The following is a rough transcript which has not been revised by The Jim Rutt Show or Frank Lantz. Please check with us before using any quotations from this transcript. Thank you.
Jim: Our guest today is Frank Lantz. Frank is a game designer with a focus on exploring emerging technology, create new kinds of gameplay. He’s the founding chair of the NYU Game Center and the co-founder of Area/Code Games, which was acquired by Zynga in 2011. And he’s the co-founder of Everybody House Games and the creator of the game, Universal Paperclips. He’s working on a book which will be published by MIT Press in October of ’23 called The Beauty of Games.
Frank: Thanks. It’s great to be here.
Jim: Yeah. It’s going to be a lot of fun to talk to you. I get a fair number of over-the-transom emails from people who want to be on the Jim Rutt Show and I will confess, a good percentage of them are nuts and what have you and another percentage of them are people I’m not all that interested in talking about their topic. But now and then, a good one comes through and when my producer forwarded Frank’s inquiry letter along, I said, “Hell yeah!” Turns out, Frank is one of the top players of my Network Wars game. And, of course, I quickly looked him up and I also find out he’s the dude at the Game Center of NYU. I go, “Hell yes.”
Frank: Yeah, I’m a big fan of the Jim Rutt Show and that’s how I found out about Network Wars and yeah, I genuinely love it. It’s a great game.
Jim: Yeah, so we are going to talk mostly about Network Wars today and we’re going to nerd out big time. I know a fair number of our listeners play Network Wars. They’re not listening to it. In fact, everybody turns the sound off pretty quick [inaudible 00:01:35].
Frank: So you can listen to podcasts.
Jim: [inaudible 00:01:38] So if anyone wants to play Network Wars, go to networkwars.com or search Network Wars, two words, on the Apple App Store, on Google Play. It’s a mobile game. Takes about 30 seconds to learn, at least a few weeks to get pretty good at. It’s a kind of hard game. A typical game only lasts four minutes or thereabouts and it only costs $0.99 with no in-game promotions. It’s a nice, clean game.
So tell me about your initial reaction. As a real game pro, I’m really interested what did you think when you downloaded it?
Frank: Okay, well, I’m going to be honest with you, Jim. When I first booted the game up and took a look at it, it looks too simple to be good. At first, you think, “Wow, this is too simple. Where’s the clever hook?” It’s a very straightforward game. It’s a little strategy game. It’s played, as people know. But if you haven’t played it yet, I’ll just quickly describe it. It’s a little turn-based strategy game that’s played on a procedurally generated network of connected nodes. And you are playing against four AI opponents and you are choosing a node and then attacking a neighboring node. And it’s just a simple kind of mathematical operation based on what the strength of the node is. You will kind of basically flip a coin for every point of strength and then, if you overpower the neighboring node, then you take it over. And then every turn, you’re generating more strength based on the size of your largest connected group.
And so I looked at it and I played around with it a little bit and I thought, “Well, there’s really nothing clever or surprising. There’s no hook.” But it turns out, you don’t always need a surprising hook or a clever kind of unexpected mechanic. If you take a solid idea and you execute it really cleanly, you can end up with something truly great. And I think that’s what you’ve got with Network Wars. It turns out to be a really fun game. It’s really interesting. It’s really enjoyable to go deep with it. I’ve been playing it a lot and yeah, it’s what I would call kind of minimalist game. And I think that it’s kind of surprising these days to come across something like this that really embraces the minimalism and goes deep with it and produces something great that way.
Jim: Well, you caught it exactly because as I was mentioning in the pre-game discussion, I’ve been a gamer since I was 10 years old when I started playing Avalon Hill tabletop games. And I went through the Atari phase and the Apple phase and the Commodore phase and the PC phase and some of the console phases. So I’ve always been into games. But I have found many games just it takes… they’re too complicated. They’re no longer elegant. There’s no elegance like checkers and chess are kind of elegant games. Go is a super elegant game. And so when I decided, I can’t even recollect quite why I decided, but I decided I was going to try my hand at creating an elegant game. And I sort of thought about the genres of games and I said, “I’m going to take the game of Risk and that family of Risk-type games and see if I can eliminate absolutely everything that’s not necessary.”
And I went through five or six design iterations and each one had less complexity, complicatedness, I should say, not the complexity. As a Santa Fe Institute complexatorian, I don’t want to use complexity wrong, but complicatedness until after a couple of weeks. I got down to a Windows prototype of what’s basically the current game and I said, “I don’t think I can make it any simpler.” I did eliminate one last thing, which was an option on how the reinforcements came in. There’s logically a couple of different ways you could have automated reinforcements. And I had two in the original prototype and then I said, “Nope, I’m just going to go with one.” And-
Frank: Just out of curiosity, what was the other one? Because currently, you just get reinforcements along your border.
Jim: And of course, the other option-
Frank: Randomly distributed evenly, but so what was the other alternative?
Jim: Well, it’s quite obvious. If you’re looking for minimalism, the other minimalist one is randomly across all the armies in the largest connected network.
Frank: Got it. So it would’ve been along all this stuff. Yeah. So again, this is an example of something that I think is one of the strengths of the game. I love the reinforcement rule and I think it’s a great example of your design process, this idea of just getting rid of everything you can and kind of boiling it down to its essence and it really worked. And again, it’s just unusual. It’s an unusual approach.
I would say that we live in an era where the dominant game design aesthetic is almost the opposite of that, right? It’s kind of… well, I think of it as Pokemon is a good touchstone for this because a lot of the beauty of Pokemon is in the intentionally complicated nature of the materials. Of all these different Pokemon, all these, each one has its own set of stats and affinities and strengths and weaknesses. And a lot of the pleasure of Pokemon is kind of memorizing this vast quantity of data. League of Legends is extremely good game, extremely popular, contemporary game, very similar in its aesthetic approach. If you sat down to try to learn League of Legends, you almost can’t. It’s not the kind of game you can sit down and learn. It’s the kind of game you have to just imbibe as part of your generational experience. You just got to grow up.
But there is this other approach of thinking, “What is the simplest possible expression of an idea or a mechanic?” And in some ways, it really is… it goes to the heart of what games are about because this idea that you can find a surprising amount of complex and interesting behavior in a very simple set of rules and materials, that’s really the magic of games.
Jim: Yeah, that was indeed my intent. It quite literally was my intent. And from that point, I had to then tune the game because as a gamer yourself, you know it’s an iterative process to get a good game. And so two other things I played with were… the first one was play balance, basically. And one of the other things that annoyed me in the mobile game world is most of them were way too easy. The quick play, easy play phone games five years ago, you could generally play them 20 times and master them more or less or figure out that they were too ridiculous to ever master. And so I chose to make two Network Wars to be pretty hard, but you could see yourself growing in knowledge and you’ll never be able to win 100% of the time. At least, nobody ever has. And I would predict that nobody ever will win every game in 1,000 plays.
Frank: Oh yeah. I mean we can talk about that, but I’m pretty sure that the maximum win rate is close to like 76, maybe 77 or 78 possibly. I don’t think it’s anywhere near 100.
Jim: Yeah, and we’ll talk about the stats a little bit, but you’re pretty close. And so interesting, the things that I tuned, the size and aspect ratio of the network turns out to make a significant difference with this particular set of gameplay. And I came up with six by seven. And then the other one is the ratio of nodes to links, turns out, or put it this way, the algorithm that generates the links and the nodes with essentially an affinity and probability of establishing a link on a node, et cetera. So called the density of the network, approximately.
And you tune both of those things and they change actually the win loss rate. Because I wrote some AIs that could play in simulation, so I’d run 10,000 games and see what the average win rate of an AI playing the human was. And I’d tweak things and run another batch job until I got to… I targeted 80%, as I said, that a genius player should be able to win 80%. But it turns out, you’re approximately right in the data that such a paragon has not yet appeared, though one guy did for a short period of time. Let me see if I can find my data here. No, actually, I don’t have that handy. But there are people who have won over 80% for their whole career, but their careers have been short, like 130 games.
Frank: Yeah. I mean, look, I can win 80% over any arbitrarily long run of games. But when you’re talking about the true long run, yeah, I think we’re getting close. We’re asymptotically approaching the maximum possible [inaudible 00:11:21].
Jim: Yeah, so I’ll give you your score here, right? 74.9% win-loss average over 79, 100, and 48 games and that makes you 14th on the all time hit parade of all players. But most of those are people who’ve played a few hundred times. In the category of 1,000 plays or more, you’re up near the top. There’s two other people that are above both you and I who have played more than 5,000 times and one of them, mysterious. I just have a user number for him, Mr. X. He’s played 11,864 times. It’s a lot. He has a win-loss record of 75.6. He’s actually, I would say, he’s the best of the many plays.
And here’s the other two things that are quite impressive about him beats both you and I. One of the statistics that I just actually ran for the first time yesterday. I’ve been saying I’m going to do this forever, but I finally got around to doing it yesterday is the longest maximum win run because again, this is a short game, takes four to five minutes to play. And you get in the groove and you win 20 in a row, right? You feel great. And truthfully, I was never quite sure what my maximum win run rate was, but I now know. But anyway, Mr. X, his maximum win run is 35, which is really amazing.
Frank: Wow. That’s pretty good.
Jim: And also, the best he ever did in 100 rolling game… one of the figures of merit that I use when I play. I don’t know about you, but I’m always looking at how I’m doing in the last 100 games because that seems to be a statistic that’s stable enough to tell me where I’m at. And also, when my focus starts to degrade, you can see it starting to go down. And his best 100, might be the highest anybody ever did, is 92.
Frank: Wow. That’s pretty good. Yeah.
Jim: That’s 92 now, Frank.
Frank: Yeah, that is impressive.
Jim: Yeah. So Mr. X is probably a bit better than you and you’re a bit better than I am. Your longest win run is 33, which is pretty damn impressive.
Frank: That’s not bad.
Jim: So you’ve won 33 games in a row at least once. And your max 100 was 91, so only one less than Mr. X.
Frank: Okay. All right.
Jim: That’s pretty good. And that ranked you 14th, the win-loss rate 14th of all players, but that includes a bunch with relatively small. All players [inaudible 00:13:49].
Frank: We all agree we’re going to throw those out.
Jim: So of the Mr Bigs, you’re number two or number three. There’s another guy who we could… statistics aren’t quite canonical, see. But anyway, you’re number two or number three.
Frank: So 75.6. Okay, that’s what I’ve got to beat if I want to get to the number one spot.
Jim: You want to be Mr. X and then myself, as the inventor who knows how it works, just slightly worse than Frank. That’s interesting. And I’ve played 9,021 times versus Frank’s 7,948 times, so that’s fairly similar. I ranked 18th on the total depth chart. My win rate is 74.1 versus Frank, 74.9. However, my max win run is only 25. I was kind of surprised and somewhat disappointed while Frank’s was 33. And my max 100 was 85, exactly where I predicted it would be because I had a sense that I get the 84 and I try to get it up and it usually doesn’t go. And I never remembered ever being higher than 84, but I said “I’ll bet there was a time I wasn’t paying attention when I made 85.” And it turns out, I did. But yours at 91 is quite a bit more impressive. So your claim to be the top Network Wars player in the world is damn close. You’re in the top three, for sure, and a bit better than me.
Frank: Well, this is scary. I mean, being number two is worse because now I have a reason to keep going until I beat Mr. X. But I mean, a lot… it’s really interesting to look at these stats and try to figure out which of them are meaningful. I think most games of Network Wars are kind of decided in the first setup. You look at a board and in many cases, there is no path to victory. There’s just certain starting conditions where you just… there’s no expert play is going to get you to win those games. Those are just losses and everyone-
Jim: It’s not quite true. It’s not quite true.
Frank: … is that not true? Come on, that’s got to be true.
Jim: Well, I have yet to find a game that if I play it 20 times, I can’t beat it. Because now, one of the things that annoys all my players is I don’t allow replays of games, right?
Jim: And I have a theory on why I do that. Also, something else, which not too many people know, a few people do, is every player plays the same sequence of games. Because I originally did this as a cognitive science experiment to see how humans learn. And so everybody plays the same sequence. So when someone gets to 7,948, they’ll have played the exact same games you had, so-
Frank: That’s so good. That is so good. I’m really glad you did it that way.
Jim: … here’s something that’s even scarier in terms of make it comparable. The dice rolls that happen under the skin all always use the same random number generator for each game. So if there is a bias in the dice rolls, a short run bias in the dice rolls at the game level, every player gets the same roles of the dice. But because they play the game differently, the impact of those same dice rolls are different. It’s the order of the dice rolls are the same.
Jim: So again, it increases the comparability across players, so we can actually look at a given game and see how good you are at a given game and know that everything else was equal. [inaudible 00:17:17].
Frank: Sure, like they do in bridge, like a duplicate.
Jim: Exactly. Duplicate. Very similar to duplicate bridge. So let me then go on to another design decision. I’d love to give your feedback as Mr. Game. The combat results are kind of like Risk, but they’re not the same as Risk. In large n, lots of trials, a bigger stack will beat a smaller stack on average over 1,000 tries or 100 tries. But unlike Risk, there’s a lot more variance in the outcome. And this drives people crazy, including myself. I’ve seen an 11 lose to a two, which would never happen in Risk, but in my game, 11 happened to two is unusual, but not out of the ordinary. So think about the statistical distribution I used as being more fat-tailed than your typical dice roll-type outcome. And some people think that’s a great feature and some people, it drives nuts, the very high stochasticity in the combat results.
Frank: It works for me because that is a combination that I like a lot. I’m a poker player, Jim. I don’t know if you play a lot of poker, but-
Jim: I used to.
Frank: … okay, so that combination of high skill and high variance is like catnip for me. I just love it. There’s something that just tickles my fancy about something where you’re rewarded for doing a very deep model and for doing a very deep analysis and for calculating. And yet, at the same time, you’re in the wildness of super high variance and you just have to embrace it and you, in all of your models and all of your calculations, have to include it and account for it. To me, that is a great place to be. My brain just loves being in that space. So I’m a big fan.
Now, is it accurate to… the way I model it in my head is that I see the two nodes. So the node that’s attacking and the node that is defending, I see them as almost like a stack of units where each unit, they flip a coin against each other and as soon as one wins, the other unit is eliminated. And then you just keep doing that down the whole stack until there’s zero left in one of the stacks or zero left in the defender’s stack and then you move the last unit or you move all of the units, leaving one behind.
Jim: If there is one.
Frank: Is that accurate? Is that the right way to model it?
Jim: Very close. It’s not quite a fair coin flip, but it’s very, very, very close because that was another little twisty parameter.
Frank: What type of coin flip is it?
Jim: Slightly weighted. It’s very slight.
Frank: Towards the attacker.
Jim: Towards the attacker by a tiny bit. It’s a-
Frank: And is there any difference in the roles between the human and the bots?
Jim: No, no, no. The one thing that I will say-
Frank: It seems like there are, but I knew there probably wasn’t because I know how my brain works, but I’m like… because you noticed the ones that are super unfair against you. You don’t notice the ones that are super unfair in your favor.
Jim: … I’ll confess. I actually went back and looked at the code because I was convinced that I must have implemented it to be more stochastic for the human than for the AIs. And nope, it uses the exact same function call.
Frank: Good. I’m happy that you did that and I think that was the right decision. It’s funny, in game design, there is a tendency to bend randomness in favor of the player and in favor of the player’s instincts. Even someone like Sid Meyer, who is obviously a genius game designer, the creator of the Civilization series-
Jim: But the really cool Gettysburg game, which is also really good.
Frank: Brilliant designer and someone who, yeah, very influential. He’s the person who said “a game is a series of interesting choices,” right? So he’s very famous and influential. But when he talks about randomness, he often… yeah, he says, “Look, I mean you want to make it feel right to the player.” True randomness doesn’t feel right because what the player has in their mental model is a kind of randomness that’s more smooth. It’s more evenly distributed. It’s not as spiky as real randomness.
And so we make it so that after you’ve had a couple of inconvenient roles, we just make sure that you get the role that you’re looking for. And it drives me crazy because I think that part of the beauty of games is that they force you to wrap your head around the reality of a surprising truth, that something… that the world doesn’t work the way you think it does and so having to update your models and realize that that randomness is spiky and has these character… that’s what happens in poker. Poker doesn’t accommodate for you and your preconceptions and your biases. It just is what it is and that is a part of its spiky beauty.
Jim: Now, it’s interesting. Now, let’s tie it back to your comment about unwinnable scenarios and my assertion that as far as I… I mean, I will say I haven’t tried every scenario. There might be one that are just totally unbeatable, but I do have a backdoor for myself that allows me to replace scenarios. And I’ve sometimes had to play it 20, 25 times. But the way you win is not by having a better strategy. The way you win is by being lucky, right? And so from my poker days and this is actually quoted in a book by Michael Mauboussin, who’s a fairly famous Wall Street guy. He happened to hear me give a talk once on it, which is as a systematic strategy for life. Learn from poker is when you’re behind, raise the variance, right?
Jim: If you get one of those scenarios that suck, if you play straightforward, the equivalent of being a rock in Texas Hold ‘Em, you will lose, right? The only way you’re going to win is get wild and wild plus lucky. But if you’re not wild, if you play conservatively or straight, exactly balance the amount of luck that you’ll need is so much that you’re not likely to get it. But if you do crazy shit and you get lucky two or three times, then you change the situation and…
Jim: … times. Then you change the situation and change the whole scenario. So even the one in particular, it’s one of the early scenarios that everybody complains about and says is unwinnable, I actually have beaten it by that approach, by taking a really radical … A strategic approach and getting lucky.
Frank: Okay. So you were saying that theoretically, it’s possible to win any starting setup because of this wide variance. So because of this wide variance, when you’re in a setup that looks unwinnable, you can just take some crazy swings and eventually … And then since you can restart it over and over again, eventually you’ll flip 10 heads in a row and you’ll be beat it that way.
Frank: But that’s different. That’s different from saying each of these … Okay, I see what you’re saying. So technically in that sense, they’re winnable and in some universe, some lucky person might end up at the far end of that distribution and end up winning a bunch like that. But in the universe that we live in, that’s not going to happen.
Jim: Well, interestingly, this famously difficult early scenario, a couple of people have won it. Just newbs who were just playing for the first time and they just have … Out of 25,000 won. Two or three have beaten it. And when I play with one of my low rent AIs that plays the human, it’s not very smart. If I play I think it’s 100,000 times, it’ll beat it once or twice. So again, it’s winnable, but it’s just very, very, very, very difficult.
Frank: There’s another way you could think about this, which is because you are using the same seed to generate all of the actions in the game, if you knew that, you would have a complete picture of exactly what’s going to happen as the result of every encounter. And in that scenario, you would be back in a universe where there are some setups that you can’t win. You can win them if you’re allowed to skip ahead until you get that extremely lucky stretch of outcomes and then use that in the right way to win the scenario. But given the fact that we all have to work our way through that same predetermined set of dice rolls, you’re going to go through stretches where you can’t win and not-
Jim: Actually, not quite true. Actually think about it hard, because it just makes it more difficult, because don’t playing blinds man’s bluff or find the needle in the haystack, but the needle is still there. It’s just You have no principled way to find the needle, but it’s still there. So it’s just a very-
Frank: So even if you exposed that, you say, “Look, I’m going to show you all of the roles in the [inaudible 00:27:16]”-
Jim: It’d be a hell of a lot easier if you could see that. But I do not let the AI see that, and I don’t let the humans see that.
Frank: Right. But in that scenario, you think that I could use that information to find a winning path?
Jim: Oh, make it a lot easier. Yeah, think about it from a calculation complexity.
Frank: Yeah. It’s certainly advantageous. Good for you to have that. I just think in some cases you’re still going to look and say, “Okay, there is no correct path to winning from this startup.”
Jim: And I would say that is possible. I like that idea, and I think I may just get psyched to try that and see if I can write an agent that does use that information and see if it can boost its win rate to a ridiculous degree, even on the worst scenarios.
Frank: And I’ll pre-register my prediction, which is that there won’t be such an agent. And we can have a little friendly bet on it, but I think that’s [inaudible 00:28:08]-
Jim: All right, let’s bet a bottle of a whiskey on that. In the first a hundred thousand games in the sequence, your proposition is there’s at least one that can’t be beat by an automaton that has look ahead on all the dice rolls.
Frank: I love that. Yeah. I love my side of this bet, and I can’t wait to see what whiskey you’re going to get for me. Yeah.
Jim: This is one that … I’ll get you something good. I hope you’re a single malt man. A single malt or bourbon. Your choice.
Frank: Oh, either one is fine. I trust you.
Jim: Okay. All right. Well, now let’s … We talked a fair bit about some of the design decisions that went into it. And you actually have sniffed them out. The fact that it’s a coin flip rather than a dice roll, very few people have figured that out. But if you play it enough and if you have the right kind of mind, it does become obvious/
Frank: Yeah, and it’s funny. It’s just like if you want to get good at a game, you end up discovering what is happening under the hood. And that’s part of the pleasure I think, of trying to get good at a game, is that it just rewards you for thinking and having a accurate vision for what you’re interacting with.
Jim: Yeah. You were correct, actually. It’s an almost perfectly fair coin, but not quite. Just tiny. Just cause again, ’cause I wanted to tune that, my guesstimate of where the maximum win loss of 80%. Now, next thing: After I wrote this, and I think maybe it was just after I had it out on test flight, was a few users of the mobile version. I think I had 50 users of the mobile version after I had 13 users of the Windows version when I was just prototyping it. I played it a bunch, an embarrassingly large amount, and-
Frank: That’s fine.
Jim: Fortunately, it does not show up in my current number of 9,021. It was probably another three or 4,000. And one of the things I did during that process is I wrote down 50 heuristics that I discovered. And in fact, part of the reason I did this was originally was to start to understand the idea of heuristic induction from a cognitive science perspective. Something I’ve been saying for a long time in the other work I do in the artificial general intelligence area, that I believe we will find that at least the low ends of artificial general intelligence are going to be in part driven by the ability to induce heuristics, rules of thumb.
The world is very, very complicated. It’s actually complex. And then the land of complexity, it turns out you can’t solve very much in closed mathematical form and you even can’t get reliable and stable results from quantitative modeling or even agent-based simulations in high complexity. But rules of thumb is how we get through life. See a lion, run. Things of that sort. And so as I played the game, I wrote down … I still have the papers someplace, I couldn’t find it unfortunately. I think it’s at my other office. I wanted to review it today. But it was really fun to record these heuristics and they were written down in a way that if I had everybody’s complete gameplay, I could actually see what heuristics they’re using, or at least on a probabilistic basis. Did you have the sense as you were learning the game, that you were gradually uncovering a stack of heuristics?
Frank: Yeah, I totally did. And I think that one of the marks of a great strategy game is exactly this feeling of climbing the ladder of heuristics. There’s a great book about games by Richard Garfield and Skaff Elias. Richard Garfield, the creator of Magic: The Gathering, wrote a book called Characteristics of Games, and he uses this phrase, the latter of heuristics. And I think one of the deep pleasures of trying to get good at a game is this feeling of incorporating these rules of thumb into your thinking, almost like cognitive tools. At first you encounter them, they’re explicit. At first, you think of them as like, “Okay, this is a little rule I need to follow and it’s going to help me win.” And then eventually, it almost becomes second nature. You don’t even think about that because now you’re at a higher level and you’re encountering a new rule of thumb, which again, at first you make explicit.
You’re thinking, “Okay, I should always do this in this situation. Oh, here’s one of the situations. I recognize it. Now let me apply this little rule.” And you’re doing it very consciously and explicitly. But then eventually, again, it gets incorporated into your thinking and it becomes second nature. And now you’re at an even higher level and you’re developing a new heuristic at a more strategic, more abstract layer. And you just keep doing this and you keep building this deeper and deeper understanding into your model of the game. And I think it’s part of what I love so much about games, because it’s a way of seeing your brain operate. Normally, you live inside your brain and you can’t see it. It’s like the fish for whom water is invisible. What’s water? It’s like, “What’s a brain? What’s a mind? What is problem solving?”
Because we exist inside of problem solving, and it’s invisible to us because in a sense, we were comprised by problem solving. And yet when you play a game, it’s like you’re able to see your own mind at work and see how it encounters the world, see how it bumps up against a problem and starts to analyze it and then develops a rule of thumb to deal with it and then incorporates that into its natural, organic, low level processing. And then re-frames the problem at a higher level and keeps going. And games are great at that. They’re little laboratories where we get to take our brains out and examine them and see them and seem them operate. And I think a truly good game really has a rich ladder of these things.
Whereas if you think about any problem you encounter, it can fail to be a good game by either being trivial and right away, you see an optimal solution, and then that solution just works all the time. Or it can be intractable in the sense that there really is no good rule of thumb for improving your results except to solve the whole thing. And so those are the two ways that problems can fail to be interesting. But a really interesting problem is one that is semi-tractable, in the sense that you can get better results by applying these rules of thumb. And you can do that in a continuous manner. So you’re not just encountering this clip where you have to solve the whole thing. You’re actually able to get better and better by taking these … What I think of, I think of a heuristic as almost like a shortcut through search space.
My basic meta rule of thumb is if you can search, you should. So if you’re in the end game of chess, there’s no point in applying a heuristic. If you know there’s a mate somewhere on the board, just search until you see the mate. You should not be doing a thing where it’s like, “Oh, I’m going to control the center”, or “I’m going to develop my pieces”, or, “I’m going to have a good exchange.” No, just find the mate. If there’s mate, just find it. So if you can search, you should search. But most of the time, you can’t search because you don’t have an infinite amount of computational resources to throw at a problem. So when you can’t search, that’s when you are applying heuristics. And what heuristics are doing is they’re really showing you these paths that are almost shortcuts through search space. They’re ways of … I don’t know, they’re like tunnels. They’re like hyper-dimensional tunnels through search space.
Jim: If we want to talk about it in complexity science terms, I like to think of them as dimensional reductions.
Frank: Yeah, that’s exactly right.
Jim: If the real space is really high dimensional, we can’t actually process formally a 20,000 dimensional space, but we can do some thinking in a four dimensional space. And so one way of thinking about heuristics at least, our set of settings in a low dimensional space, which are tractable to human reason. And humans aren’t that smart. There you go. We can be beat by a $1.95 calculator when it comes to adding numbers up. And as it turns out, our phone can beat us at chess, and our large language models will soon be able to write better than anybody, less than the very, very best of us. And so we have to these heuristics, because we’re not that smart. We can only operate in a certain level of complicatedness. And inducing those … But that’s also going to be true of computers because the world is just amazingly complicated and complex. The ability to somehow find these heuristics is still a human superpower that-
Frank: Yeah, that’s exactly right. Even in a domain where it is completely deterministic. And the problem … In a sense, games exist to create problems that are the opposite of wicked problems. Because in a game, the problem is well-defined, incredibly simple, it’s clearly stated, and we’re going to reduce as much of the fuzzy, complicated noisiness of the world. We’re going to isolate a very simple problem and create a toy universe in which this problem is clearly stated, and the outcome of success is clearly defined. The criteria is right in front of you. And even then, you find yourself in a thing that is almost infinitely rich and complex and surprising, where we still have to use all of our creativity. We still have to improvise and be creative and come up with ideas. And that’s one of the wonderful things about games, is they show how deeply rich and complex the world is, even when you isolate a single corner of it.
Jim: And as you pointed out, your initial reaction to Network Wars is, “Damn, this is really, really simple.” Were you surprised at how much emergent complexity there turned out to be?
Frank: I expected there to be more there than it looked like at the … Because I knew you and I knew the kinds of topics you were interested in. I knew your background, and I just trusted that there was going to be some meat on the bone. But I will also say that even I who have a taste for this stuff, I have a taste for simple games, minimalist games, even I was a little bit surprised. And I think it’s partly because in game design, we tend to have a reliance on cleverness. That part of when you encounter a new game, even if you’re a game design snob that has a very developed literacy about games and genres and stuff like that, you’re often looking for the novelty in the mechanics. Do you know Reiner Knizia? The board game designer?
Frank: Yes, he’s a famous board game designer. He’s one of the people who ushered in this new era of Euro style board games that-
Jim: Like Settlers of [inaudible 00:40:31] or something?
Frank: Yeah, exactly. And Euro games can be very simple, but they often have this interesting twist where a mechanic is unusual. Reiner Knizia is especially good at this kind of thing, where you’ll say, “Oh, okay, well your score is the result of the difference between your two closest neighbors”, or something like that. Some little strange, surprising hook. And so I’m often looking for that, “Okay, what’s going on here that is going to introduce this novel surprising thing?” But instead, I think in your case, I think in the case of Network Wars, it’s the graph itself. It’s the play field itself. What you have done, you took this tradition of connection games. There’s this existing tradition, go is in this tradition. There are also games like Nash or Hex y, Mudcrack Y, all of these games. Twixt is an example. Games that involve understanding the network topology of a graph and building a game out of that.
And I would say, yeah, Network Wars is definitely. In this tradition, there are computer versions of this. Slay is an old PC game that is … I think there are mobile versions of that too, where again, it’s a minimalist version of a traditional strategy game on a map with interesting connections, like a Risk like map. But I think what you did by really going deep on the graph itself, on the play field being procedure generated, and so the connections are always different. The topology is always different. I think that is really the heart of the game. But I will say that for me, there’s one thing in particular that makes Network Wars, was the biggest and most surprising insight that I got out of this game. And it has to do with the AI. So the fact that you had this minimalist approach and you reduced everything, means that the opponents in the game are also incredibly simple. You made these very simple models. They don’t really take a strategic approach.
Instead, they have very simple deterministic behavior. And because you did it that way, as the human player, that ends up being, in some ways, the heart of the game. The heart of the game is this bot diplomacy where you are manipulating the opponents against each other to your own advantage. Because you are able to model them very accurately because you know exactly basically the algorithm that determines their behavior. And as a result, you can often be vastly outnumbered. But because you can model them and they can’t model you, you’re able to direct their behavior. And so you’re like this tiny … I’ve had games where I’m on a single node and know these two vast armies surrounding me. But because I know that they’re going to fight each other, I’m able to sneak in and then eventually at the right moment, take over the entire map. There’s nothing more fun. It’s the absolute peak fun of Network Wars, is to do this.
Jim: Yeah, I had one of those about a month ago, and I sent it to some of my Network War friends. I had one node. I will say the node had 11 strength points on it, and I had an intuition that this was winnable. And so I started taking screenshots. I had a series of five screenshots and sent them to my Network Wars buddies. And indeed, I did win exactly the way you did. I saw that there was probably going to evolve a situation where I could break out, make the guys fight each other, and then isolate one of them, kill him, and then come back and kill the other guy. And indeed, it worked out.
Frank: So realizing this really led me to an epiphany. The epiphany that I had was that the bots in Network Wars aren’t opponents at all. They’re really just rules. They are in the game the same way that gravity is in Tetris. You don’t think of gravity as an opponent. You don’t think of gravity as a player that you’re playing against in Tetris. It’s just a rule of the game. It’s just one of the material facts about the game, is that pieces behave in this way because gravity causes them to go down. Well, the opponents in Network Wars are just like that. They are part of the game. And this was like, “Wow, this is a new way for me.” In some ways, this is true of all games that have deterministic AIs in them. But because of the way that you implemented these AIs so simply, in a way that made them transparent to the player, this was the first time I’d really thought of this fact, that in a sense, any deterministic AI can be thought of not as an opponent.
It’s not that you and the AIs are both playing this game. It’s that really, you are playing the game and the AIs are inside the game. They’re just pieces in the game. They’re just rules of the game. And then it led me to another thought, which was, “Well, this is true of any AI. Any AI is like this in a sense.” And then ultimately, it led me to realize, “Well, in a way I’m like this.” Do you know what I mean? My behavior in the game is produced by something. It is the result of something. You have a strategy that determines your moves. Now you can’t see that strategy clearly. But in a sense, someone could. Someone could look at you and see that you yourself are also like this. In a sense, there is … At first, once you realize this about the Ais in Network Wars, it feels great because it feels like you’re a sphere visiting flatland.
You can see along a dimension that they can’t see because they’re embedded inside of it. You can see them from above and they can’t see you. And it’s just this amazing feeling. But then you realize, “Well, I’m like that myself.” I have this advantage over the AIs and Network Wars. But in a sense, I’m also embedded in whatever deterministic space that I’m in that is causing me … Whatever strategy I’m applying to determine my actions within the game. And in a sense, all there is really is the game. There are neither opponents nor players. There is just the game. That’s-
Jim: I like that. That’s an interesting-
Frank: It was a lovely epiphany. It really changed my perspective, not just on your game, but on games in general. And it blew my mind a little bit.
Jim: Interesting. Now, I did made that decision because I actually have four or five other AIs that I tested, and this one is definitely the stupidest, of the ones I’ve used.
Jim: And this one is definitely the stupidest, right? Of the ones I’ve used. But I had the epiphany of what I was trying to accomplish. Of course, after the fact right? Because often some of our best insights are after the fact, we just stumble into. And I realized what I’d created, I’d created the equivalent of swarming behavior in complexity science, where it turns out you can model fish schools and bird swarms with remarkably simple algorithms. I mean, ridiculously simple, where all the elements are running the same algorithm and they produce what looks like highly intelligent, coordinated behavior, but they’re running equivalent of 10 lines of code, right?
Jim: And one of the things people are always, offense, right? Often takes people a long time to realize how dumb the agents are because they do things that seem smart. They fill up their end of the space and then they build a wall and then they build up, and then they break through. Kind of the same things that humans do, but they do it through a swarm-like behavior rather than a deeply, there’s no analysis at all. They’re just amazingly shallow. But nonetheless, the behavior looks smarter than it actually is in the same way that bird flocks and schools of fish are smarter than they are. And the wars between the bots also just seem a little smarter than you’d think once you realize what the agents are actually like.
Frank: Yeah. One of the things that happens often is that you’ll look at a board and you’ll be like, “Okay, well I can tell what’s going to happen next turn, because this guy’s going to go here and this guy’s going to go there and I’m safe because he’ll go over here.” And then the bots manage to do a thing where they get in their own way, you know what I mean? Where they cut themselves off, and as a result, they end up attacking you in a way you didn’t expect because.
Jim: Well, that-
Frank: They cut off their own best move and it’s just so aggravating.
Jim: The famous one that happens to me, well, yeah, the most likely thing they’re going to do this, it’ll leave me safe because there’s twos next to my twos, therefore they can attack me. But one of them ends up attacking the two and turning it into a zero, and then somebody else attacks the zero and then attacks me, and I go, “God, damn it.” Or occasionally a bot will just get this incredible run, we call it the Prussian style, where they’re just unstoppable. They just plow through far more intermediate little ones than you would ever think that was possible. And then they’d hammer my critical but weak position and then I’m screwed, right?
Frank: Yeah, I hate when that happens.
Jim: Well, that happens. It does.
Frank: When you talk about heuristics, I think one of the first heuristics that someone develops in Network WARS is the idea that you sometimes want to skip your first turn.
Frank: Knowing whether to skip your first turn, I would say is almost like the first step of getting good at Network WARS.
Jim: And actually it’s the reversal of the anti-pattern that most people start with, which they start with trying to expand as fast as they can on the first turn, right?
Jim: And by the way, there are times that is correct, but not very often. And in fact, in the next, if I put out a next version, one of the things I’m going to capture statistically is how often do people not move at all on the first turn? And my guess, the good players, it’s 40 or 50% of the time.
Frank: Yeah, I’d say that’s about right.
Jim: On the other hand, every once in a while you see a first move which is just devastating, and you take it and you do all kinds of crazy shit. And sometimes you just move one piece out of the way to let two guys get into, I’ll hold your coat, let these two guys fight, right?
Jim: And see those things. But probably 40, 50% of the time the best play’s to do nothing. But knowing when to do nothing and when to do something is what moves you from being a low sixties percent player up to a high sixties percent player, I suspect.
Frank: Yeah. And another thing that happens as you get better at the game is starting to think about it almost in thermodynamic terms. You’re looking at pressure building up and you’re like, “Ooh, I really need to puncture this to let this fluid out of here so it can flow over here.” And you’re really looking at these things in terms of flows. It’s all this chunky mathematical stuff, but it really does feel like, “Ooh, I need to deflate this thing over here so that this thing can expand into it,” and thinking holistically like that.
Jim: Yeah. Let me think of another kind of chemistry analogy that I sometimes use, which is fluid dynamics in viscosity. Sometimes, the opponents have lots of big chunky pieces behind the lines and go, that’s going to be hard to break through, shit. But other times, even though they have lots of nodes, they’re really thin on the ground, so their viscosity is low and you can puncture through and other times you can’t. And then the meta-meta-metagame is up in the 40th of the 50th heuristics is say, how is it likely the situation will change over the next two or three turns with respect to this viscosity? Will the density go down because the guys are fighting each other? Or will it go up because these guys have reinforced positions where the reinforcements will make them stronger and stronger.
And that’s to my mind, almost like the zen, you’re now getting really high up near the top of the stack. Some of us talk about the need to be both mentally tight and loose at the same time, very much like in poker. You have to know when to wait, and you have to know when to go. And early on, there are people who are biased towards going too often, but then they’re often in their intermediate trajectory of learning, seem to wait too long. And so having the ability to analyze the situation, know when to wait, and when to go.
Jim: Does that resonate with you and your game plan?
Frank: Yeah, it totally does. Every turn of Network WARS involves modeling what you think is going to happen on the next turn and looking ahead a few turns to predict what the likely outcomes are, and then thinking, okay, what can I do to improve my position given this prediction of what’s going to happen? And then knowing how much risk I need to take. So going back to this idea that you had of knowing where you are, how safe you are, right? Yeah.
Jim: Don’t lose a winning position, but if you have a losing position, raise the variance, right?
Frank: Exactly. So if you think, oh, things are not looking great, then that’s when you need to start taking some risks. I need to get lucky. I need to get myself an opportunity to get lucky, right?
Frank: So then you’re finding situations where you’re maximizing your chances of getting lucky in both directions. Both getting lucky as an aggressor, and then maybe getting lucky on the following turn when the bots are moving, getting lucky by having them be unlucky. And knowing when to put yourself in those situations, I think is the real edge that you get in as a long-term player.
Jim: Yeah. Now, you’re getting up into probably the upper forties of the heuristics that once you start to see those things, then your gameplay moves up another couple of points.
Frank: Yeah, I think that the earliest heuristics are understanding this idea of when you want to skip a turn, also understanding the strength of position versus raw power. You want to be in a corner, you want to be on an edge, you don’t want to be in between two opponents, and being in a corner… And there’s also this idea of surface area. How much of your territory is bordering other people’s territory? Because the smaller that ratio, the more power you’re going to have in the nodes that you’re building up reinforcements in. And so understanding the strength of that and knowing when to trade off size of territory for position, I think is one of the earlier heuristics that’s really important.
Jim: Absolutely. And some of that, as I was learning the game, I adapted from Risk. When I play Risk, I tend to be a grab Australia or South America early kind of guy because they have the least surface area. And then you can build up your power at the borders more rapidly than the opponents who have [inaudible 00:56:45]. The worst one I think is Europe maybe, or Asia, has just too many entry points. Even though they’re bigger and stronger, the ratio of area inside to area outside turns out to be important. And it’s often, I find it an interesting play on seemingly difficult scenarios when you fight your way through, get some luck on your side, and grab a peninsula that you had no business grabbing, for instance, right?
Frank: So satisfying. One of the things I think about sometimes is how the game would change if it were played between five human opponents.
Frank: And I think what you would find is that in some ways I think it would be interesting to do. I think it would be a totally different game. I think ultimately the game would become much more a game about king making. It would be a game with lots of politics. It would be a game in which you’re always incentivized to not attack, right? Because when you attack, you’re going to make yourself weaker and the person you attack weaker. And so you’re really making three of your opponents stronger. So everyone’s disincentivized to attack. And you’re going to have these strange political and psychological and social norms that emerge over the course of many games, right? There’s a wonderful game, another mobile game called Galcon, which I played, which is.
Jim: How do you spell that?
Frank: G-A-L-C-O-N. And Galcon is similar in some ways. It also comes from this heritage of connection games. It is real time instead of turn based. And you have these planets and you’re sending ships back and forth, but it seemed like minimalist abstract strategy game with reinforcements. But in Galcon, it’s human opponents. And again, it quickly becomes this very elaborate game of diplomatic negotiations. It is really a king making game. And the only reason it works is because it has short game sessions. I’m a huge fan of short sessions. I think that’s one of the reasons poker works. Poker would not work if every hand took half an hour.
Frank: Poker works because despite the fact that you can just get unlucky, every hand takes a couple of minutes, and so you can play thousands of them. And over the course of many hands, the overall shape of the game emerges and that’s really beautiful. And a game like Galcon is the same thing. It would not work if you had to invest half an hour or 45 minutes in a game, and then the results were just determined by someone decided to attack someone else and that was what determined the outcome. But because the games are extremely short and they’re iterated, it’s like the prisoner’s dilemma. The prisoner’s dilemma as a one-off, very, very different from the iterated prisoner’s dilemma.
Jim: Yeah, exactly. We know what the answer is for one-off. It turns out there is no staple strategy for iterated.
Frank: Yeah. And that’s the beauty of the prisoner’s dilemma is when you have these long tournaments where the strategies involve not just what you’re doing in the moment, but how you respond to other strategies over the course of an iterated session. So now in the case of, yeah, I don’t know what would happen in Network WARS. I think part of its beauty is the dumb AIs as they exist, because it leads to this deeper understanding of the kind of epiphany that I had about AI in general and about games and being embedded in them, knowing where you are in the metagame stack. I mean, that’s the secret. It’s not just outplaying your opponent. It’s where am I relative to my opponent in terms of my actions being determined by the rules of the game, and even a kind of optimal response to the rules of the game versus having a perspective that is outside the game and able to see it.
That’s why Magnus Carlsen is still the best chess player on the planet, even though he’s chosen not to play chess in a way. The fact that Magnus Carlsen got bored of chess is a demonstration of his genius, right? The ability to get bored at a game is the precious thing that we have that is ultimately the most valuable thing that we bring to a game. It’s not just the ability to optimize within the game, but the ability to pull back and see ourselves in the game optimizing and just choose not to. Yeah, I’d rather play this other game.
Jim: Yeah. That’s it. Yeah, that is true. And he does seem to be at that point. Now, let me get to another point, which many of us have thought about, and truthfully, I could do the analysis and see, but I haven’t yet. I’d love to get your subjective sense.
Jim: All of us players who played a bunch get to these places where we’re winning more than 80% of the games. And sometimes, we can keep there for a few hundred games, but inevitably we crash, right? And I’ll often crash down to the high sixties. This is again, for the last 100 games, which is my figure of merit. And so the question is it a run of bad luck, i.e., a long sequence of hard scenarios, or does our attention break in some sense? Or do we get sucked into anti-patterns on some of the heuristics? What’s your thought on that? First I’ll ask, have you had that sense that, okay, I can maintain the plus 80 for a while, but inevitably I break?
Frank: Yeah, no, when things are going well, you feel like, “Oh my God, I’ve cracked this game wide open. There’s no way I’ll ever lose again.” And then you can go on a huge downswing and you’re just like, “What happened?” And it’s sobering to be all of a sudden you’re just losing over and over again and you feel almost like you’ve been kicked out of an airplane or something.
Okay, so my answer to your question is it’s bad luck. I’m not saying that I’m also not susceptible to bad play or tilt or psychological going on a streak where I’m not focused or not paying attention. But I think that this is just the reality of the game. Like poker, the long run is bigger than you think. The terrain of the curve that you’re on, even if that curve is going up, it has bigger spikes and valleys than you expect. You expect it to be smooth, and it really isn’t, and you’re going to go on a stretch of a bunch of games that you lose that you really, you played fine. So I don’t think that those are the result of having a cold hand. I think they’re just a result of randomness.
Jim: Yeah, my guess is it’s both, but I don’t know what the weights are between the two, because I will notice sometimes that I get, if I start playing too automatically and too quickly, that seems to be correlated with these declines where I get into this zen state where it’s just flow. Well, it turns out flow isn’t quite as good as flowishness plus some analysis, something like that.
Frank: But here’s the thing, Jim. As soon as I lose one or two games, I’m pretty focused because I care a lot about my long-term stats. I’m going to go on The Jim Rutt Show, he’s going to look at my stats, he’s going to talk about everybody. I’m really incentivized to make them as good as they could be. And so I start to pay attention. And I don’t think I’m playing worse or making suboptimal decisions during those times. I really think that I’m on a downswing. And look, you played poker. You know what that’s like. Poker is another, you can play for a year and think, oh, I’m never going to lose at this game. I have a huge edge. I’m playing in these soft games, whatever. And then you can just go on a huge run, a terrible, terrible downswing. And the reality is that that can happen to you even if you’re playing optimally. And it’s one of the beauties of poker is that you can never really know. You look at your play and you’re like, “Well, wait a minute.”
Jim: I played right, but I lost anyway, it happens all the time. And of course, why people love Texas hold’em, because you can see all the mechanics, but only after the fact, right? I actually sat at a talk about the overlays, the soft tables. For 10 years, I lived in Santa Fe, New Mexico when I was at Santa Fe Institute. And the casinos, the Indian casinos in northern New Mexico are like shooting fish in a barrel. These are the most ridiculously loose tables ever. And on my book, and I’d only play three or four times a year because it embarrassed me to take money from these Hispanic business guys. They were mostly successful contractors, so they made enough money. But if I went to the casino four times in a year, and I saw those guys there three times, a little bit of induction says, oops, they’re probably there every day.
Jim: And consider this, it seems impossible. But I typically played a one two, no limit Texas hold’em table, and my win rate was $185 an hour, which seems impossible. That’s how soft the table was.
Frank: But that’s an example of you having a big edge and also getting lucky. There’s no way that your actual edge was that high.
Jim: Probably not, but it was big enough. But on the other hand, because they were so wild, I literally did see, fortunately it wasn’t in the hand, and because I knew this one guy’s behavior that he always stayed with seven two unsuited, which for those of you who don’t play Texas hold’em, is the worst starting position. He just had this stupid superstition that he did, and the son of a bitch pulled a boat and beat a guy with a big flush.
Jim: It was a gigantic stake, but I knew I didn’t have a hand to play anyway, but if I did, say I had trips, I would’ve, and I started to see these cards fall. I’d go, “That’s all, bet you’re probably going to get the boat and get the hell out of there.” But even with me with a big, big, big edge, I could certainly go for an eight-hour session and lose money. Without a doubt, right?
Jim: Because again, these guys played a very high variance game, but fortunately, it was tilted in the very, very, very loose. I mean, I think it was an average of 5.5 guys after the flop.
Frank: There’s a certain style of play in some live games where the people are just there to get to play their cards. For them, the game of poker is we all get dealt cards, and then some of us have, you get lucky and you get a good hand and it connects with the flop, and then you win that pot, and now it’s my turn because I got luck. Now it’s your turn because you got lucky. And then occasionally you bluff. They’ll throw a bluff in there some of the times. But mostly, they’re there to just enjoy the churn of numbers and to feel the fluctuations of fate pass through the game. And it’s almost like it’s a moral… It’s like a norm for them.
Okay, now it’s your turn. Now it’s my turn. And if you play the game in a more analytical way where you’re trying to find optimal moves and maximize your chance of winning, it’s almost like you’re disturbing this collective agreement. And that is, it’s a little bit like playing Network WARS. It’s a guy recognizing that the opponents in Network WARS are following a set of rules that you don’t have to follow. And because they’re constrained to those rules and their behavior is completely determined by them, you have this huge advantage because yours isn’t right. You don’t have to sit around waiting until you get lucky. You can make your own luck.
Jim: And as a really hardcore materialist, I don’t believe in luck, right? But those guys clearly do, right? They clearly sense that they can fuel the force and they can’t, I would say.
Frank: It’s so funny because you don’t believe in luck, Jim, but you made a little toy universe that has luck in it because you predetermined this seed, which all of the random roles in Network WARS are actually predetermined by this deity, which exists outside of the game, which is you.
Jim: We live in. Yeah, we live in a Presbyterian universe as it turns out.
Frank: And you have access to this information. The poor bots are just struggling to keep afloat. Whereas you could just look at the code and say, here’s what’s going to happen. You really, you’ve given yourself these superpowers.
Jim: All right. Let’s move back to the topic you brought up, which is the idea of live Network WARS, five human players.
Frank: Yeah, okay.
Jim: One of my biggest fans, and he was all the way from my Windows days as a volunteer tester. He’s played an embarrassing number of times. He keeps lobbying for me to do the five live players, but I don’t know if the game would even make any sense with five players. But because you did hit on a key, key distinction. Is there communications or not between the players? If there’s communications between the players, it’s a totally different game than if there’s not. One of the games I used to play when I was a teenager. We used to, we was Diplomacy. Did you ever play Diplomacy?
Frank: Absolutely. Yeah.
Jim: We were stabbing him in the back. But when we were about 19, we got together a couple of times in the summer, and we were all home from college and played, I thought the ultimate version of Diplomacy, which was we had a $20 buy-in, which back in the days, the minimum wage was a 1.60. And both of us were working for a 1.60, was significant money. And we made the rule that cash bribes were legal, but they had to be recorded with the referee. And the winner of the game was not the person that took over the map.
And we had a rule for allocating the $20 buy-in pot based on the rules of Diplomacy. If you have a four person coalition, you’d divide the pot evenly, all that stuff. But the meta winner was the person that got the most money, whether they won or lost the game of Diplomacy because they were able to suck people into paying them bribes to either augment their part of a winning coalition, or to win more money than anybody in a winning coalition might have via soliciting bribes. And I put that idea back to my friend who said that would be quite brilliant.
Jim: … put that idea back to my friend who said, “That would be quite brilliant if you could figure out how to gamify that,” where you’d have open communications, pipelines, allow cash bribes, require cash buy-in. And then the real, real winner is the person that collected the bribes plus whatever they want as being part of a winning coalition, if they were part of one. That’d be one way.
The other way would be, and this is the war that the poker rooms have. I have never played online poker for money, not even once, because I’m so aware of what leverage even a small amount of collusion provides. And of course, they spend vast sums trying to detect any signs of collusion, but I would say they can’t possibly totally stop it. And so if I were going to do a zero interaction version of Network Wars, I would want to do it with some pretty rigid anonymity features and make it very difficult and have to have a large enough number of games going on that it becomes impractical for people to collude.
Of course, unlike a poker table where each game would be fresh, you could mix and match the users together when they’re ready to play. Any thoughts on those distinctions between the attempt to make it a isolated, no communications, a communications-like diplomacy, or even crazier, a meta game where cash bribes are actually part of the game and part of the strategy?
Frank: Yeah. No, I think it’s a fascinating question. I mean, I think the first thing that you have to acknowledge is that it would be a totally new game, right?
Frank: Because like I said, it looks like Network Wars is a strategy game that you are playing against a bunch of opponents and the other opponents just happen to be AI. But then you quickly realize, no, it’s not really a strategy game. It’s mostly like a puzzle. It’s like a Rubik’s cube and it looks like a strategy game with opponents, but really it’s kind of a Rubik’s cube because these opponents are just rules inside the game. And this is the sense in which the John von Neumann would say, “It’s not a game at all,” but he doesn’t think of chess as a game. For Von Neumann, games are things like poker that involve modeling your opponents and making decisions based on what you think your opponents are going to do. And chess is just a puzzle. It’s just a big math problem. Whereas poker, now that’s a game.
You’re talking about making a new game, which kind of uses the same underlying mechanics, but would quickly, I think, be overwhelmed by these kinds of issues of coordination, collusion, what we call politics in game design in general. Anytime there are opportunities for people to work together and steer the outcomes by colluding in some way, either explicitly or implicitly, that’s politics, right? So, quickly, it would just become a game about politics.
In terms of whether or not you allow communication, I think that’s huge, obviously. I don’t think it would be that different because ultimately, even if you don’t allow communication, your actions in the game are a form of communication. For example, if I want to punish someone because I think that they’ve made a move that is unfair or it’s a move I don’t like, there are ways I can attack them that are demonstrating punishment. I’m like, “Okay, I could take you over here, but I’m not. I’m going to attack this one node and I’m going to leave it there as a warning shot.” You know what I mean?
So, people would find ways to communicate within the actions that you can take within the game, and there would be these elaborate kind of norms, I think, that evolved for communicating and keeping people in check. I think a lot would depend on whether I’m playing multiple sessions with the same people over and over again. If we sit down, then it gets interesting. I would definitely be interested in that. I would like it to be, again, short sessions so that there is some timer. I think you’d need to introduce a timer to keep these things moving. And then I think it would be interesting to see how that evolved. And I love the idea of doing it with money. I actually think that there’s something that happens when you introduce betting that often elevates a game because it encourages people to think not only what is the optimal move in this situation, but to think abstractly, how strong is that move?
To understand the whole range of possibilities. If you think about the doubling cube in backgammon, right? Backgammon, one of the oldest games on the planet that is still currently played for fun, thousands of years old. People have been playing backgammon forever. The doubling cube, which many people think as the thing that really makes backgammon a masterpiece is this ability to decide at a certain point, “I’m so strong in this position that I can offer you double stakes, and you can either have to accept that or not.” And all of a sudden now we’re thinking of it in these meta terms about how the game is going to play out. The doubling cube was just invented a hundred years ago, like early 20th century in the Lower East Side. You know what I mean?
Jim: Oh, I love that. I did not know that.
Frank: Yeah, absolutely. What it did for the game… And another example of this is contract bridge. The difference between playing bridge, playing wist, playing bridge, just trying to find the best moves versus when you have bidding that allows you to make predictions about what’s going to happen and the overall game. And the real skill is in this ability to model what’s going to happen-
Jim: And fault signals, obviously that’s a huge part of contracting.
Frank: Exactly. Now you’re opening up this whole channel for communication. You’re giving moves extra meaning because they now are referring back to this larger system of what your prediction about the overall game is going to be. So, I think that there’s something really interesting about human Network Wars with cash. That’s the version of it that I would be most interested in playing.
Jim: Unfortunately, damn hard to do in the United States due to the sons of guns and their regulations against gambling. I have to put it on the Isle of Man or something like these-
Frank: Or mint a coin?
Frank: Is that how the crypto works? I don’t know.
Jim: We could possibly do that. It’s possible. I’m going to go back and think again about multiplayer Network Wars. Maybe it’s not as hopeless as I think, and I’ll have to explore the design space in a larger way, including how I think about signaling and implicit politics or explicit politics, et cetera. I have a list of 20 enhancements for Network Wars 2.0, but none of them have anything to do with the core game itself. They’re all about leaderboards and ELO ratings and things of that sort.
Actually, I have one called the Advanced Duplicate Tournament. They take off the idea of duplicate bridge, ranks. People work their way up by, if you win 20 games in a row, you become a lieutenant colonel or something. But nothing in the game itself, zero. I’m actually quite happy with the game as it is in terms of a single player game. Is there any ideas that you had, as the king of games, when you were playing this thing that you would do differently that might make it a better game?
Frank: No. I think you’re thinking along exactly the right lines. This, first of all, I think the domain of metagame design is itself really fun and fascinating, and a place where a lot of people don’t innovate, right? They’re not thinking about how to take an existing game or existing set of mechanics and find the right context for it, frame it in the right way. Think about the kind of long-term experience. Unfortunately, we have the kind of business of games, especially the business of mobile games, is a real cesspool, and it just, it’s drags… There’s an strong incentive that kind of drags games down into this kind of Vegas style approach where you’re just trying to maximize time on device because with the way the revenue generation works.
Jim: Ad based the revenue, which I hate.
Frank: If you’re just trying to entrap people and maximize time on device and get them to play as much as possible, that is something that is counter to discovering the kind of really genuinely rich and interesting and cool kinds of contexts that you could make. I’ve often thought that Las Vegas… I mean, I love Las Vegas. I like gambling. I think it’s fun, but I also recognize the way it degrades certain kinds of experiences. It kind of drags them down to this simplistic and rote kind of dynamics. I think that it’s an unexplored terrain, unexplored design terrain, like smart, interesting gambling games. Games that combine randomness and fluctuation and variance and gambling and betting and wagering with strategy and calculation and interesting social interaction.
Imagine if the lottery was an interesting puzzle that people could do. And in a sense, it is, right? Because people can sometimes discover that there’s an edge in the lottery that people have overlooked and that they find certain techniques or certain ways. But imagine designing games like that, that would encourage people and incentivize people to be creative and to think about the thing, not just as, “Am I going to get touched by the hand of God and be the lucky one, be the chosen one?” But also, “Can I apply myself to figure out what’s going on here and improve my chances?”
Jim: Yeah, it’s an interesting idea. Yeah. Large scale emergent gambling that has the idea of gradually uncovering heuristics.
Frank: Yeah. Imagine if you and your friends could go to Vegas and there was a table game that you could play where you were actually collaborating and working together and making decisions the same way you would if you were on a League of Legends team, where it’s like, “Okay, your role is to do this, and my role is to do this.” I mean, that’s what it’s like when you’re part of a card counting group, but I’m talking about something that’s explicitly designed to allow for that and encourage that.
Jim: And to be fair-
Frank: It’s a hard design problem, but it’s not impossible.
Jim: That’s interesting. God damn it, now you’ve put another worm in my head, right? I’ll definitely be thinking about this. All right. Well, we’ve talked a bunch about Network Wars. It’s been a wonderful conversation. I don’t think I’ve ever talked to anybody about Network Wars quite this long. I do have one guy who I talk to about this.
In the last 10 minutes, why don’t you back up a little bit and tell us a little bit about your thinking about games in general, and particularly maybe some of the ideas that you’re going to be putting in your new book, the Beauty of Games or otherwise. What is the Frank Lance big picture about games? You’ve certainly alluded to bits and pieces of it, but let’s see if you could give us a few minutes, 10 minutes, maybe worth of straight from the dude himself.
Frank: Okay. Well, before we move off of Network Wars, there’s one more thing I want to say about it, which is that a year ago, I had a really bad year. My wife had some health issues, and she’s fine now, so everything’s good. But it was a really, really, really bad year. The kind of stuff that you’re in the hospital, you’re going through surgeries, you’re going through treatments, and it was really rough. I played a lot of Network Wars during that time, and it kind of kept me sane. I think there was a quality of games that they can do that for people. I think it’s a real value of games. And I think it’s sometimes looked down on. We think of it as like, oh, escapism, this ability of games to absorb our attention and to expand in our heads to such a degree that they crowd out our own thoughts and feelings.
I think there can be great beauty in that. And I think that the fact that a game can be a source of comfort in addition to being a source of ideas and intellectual exploration. That sometimes games can be a thing where you’re standing in a dark place and you need to have a little warmth and a little light. And games can be that source. I made a game called Drop Seven, that I think a lot of people have this kind of relationship too. It’s a very abstract game. It’s similar to Network Wars in some ways, in the sense that it’s minimalist game. It’s very abstract. It’s a little puzzle game.
A friend of mine once said that he was on an airplane, and this airplane hit some very bad turbulence, the kind that you’re not sure if you’re going to make it, right? And this was really, really bad. He looked over and the woman sitting next to him was clearly very upset, very scared, and she was playing Drop Seven. That made me really happy when he told me the story. It made me very proud that at this moment, at this dark moment standing on the precipice looking into the void, that this woman turned to this game that I had made and found some comfort there. That made me really proud, and your game did that for me.
Jim: I’m really pleased to hear that. That really warms my heart because I knew it would be a game people would enjoy, and it’s a pastime that is nonetheless kind of good for heightening your cognition. I don’t feel terrible about sucking people into it, but I’m also glad I deliver this kind of experience that I really didn’t even think about. So, thank you for telling me about that. I really made my day.
Frank: Well, thank you. Thank you. I needed it and it was there for me, so I appreciate that. I wrote this book. I’ve been teaching game design for a long time. I love games. I love thinking about games. I love playing games. I also think that games are deeply confusing. I think a lot of people who aren’t inside of the world of games, who don’t play a lot of games, I think they look at games and there’s something, they realize there’s something going on there. It’s kind of interesting. It has something to do with computers. There’s a lot of people doing it, but it also seems very confusing. What is it? What are these things? Are they hedonistic appliances that you kind of plug into the wall and then they take you to a dream world and you get to imagine that you’re a fireman or a dragon or whatever.
I want, with my book, I want to share my way of looking at games and thinking about what makes them interesting and beautiful and meaningful and important. I think it really starts from the idea that they are something like music or literature or film, that they’re a creative form. They’re something we do for their own sake that we do because we find them interesting and meaningful and beautiful and expressive, but in a very particular way. Unlike any other art form, games have so much engineering in them. They’re so much about problem solving that they really are the art form of problem solving. They’re the art form of interactivity.
They have this deep relationship to computers, like games predate computers, obviously. Games have been around since before the dawn of civilization, but I think in some ways in games invented computers. You know what I mean? Games were doing computation before computers existed, games were one of the inspirations for computers.
Jim: And we know for sure that gambling motivated Pascal-
Jim: … to understand probability for sure, right?
Frank: Yeah, for sure. I think looking at the Mechanical Turk was supposedly one of the things that inspired Babbage to work on the analytical engine. And Turing, one of the first things he did was write a chess program. Before he had invented the idea of computers, he was like, “Well, here’s how you would write a chess playing program.” So, this idea of algorithmic thinking, of the rules-based behavior exists in games. And when computers finally came around, I think the intersection of games and computers obviously brought about this huge explosion of creativity and led to the world we live in now, where digital games and video games and computer games are arguably one of the most complex, interesting, important forms of culture that’s on the planet.
But I still think poorly understood, right? There’s a sense in which they are so complex and hard to wrap your head around that kind of building this shared literacy that is an important part of having them continue to kind of evolve and grow and fulfill their capacity, I think is the project that I want to be a part of. Is one of the reasons that I was drawn into teaching games and game design and helping develop the program at NYU. That’s essentially what the overall project of the book is. It’s about trying to understand how to think about games within this larger context of them as a form of culture that have the same kind of expressive power that these other creative forms have, but their own unique version of it that has to do with problem solving and logic and rules and all of these things that we think of as in some ways of being the opposite of creativity.
We think of creativity as being about spontaneous, emotional responses, intuitive responses. And then we think of games about being these kind of deterministic things where the rules create this limited set of possible behaviors and you’re trying to extrapolate from that. But the reality is that, that is I think the thing that gives games as a form of culture, their angle on the contemporary world that is unlike any other form of culture. I think that they have a power to help us understand the world we live in, ourselves, each other in a way that is just profound and meaningful and beautiful.
And it is this contrast, I think, between logic and reason, instrumental reason, the things that we apply ourselves to when we’re trying to solve a problem and the things that we find irreducibly complex and beautiful about art. The things that art is by its nature, it kind of always eludes our attempts to explain it and to form an objective explanation about how it works and what it’s for. It’s because it exists a little bit outside of the ordinary course of events where we’re explaining things and putting values on things and determining. And so it is in this combination, I think, that games have, they open up a window on the world that is truly special and different. That’s kind of what the book is exploring.
Jim: Sounds fascinating as can be. Let’s definitely have you back on the show when it comes out. We’ll go into the book in great detail. ‘Cause as I mentioned, I’ve been playing games and designing games, playing strategy games since I was 10 and designing them from the time I was 12. Not that I’ve ever made a living at it, but I’m sure we can have a great conversation. I really look forward to it.
As we sign off here, people that want to play Network Wars, networkwars.com. We’ll provide you links to the Apple Store, app store and to Google Play and you can search there for Network Wars, two words. Download the app, 99 cents. And you can have as much fun as Frank and I have had playing this game and having this wonderful conversation today. Frank, it’s really been great.
Frank: Well, thank you very much, Jim. I’ve enjoyed it.