CGD Podcast Episode 24 – Execution and Single-Player Games, with Frank Lantz

cgdplogo_superwideToday we have another episode with Frank Lantz, game designer, writer, and Director of the NYU Game Center. Today’s show involved two major topics: execution, and my seemingly crazy idea about how single-player should probably be the “default” number of players for a strategy game (something I’m going to be writing an article about soon). Also, Frank gives some of his own game design faux-pas thoughts near the end of the episode.

There were some technical issues during the recording, so please forgive the somewhat strange format for this episode. Hopefully it’s clear enough what we were both trying to say.

 

If you enjoyed the episode, please spread the word on social media! Or, become a patron at Patreon.com.

  • hilbert90

    I didn’t find this one as interesting (but still very much enjoyed it). I think one thing that did this is that Frank seems to over-romanticize classic games like Chess. High level chess is almost entirely a huge memorization problem followed by a huge calculation problem. There’s pretty much no room for interesting and creative play. This was even said by some famous pro player (Fisher or Lasker maybe?). It had to do with how he was drawn to the game for this reason as a kid, but then the game lost this magical quality when he got good.

  • Nick Splendorr

    Hi! I haven’t been listening long, but I really enjoy your podcast, Keith! I’ve also enjoyed Frank’s work in the past.

    I don’t want to be overly negative here, but I’m halfway through this episode and felt a need to comment. I felt like Frank was being very dismissive throughout this conversation. He wasn’t listening to you very well. You would say something with clear (to me) distinctions from what he said, and then he would reply, “But aren’t we saying the same thing?” Or, “Isn’t that a point for me?” When, no, it wasn’t; but it seemed that anything Frank heard, he could only hear in the context of the point he would prefer to make. This was frustrating, and I wanted Keith to know that he was making good points; they just weren’t being heard.

  • Nick Splendorr

    And to be clear, I don’t think Frank is a bad person or any other extreme extrapolation! I just wanted to say this particular conversation didn’t feel as constructive as it could have been, for this reason. I still enjoyed it overall, and found the different points you each made to be interesting and useful in their own contexts.

  • Jimmy Mac

    I think I disagree with you about single player games. I don’t think the point of games in general is simply to learn, but it’s to learn something useful. The psychologist Vygotsky had a phrase for the things a child could do with the help of an adult, which he termed “the zone of proximal development.” I think that’s where you want to be when you are learning through games, just a bit past the edge of what you could think, strategize, or analyze alone, learning from the game and your opponent. To that end, I think learning how to play against other people is more valuable than playing against a computer. People do all sorts of odd, counter-intuitive, or irrational things that no machine would, and to gain learning that would be applicable outside of games, you need to play against people. I think playing against computers can be valuable, interesting, and fun, I just don’t think it’s categorically better than playing against people. Playing against people isn’t always better than playing against a machine either, but when the situation is a good test, like evenly matched skills, then human opponents I think are best.

  • Jake Forbes

    Really enjoying these chats with Frank.

    There’s something that came to mind during the multiplayer discussion that irked me during your mind games podcast a few weeks back, and that’s the use of feints and deception. In that past podcast, you primarily focused on a player’s ability to read the opponent rather than player’s agency to fool the opponent. In a two-player or team game, be it chess, MTG, League of Legends, a war game or anything of a tactical nature (and most sports) the possibility of deception is inseparable from the possibility space. When playing a human opponent of approximately equal skill (ability to identify “optimal” move is equal to your own), then at any point in time, if your opponent does something that doesn’t gel with your perception of optimal, you have to ask if they were “wrong” or if they’re up to something. How do you reconcile the concept of laying traps or making feints with the degenerate RNG nature of mind games? And accepting that deception has a place in games, would you concede that it is more FUN to deceive or be fooled by a human than an AI?

  • Keith says he’s skeptical of any feature of a game that makes design easy (and gives randomness and multiplayer as examples of both). Yet he simultaneously explains that single player games are great because they make things easier to control for the designer. All three of those claims are breezed by, but with such a big contradiction none of them seem justified. While I get Keith’s point about the allure of randomness and multiplayer, it’s worth remembering that these two approaches have a slew of design/testing challenges of their own!

    In general, the participants often appeared to be talking past each other. Frank is harping on the value of games (a multitude of games!) and Keith doesn’t see the disagreement. I think I get it.

    I really enjoy this podcast and the discussion that comes out of it, but I have to say there’s a consistent theme at play that bugs the hell out of me. The theme of the podcast seems to be:

    “There is an optimal form of game and we (possibly *I*) will be discovering it soon. All else is pointless.”

    We talk about mountains here, but it sounds like Keith is always talking about a *single* mountain. One global optimum that we’ll soon achieve (or get vastly closer to than we ever have).

    It’s gotta be strategy because everything else is a toy (read: a waste of time).

    It can’t have reading because reading doesn’t exist.

    It can’t have randomness because learning is artificially stretched out.

    It can’t have asymmetry because… well, you get the point.

    I understand having particular interests and neglecting other things, but the podcast tends to go beyond that and trashes everything that Keith is not particularly interested in at the moment. There’s a big focus on dismissing alternatives and, if players enjoy those things, attributing it to naivety on their part. Keith wrote a few roguelikes and then proceeded to categorically slam the genre. He wrote about how League of Legends was the best game in the world and a few months later claims that single player games are optimal. I don’t think it’s intended, but such discussion sometimes comes across as a tad arrogant and erratic.

    The thought that there’s some objectively optimal game strikes me as totally absurd. Even if you apply something like Sam Harris’s “moral landscape” reasoning and assume that game design is a field that is subject to objective analysis, you haven’t demonstrated that there aren’t multiple (or even infinite) equally high peaks.

    And even if there was a single game that was objectively the best by a mile, many of us would get bored of just playing that one game. I know I would.

    I could continue with the “classics” issue, games as art vs science, my skepticism about an impending guaranteed advancement in design, but it’s beating a dead horse.

    Anyway, maybe that’s all a big strawman, but regardless it’s the taste I get after listening to each episode. And that’s the attitude Frank seems to be fighting against. I really do like the podcast even if, admittedly, I’m sometimes raging while listening. It’s not very productive to always agree. đŸ™‚

  • >”There is an optimal form of game and we (possibly *I*) will be discovering it soon. All else is pointless.”

    No, it would be: Games can be better, maybe significantly better, and we can discover better guidelines that will help us create better games.

    >Keith wrote a few roguelikes and then proceeded to categorically slam the genre.

    There is no hypocrisy here; as someone who has made Rogue-likes I feel like I understand them well enough to make a good judgment about them. My opinion did change, if that’s what you’re getting at. I originally thought Rogue-likes were a promising game design pattern (back in 2006-2011ish) but have come to understand that that was a mistake. My opinions do actually change over time, which I think is something videogame players could do well to emulate. Many game designers in their 30s and 40s are walking around with the same opinions they had about videogames that they had when they were 10, and to me that seems highly suspect.

    >He wrote about how League of Legends was the best game in the world and a
    few months later claims that single player games are optimal

    Right now, a game doesn’t have to “be optimal” or “have optimal qualities” to be the “best game in the world” – it only has to be better than the other things which currently exist. League, despite having massive problems (did you read my article? I was pretty harsh on League), is still overall better than the other things which exist. So again: that’s *despite* the fact that it’s not single player.

    >The thought that there’s some objectively optimal game strikes me as totally absurd.

    Ok. I am pretty sure I’ve never suggested this. All I ever say is that there are *better* games, and there are *better* guidelines for producing better games. I don’t know, and also don’t care about whether there is some theoretical “optimal” game – it’s pretty irrelevant to the work anyone does. When I say “ideal game design”, I’m referring to: what are the currently best known guidelines/criteria for making games?

    Thanks for sharing your thoughts.

  • Because feints don’t actually add anything to a system, the same way an RNG doesn’t actually add anything. All it does is make there be some game outcomes that are determined by a basically random thing. Take a certain game situation, where I have an optimal position. But then you “feint”, and I actually fall for it, and you end up winning. Basically, that’s “false feedback” for that match. My general position and strategy was superior.

    >would you concede that it is more FUN to deceive or be fooled by a human than an AI and there is merit in that?

    I am not sure that it’s more fun to be deceived by a human than the AI. I think we haven’t written much AI that tries to “feint” – in fact, I can’t remember *ever* seeing an AI that did that, in any game. If they did, I think it would basically be equal to the amount of fun you get from a feint that comes from a human. Also, I think the kind of “fun” it is is pretty similar to the kind of “fun” of rolling a 20 in D&D or the “fun” of drawing a card. It’s “random surprise”, so it’s a bit cheap and has a cost on the efficiency of the game.

  • I agree that if your goal is to learn about *people* and how they work, as in some kind of sociological experiment, then single player games are less ideal than multiplayer. But I think that strategy games are generally about understanding the system of rules, and not really about understanding psychology.

  • Jake Forbes

    I guess I just fundamentally disagree with a definition of games that excludes the albeit messy tastes and tactics of those who play games. One of my very favorite games is Agricola (or most Uwe Rosenberg games) and he’s a designer whose aesthetics are very close to your stated ideals, to the point where his mechanics easily map to single player experiences. Puerto Rico and Terra Mystica are also wonderful games where randomness and secret information are removed. And with games like those, it’s true that human fallibility can spoil an experience where one player’s sub-optimal choices are inadvertently king-making for another player who randomly reaped a windfall from turn order. Those games have less room for mind-games because they are fundamentally about building engines of efficiency. They’re modeled on economics in the same way that many tactical games with a deception component are modeled on combat. Puerto Rico is maybe a 2 hr game. When the same concept was adapted for cards as San Juan, it suddenly introduces more RNG and secret information. Race for the galaxy built on that foundation and added a blind role selection process that also adds a mind-game component with blind role selection. At the end of the day, Puerto Rico is less susceptible to lucky swings, but RftG offers a lot more variety of encounter; they’re both economic engine games, but the messiness of RftG allows for a much greater variety of encounter. The systems to master — identifying efficiencies, grabbing windfalls while denying them your opponent, and anticipating endgame — stay the same across the different interpretations. What is wrong with accepting a spectrum of RNG and mind game elements allowing for tastes and aesthetics? For all your emphasis on categorizing games vs puzzles vs toys, it seems like puzzles are your true passion, which is totally okay. It’s the messy human stuff that makes games beloved.

  • Jake Forbes

    re: AI use of deception, I expect that the reason we don’t see more of this is that players don’t respond well to it. If a human behaves erratically, we try to rationalize it and figure out a motive. When an AI does the same, and we know it’s an AI, it seems broken or unfair. When players are up against an AI opponent, there are a different set of assumptions at work. I doubt many players wrap a narrative around a single player game session and say that they were “outsmarted” but rather that they frame it as if they were unlucky or the opponent was unfair, even if luck is minimized and fairness is greater vs the AI.

  • Jeff S

    From what I’ve observed, Frank has a need to “score points” in the conversation, or one of them isn’t playing with the rules in mind (is it an argument/conversation/debate/etc). In the beginning, you can even hear frank say “tell me what you think and I’ll tell you why you’re wrong.”

    Their convo’s in a nutshell: “Hey, I’m trying to make a great game but when I used “this” I got the negative side effects.” (Instead of asking for a goal or suggesting a way to remove or mitigate side effects) “Hey, man these are great things and they’re totally awesome and a bunch of people love them.”

  • Nick Splendorr

    I’d like to follow up and say that I finished the podcast later, and thoroughly enjoyed the latter half. This was a great conversation to listen in on, even if it started out with you guys trying to talk about different things. Thanks for recording this, and I look forward to hearing more from both of you!

  • Ben Ritter

    This raises an interesting question for me: How would a strategy game avoid degenerating in the face of years of study? Ideally one would still be able to learn more about the system. If the game has no hidden information then the possibility space has to be huge and the game has degenerated into a look-ahead contest. If there is an information horizon and possibility space is small enough, then the game has degenerated into a probabilistic reasoning contest (with all the associated trade-offs). At this point the calculations that expose the underlying probabilities are just a mandatory prerequisite to making the “interesting” decisions in the system.

    What am I missing. Or is an ideal strategy game in Keith’s design just a puzzle that no one has solved yet?

  • Ben Ritter

    > > “There is an optimal form of game and we (possibly *I*) will be discovering it soon. All else is pointless.”

    > No, it would be: Games can be better, maybe significantly better, and we
    can discover better guidelines that will help us create better games.

    I understand that the podcast has this theme and all of your points are made in this context. I think it contribute to that goal too, that’s why I love the podcast.

    That being said, I agree with @nluqo:disqus that the podcast often *seems to be* a quest to find the optimal strategy game. To keep the feeling in check, every time I hear you say “a game should …” I read “a strategy game should usually …”.

  • There’s a middle ground you want to achieve between “no information horizon calculation contest” and “information horizon right up in your face randomness”. Beyond that, it’s a matter of strong design and careful balance – probably taking years of testing and iteration, at least – that will leave you with a *pretty* strong system. But yeah, if you’re looking for something that is like, “go/chess levels of unsolvable”, I think that’s very very difficult. Luckily, games don’t have to be anywhere near that “deep” to be good strategy games.

  • hilbert90

    Oh. I missed this response somehow. What about something like StarCraft? Forget for a second the insane UI barrier to entry and pure mechanical training that goes into it. As a strategy game it certainly isn’t a look-ahead contest because of certain circularity issues (I counter your unit, but you know I’m going to counter with that so you counter my counter so I know that and I counter your counter to my counter blah blah blah until you’re back to the original unit and the circle starts again). But I also don’t think there is a big set of probabilities to be memorized. Yet there are all sorts of interesting decisions to make still.

  • Isaac Shalev

    I’m not sure I understand what you mean when you say feints don’t add anything to the system. Are you saying that they don’t enhance the system? Because they clearly are part of the system – the only way to feint is to manipulate the game mechanisms.

    You also have a narrow definition of feint, which seems to be ‘a suboptimal move intended to goad the opponent into making a worse move, thus benefiting in the tradeoff.’ The point of contention for me is the ‘suboptimal’. Games only exist in the space between dominant and dominated strategies. In that space, there can only be contentions about the utility of a move. Moreover, the payoff of any given move depends on the choices of your opponent. The word ‘feint’ is a retroactive description of a move or series of moves in the context of the game. But if an opponent had ignored the feint, perhaps I could have expanded it into a profitable line of attack.In retrospect we’d say that my opponent ignored the early signs of a looming threat.