express ideas from like the right side of your brain.
3:16 Um, a lot of my life I spent building startups and this maze was about kind of my experience of building companies. Um, and it was specifically inspired by Peter Thiel’s book 0ero to one where the goal of this maze is to get from 0 to one. And the first milestone is to create a product that people love um which is in the maze called tech user fit. But then if you try to prematurely scale that technology uh you would hit a dead end in the maze. And to me this is a lot of the experience of building a company is that you’re always tempted to want to skip stages ahead of where you are. And this maze and you get punished for doing so. This maze tells that story.
3:59 This might have been one of my more elaborate topological mazes uh inspired by Gutenberg’s invention of the printing press. Um here Gutenberg had a vision where he wanted to print copies of the Bible at scale. But if he tried to go straight using existing technologies to achieve that vision, he wouldn’t be able to find a path through. So we had to take a step back and get inspiration from other domains. By the way, this is uh based on Arthur Kesler’s telling of the story in his book, The Art of Creation. And Arthur Kesler has this idea of by association. Um so Gutenberg got inspiration from both coin making and the wine press. Um and that’s how he ended up making uh having the invention of the printing press. Um so I won’t go into too much detail, but if you kind of follow your way through this maze, you go through coin making. He then tries to um create the the invention but it was too uneven how the books were being pages being printed. So he then had to get inspiration from the flat surface of the wine press uh to eventually make his creation come true.
5:08 U and finally this topological maze was inspired by Copa’s novel the castle um which tells a story of a man trying to navigate an impossible bureaucracy. And to tell that story, this maze is impossible. Where if you were to try to get from the um the person here at the bottom to the castle, you’d find there’s no pathway through.
5:33 Um so now I want to be more explicit about the connection between mazes and protocols.
5:41 So here is a simple maze. Uh, but the simple maze will tell you what valid movement is. It’ll show you invalid movement. For example, you can draw a line from start to end, but if you ignore the walls, that would be considered invalid. It shows you reachable states, unreachable states, junctions, loops, uh, terminal states or otherwise known as dead ends. And like protocols, mazes show a full possibility space without prescribing the behavior inside of it. And uh to go a level deeper, mazes and protocols can be classified by the same properties. So consider the difference between a paper maze and a physical maze. With the [clears throat] paper maze, you’re able to cross the lines even though you’re not supposed to. So it’s like a soft protocol. With the physical maze, you’re not able to go through a wall. So, it’s like a hard protocol.
6:43 So, consider like an expense policy. A hard expense policy uh would actually decline your payment if you tried to go over like the the budget allocated for something. Where a soft one, it would flag it for your manager to approve it, but the payment would go through. With this maze, it’s a unic maze on the left, which means no junctions. On the right, it’s multicursal. uh two um airplane protocols might have that same distinction between a universal or linear checklist
7:14 um and a decision tree for troubleshooting an issue. Uh now we’re getting a little bit more nuanced and the difference between a scale lined maze and a scale entangled maze. On the left, if you follow the outer structure here of the orange maze, you can basically see your way through the maze. The local details don’t interfere with your ability to travel through. where they’re just like trivial to navigate. Here, the local details matter. Um, where you’ll find yourself going forward and backward in the high altitude maze to get to the end. Um, so you’re not able to basically tune out the details. You really have to constantly toggle your brain between lower level details and higher level structure. um similar to maybe different like shipping protocols where if you’re shipping like a printed book, the geographical direct geographical path might show the way that it’ll ship. Where with a more sensitive item like a lithium ion battery, the local details matter, you might have to go backwards uh to a dangerous good origin before you can kind of go forward to your destination. like lots of I think distinctions here in terms of protocols where local details will affect your ability to navigate through the bigger picture space.
8:29 Um in this maze um they both have the same number of dead ends. I believe they both have seven dead ends. But on the left um it’s trivial to see where the dead ends are. You can just easily ignore them. on the right, you have to fully uh kind of commit to a path and only there do you see the dead endex.
8:49 Um with protocols, this might be the difference between raising money via as a small business u where you can quickly get feedback, oh like we don’t have the financial qualification to get the small loan, you immediately get a rejection. Uh but with raising money from VCs, um I know this from direct experience, you have to like fully commit to a path and try really really hard, spend lots and lots of time before you know whether or not that path is viable. So it’s that same distinction we saw between those two mazes. Um here the maze on the right might look simpler than the maze on the left, but the maze on the left um can actually be solvable with one local rule. People often talk about how if you’re in a maze, you just have to put your hand on the right side of the wall and just keep your hand on that wall and then eventually you’ll get to the end. Uh that works with the maze on the left. It does not work uh with the maze on the right. With the maze on the right, you’ll go into a continuous loop forever. Um similar to the distinction between maybe two different customer service protocols. On the left, you might have an issue and it’ll get escalated up the chain and eventually it’ll get resolved no matter what. On the right, you get um it gets passed off between different peer departments and you might end up in a loop between billing technical support and account security kind of over and over and over again.
10:16 Um, and here is like a similar um idea where it’s a web crawler where on the left you have no memory and if it follows the highest priority link it’ll just go into a loop forever potentially where on the right now we’re just getting into the topic of memory just the simple idea of knowing what pages have been visited before the crawler can break that loop.
10:39 So now we get into kind of the next level and this is really what the protocol uh kind of idea stimulated in me especially when I got involved in the stigmi group. Um if you have a fog of war maze, we’re going to add fog of war to this maze. Um now we’re not just thinking of protocols as mazes, but it’s unlocking our ability to use mazes to study navigation protocols, coordination protocols, and memory protocols. Um, so now for the next part of this, I’m going to go into some recent u maze games um that I’ve been building to really play around with this concept.
11:29 Um, so these are mazes that um play around with the concept of stigmargy or leaving traces in the environment.
11:38 Um, this one, I don’t know if anybody else is seeing a lot of videos in their various feeds of cave divers being tragically lost in underwater caves. U, but for better or worse, I was seeing that it inspired this game. Um, so here I have a character. I’m going to go down [clears throat] and I’m going to try to find the treasure. And in this game, I’m actually running out of oxygen as I go.
12:06 Um, I found the treasure. Now I’m going to get out.
12:13 Um, that one was pretty simple. I was able to even do this one without an oxygen tank.
12:22 I’m going to deposit that money in the bank. But now I’m going to get some oxygen tanks. And now I’m going to try to solve a more elaborate underwater maze. Uh but here I’m going to use a a memory protocol which is what cave divers use called guidelines.
12:41 So I’m leaving this guideline. In this case it’s only for me. It gets really interesting if you think about how other people have left guidelines and can you trust them or not. But for now I’m just going to use it for myself. I admit that I have solved this maze before. So I kind of know where I’m going. You can imagine the drama of somebody doing it without knowing where they’re going. But I’m going to try to find the treasure. Collecting the gold. Um, and now because I have this guideline, I’m not going to be tempted to go down that wrong path. I can find my way out of the maze really quickly. Without the guideline, it would be very easy to take the wrong fork back and then die in this underwater maze.
13:30 All right, now I’m out. Um, so the next one I’m going to show you is playing around with the ideas of um having sharable kind of memory traces.
13:43 Um, so here the goal of this maze is to get as far north as possible. By the way, I’m sorry this is a little bit dark on your screen, so hopefully you can see it well enough. So I’m trying to go north. Um, what you can do is you can leave uh pebbles by hitting P. Um, and it’ll tell you uh kind of where you’ve been. But the tricky part is that, you know, I might have a convention for these pebbles that other people don’t have. For example, here’s a pebble that somebody else left. Um, I don’t really know what it means. I actually can see like who the person is that dropped it and when, but I don’t really necessarily know their language for when they drop it. But the game kind of rewards you for learning the language of other people. um and seeing who can get to the northernmost point in the maze in the fewest number of steps, which really rewards your ability to interpret other people’s protocols. And of course, you could try to be deceptive and leave ones that are um going to take somebody down a wrong path.
14:42 Um this next one, we actually had a whole workshop about it um kind of earlier and hopefully I think some of you in the audience were were part of that. Um, but it’s using um the stigmary concept uh very literally um with ants that are trying to solve a maze. Um, and here it’s just so exciting how these ants are really dumb. They have no memory of um of anything besides their last previous spot, but they’re leaving traces of how to get back home when they leave and then traces of when they find food, how to get um find the food and then eventually they’re able to form ant highways and basically solve the the maze through stigmar alone. Um this is actually another interesting one where you can see a failure case where they got trapped in an antl um over here. Um, so this is a little bit of a hard one, but actually it also teaches you about how the properties of pherommones
15:38 um work. So I could crank up the evaporation rate, remove the trail bias. Um, and now we’ll see this antill uh disappear because there’s no longer um uh because their their pherommones disappeared. Now I’m going to crank it up again and see if these poor ants can solve this hard maze. Maybe just to give you guys the satisfaction of seeing the ant highway, I’ll place another food source there and that should be a little bit easier and we’ll see the mazes, the ants form this ant highway. Um, I won’t go into the full details of this game. We did loss on it as part of the other kind of workshop, but there’s like a version of it where you can go at war against one of your opponents against an opponent to see who can more effectively outlive the other. But now we have the strangely addictive satisfaction of seeing an anh highway form. Um, here’s another spin on how to use um ant uh how to use like traces for mazes where this is a game where you can design a maze and then you can solve other people’s mazes. Um, but one of the things that is fun about it is it it shows you like the hardest mazes by um how long it took people to solve them.
16:45 Um, if I go into one of them um I this is actually helping the person who designed the maze understand how people navigate it. So imagine you’re a protocol designer and you want to understand how people are using that protocol. It shows a heat map of what are the most popular places in this maze and like where are people getting trapped in the most dead ends. So it kind of flips the trace from something that a navigator is using to something that the designer is using.
17:13 Um and I think I’m going to show uh just one more here um which is uh a maze. And this actually was inspired by a conversation with with Vancat on using mazes as art galleries. Um where I’m going to start exploring here. Um on the left we have a kind of bird’s eye view of a maze. On our right we have the first person view. Um and now I can navigate this maze and I’m going to encounter many of the same images that we saw during the course of this presentation. Um, so another it’s another protocol for really exploring a in this case an art space.
17:57 Um, yes. So that’s about it. So I think I’m going to check the chat to see if there’s any questions or I could always spend more time showing off more of these games in in detail.
18:13 So that’s it I have for now. Does anybody have any any questions?
18:21 Do you have a favorite maze that you’ve made? Um, I would say actually one of my favorite ones I’m going to pull it up that was not in this presentation.
18:42 Um, this was a maze I actually collaborated on with with Venitesh. Um, where he had a very he actually gave me a detailed set of requirements for a maze that would represent his clockless clock um idea. Um, so the challenge was to take um like a PRD for a maze and to represent it in this form. And you know honestly I won’t be able to remember all the details of it right now you can feel free to to jump in. But the way this maze works is that um you [clears throat] have a danger of getting these are different kind of stages of of life basically or different modes you can be in. You can get trapped in this outer loop like a zombie going around without real intentionality. Um but the goal of the maze is to um kind of find the hidden crevices in each area and to wind your way through and eventually make it to Nirvana. Um, but the reason why I like this maze so much is that um it just this amazing like very detailed complexity in terms of representing very subtle nuanced ideas and how do you express that as a maze?
19:46 Yeah, this is a really cool one. Um, I don’t see any questions, but personally I would like to see a little bit more explanation of kind of the games that have been made and how those work. And maybe somebody can put their hand up if they have a question. Okay. [clears throat] Um, I’m going to go into um I think I think I assume that most people here were not in the the working group that we did before on kind of an amaz. So, I’m going to spend some more time with that.
20:20 Yeah. I was wondering if there’s actually a solution to the 0ero to one [clears throat] startup map or if you just constantly bang your head into a wall until you decide to quit.
20:32 Um yeah, for the startup m for the startup one actually let me tell you a little bit more about that one.
20:43 This is one of my old maze websites. I have a lot of them here.
20:49 Um yeah, so this one was actually inspired by um like a I created um like a model. It sounds really cliche now called the product management triangle. It’s basically mocked very directly in Silicon Silicon Valley. You can join triangles of success. Um but the idea is that if you follow these stages kind of really naturally, you start at zero. you get to um a technology that users like.
21:13 Um then you figure out how to kind of create that [clears throat] technology um reliably. Then you figure out how to monetize it. Then you scale the user base. Then you scale the tech. Then you scale the business. If you kind of follow those um sort of steps, then you eventually get to the to the end.
21:39 Um, yes. I think the most the project that I’m find to be really fun right now is what we were doing for the uh for the workshop before. We’re actually Let me go back to the index page.
21:57 Um, we’re here. We can we have uh this is like a maze game, but you’re not an individual person in the game. um you have two colonies competing against each other and what you’re trying to do is outlive the other one. Um so it’s all about I’m going to make the simulation speed get higher. So it’s all about each side’s ability to control these parameters of the pherommones. Um, [clears throat] and we have these concepts in here where you can um like try to sabotage your opponent um by leaving sort of pheromone trails that’ll lead them back uh to your own nest and into the food source. Um or um if you want to like explore more, you can go into scout mode and have your ants uh less kind of commit to existing trails and do more exploration. Or if you get in an antmill, um you can go into volatile mode and they will like break free of the um uh of whatever kind of pattern they might be stuck in.
23:02 Um and actually um Vanites, if you’re willing, I think it might be kind of interesting for you to explain the concepts of the of the clockless clock maze more. Would you be willing to do that? Sure. If you want to bring that up.
23:19 Okay. So, this is based on a series of articles I wrote that are on ribbon farm. I’ll see if I can find the link and share later, but it’s it was an attempt to create a map of um every possible thought you could think about. And the way to read the map is the interior of the inner triangle is kind of like inner space where you have unstructured thoughts. So they’re meant to map what’s called the default mode network. So uh when your mind is wandering and so forth. Whereas each of the outer segments outside the triangle, they’re more structured kind of domains of thought uh which are usually called task positive network in the brain. So when you’re um focused in doing something uh the three vertices, the crow, the rat, and the goat. [clears throat] There’s a whole theory behind why I chose those motifs, but think of the rat as uh interpersonal subjective space um with um like you know, if you’re sick and feeling sorry for yourself laying in bed uh nursing your um body and mind, that’s the rat. The goat is meant to ex u u symbolize exploration on the frontier. So if you’re out there in the wilderness uh standing on a high mountain [clears throat] ridge or something and the crow is meant to symbolize um the civilizational core. So where you’re uh maybe I don’t know giving a big important speech in a big auditorium or something like that, right? And you turns out you can actually kind of represent everything you might do or think in your life along those vertices. And you can represent your life journey through them. And in between you have like transitions, right? So from the personal rat vortex to the goat vortex. Uh that side is kind of like uh all your builder maker type activities between the goat and crow.
25:21 You kind of have your public life. Uh so degrees you might earn, awards you might win, things like that. And the bottom edge is stuff like personal life, community, family, friends, things like that. So the maze uh topology reflects journeys through that and how you can have arrested development by getting stuck on one side or the other. How you can circle endlessly without ever like um you know having any kind of personal growth. So personal growth is spiraling inwards towards the void. So a very Buddhist kind of like interpretation here. So [clears throat] the solution to the maze is you start from the start point which I believe is in the public over there and if you solve it right you’ll avoid all the arrested development stuckness loops and eventually spiral inwards and uh achieve nirvana. So that’s kind of the concept of the maze. [clears throat]
26:12 Yeah. Thank you. And I also loved um go ahead.
26:17 cover what I what you wanted me to cover. Yeah.
26:20 Yes I did. And I think there’s also the element of the knock. Am I saying this right? Nakatomi spaces where you have the kind of main um corridors that like obvious but then you kind of have to slip your way out of them into kind of like the the pipes the inner kind of pipes of the building to kind of get through it as a way to break free from each of these zones. This one I think we can actually do better now with your new digital virtual mazes because actually the right way to represent busting through walls in Nakatomi space like in the movie Die Hard is for Bruce Willis to actually punch a hole in the wall and get into the duct right so you can kind of do that now in one of your new simulators where you can say if you expend enough energy you can break through a wall right uh but this is our attempt to represent this when was this 2019 or something so long before AI so in this case it’s like the very obvious and legible. Bold outlines represent kind of the official map and everything else is kind of the unofficial map of like duct work and sewers and things like that.
27:19 And it was the one time I ever participated in selling an NFT. So, thank you for that as well. Um, and actually have just realized one more thing since we have uh four more minutes left that I think would be um appropriate.
27:43 Here we go. So, here was like a very literal attempt that I made um at making a connection between protocols and mazes. Um and the idea here is that you can select these different attributes um like the topology, uh the visibility, the memory, like how hard it is um and then generate a maze like based on that criteria. Um and then it’ll kind of show you what what does that say about the protocol that you’re designing that fits that maze. Um so here we have one where meaningful junction is delegate judgment operators must infer global structure from local evidence um etc. Um, and then actually what bank was just talking about and why I thought of it is here’s like a a soft maze um where you can um literally go through the walls um versus actually no it’s down here where it’s participatory where somebody Oh crap, now I can’t remember. Oh, here it is. Um I actually have to refresh it where there’s a there’s a version of this here where you can actually like change the shape of the maze by your navigation through it. So you if you like bang your head against one of the walls, you can eventually get get rid of it, which is maybe like the die hard example. Um, but I think maybe just the final thing I’ll say is I like how with mazes, it’s almost like you can create the feeling of what it’s like to be inside of a protocol and create the emotional experience of living inside of a protocol, which is something that I’m excited to explore further. And yeah, thank you everybody for for listening.
29:17 Awesome. Uh, thanks Dan. I every time we have one of these like um, you know, symposiums or schools, there’s like a couple talks that are just like totally out of the blue in terms of topics and they’re always a crowd favorite and I feel like this is this is one of them. So, nice job.
Sean Stevenson argues that AI is undergoing a 'descent' through Stewart Brand's pace layers—moving from fast fashion/commerce layers down through governance and culture toward nature itself—becoming less dramatic and less contestable as it sinks, following a pattern he traces via Clayton Christensen's theory of attractive profits migrating to the least-optimized layer of a value chain. He proposes a four-stage model (integrate, coordinate, embed, inherit) for this descent and introduces 'appeal foreclosure' versus 'ossification' as a way to distinguish genuine new nature from captured systems, while drawing on Spinoza's natura naturans/naturata and Philippe Descola's ontological framework (naturalism, animism, totemism, analogism) to speculate how AI-plus-protocols might shift human categories of inner life and matter.
Nicolás Madoery presents 'divergent protocols' as a counter-model to standard protocols that converge toward homogenization, arguing that AI and robotics now function as infrastructure (not tools) through which culture becomes 'liquid' and infinitely re-versioned. Drawing on examples like AI-generated Argentine folklore, a dancing folk robot, and his PIACH framework for cultural organizations, he proposes protocols that preserve diversity by naming what is lost, keeping source names, and leaving room for noise rather than optimizing transmission.
Neilson Koerner-Safrata (introduced as Nelson Bershad, working at World Labs) argues that photorealistic image and video generation functions as 'a protocol for knowing'—that creating a photoreal image or 3D world from a single photo demonstrates a form of knowledge about reality, inverting the historical arrow from reality-to-realism into realism acting back on reality. He traces this through his own tools (image-to-3D-world generation with physics and sound, camera-roll 'telepathy' search) and historical precedents like Muybridge's chronophotography, Duchamp's response to it, Ed Catmull's hand animation, and AlexNet, concluding that the same technology deepening our computational grasp of reality is simultaneously destroying the shared trust in photorealism as a stable representation of the real.