CGD Podcast Episode 22: Mind Games and Reading the Opponent


What does it mean for a game to involve a lot of “mind games”? Can you really make “reads” off of an opponent and predict what he’s going to do? What’s the difference between “reading the opponent” and “a lucky guess”? This episode explores these questions, discussing games like Poker, Street Fighter, Rock Paper Scissors, Yomi and more.

Support the show on Patreon by clicking HERE!

  • tinytouchtales

    I’d like to see a Frank Lantz episode, you guys seem to be in an interesting love/hate relationship. Could be good podcast material ;).

  • Tuomas

    In games where reading isn’t a big part of how the game plays out, reading your opponent definitely can occur. If one player isn’t paying much or any attention to the reads they are giving, there’s a good chance their opponent will get actual benefit from that. You know the opponent won’t be bluffing, so picking up a read can be advantageous.

    That said, if bluffing and reading are a core part of a game, then players will expect that, and the mind games start to happen: He knows I might be bluffing, so I’m going to do X etc… At that point the outcome is basically random, like you said in the episode, Keith. Just wanted to point out a scenario where reading is an actual thing and a skill that can be utilized.

  • Rob Seater

    I agree. Reading an opponent is a real thing when dealing with new players or players who are distracted. I agree that it is an illusion if dealing with attentive experts, which is usually how I judge if a strategy game is any good.

  • When it’s me and Frank Lantz, it’s always an “episode”. But yeah I agree!

  • Viper

    At least when it comes to ‘rock paper scissors’ robots can read humans perfectly and win every time:

  • Venom

    even with expirienced players it’s a thing, at least in some fighting games like Smash, where it’s often part of the metagame to condition an opponent (and thus be able to read them). For instance it’s not an uncommon strategy in smash to get hit a few times by a specific attack so the opponent gets consistent in that situation and then later punish it hard when it happens again.

    And in some cases its even part of the metagame to hear the other player input moves and thus be able to read the opponent (which is why american players of streetfighter tend to play softly while that doesn’t matter for japanese players, as american players are put next to eachother while japanese players not. Which is also why americans players purposely play loudly sometimes.)

  • Isaac Shalev

    The robot is actually seeing the hand starting to take a shape and is able to read that information and respond to it very quickly. That feels distinct from reading a player’s intentions – it’s just a more rapid recognition of what a player is DOING, not an accurate assessment of what they’re about to do.

  • Isaac Shalev

    At one point you said that player’s style of play is information that exists outside of the game state. Yet a player’s style is simply the generalization of their in-game choices, across many iterations of playing that same game. In high-stakes strategy games like sports or tournament games a great deal of time and effort is spent studying opponents’ game styles, suggesting that there is a lot of value in knowing how opponents are likely to respond to a set of circumstances. More broadly, isn’t this kind of meta-game part of the game itself? It’s player interactions that arise from the game system.

  • Yeah, you can do statistical analysis – that’s basically what you’d be doing there. I think ideally you would not be able to use information from outside the current match to gain an unfair advantage – imagine a scenario where I researched your history and you did not research mine. That’s kind of unfair and silly to allow. Other than just making games single-player, another way to solve this would be to have anonymous matchmaking online – during the game you wouldn’t know who you’re playing against, until perhaps after the match is over.

  • Isaac Shalev

    Why is it silly? Players show up to a game w many relevant asymmetries. Why is a sense of the other player’s tendencies so problematic? It seems like their is some underlying principle of fairness that you haven’t articulated, but which would be interesting to talk about.

    Having listened to 20 episodes in a couple of days, I got the sense that you believe in a kind of Platonic ideal of a strategy game, where external factors don’t influence the game contest, and where every interaction and consequence has full player agency and intentionality. In a way, it’s not only the game that is Platonic, but the players must be too. It’s an interesting model, but it’s so different from the experiential Iens I use to think about games.

    (PS still Isaac here, Discus is being rude to me)

  • It’s not an un-articulated sense of fairness, it’s a very-articulated concept of what a game is and how it works. You put in input and you get meaningful feedback. If you put in strong inputs, but you get negative feedback because “your opponent googled you and knows some personal stuff about your play style”, that’s as bad as “your opponent has more money than you do in real life” (pay to win) which everyone knows is bad. In both cases, some external-to-the-system thing is determining who’s winning, and players are not getting the correct feedback based on their inputs.

  • Isaac Shalev

    Let’s assume that an opponent knew nothing of my play style. It’s conceivable that they would enact a strategy that would be identical to the one they would have chosen without that knowledge. I might zergling rush in Starcraft if I know my opponent likes to turtle, but I might choose to do that anyway, with no knowledge of my opponent. The ‘negative feedback’ of losing to the rush is valid system feedback, irrespective of WHY my opponent chose that strategy.

    Let’s take an extreme case. I have a quirk that I never play Rock in RPS. My opponent, knowing this about me, decides to play mostly scissors and rock, and doesn’t bother playing paper. The game system itself hasn’t provided any bad feedback when I get trounced – it has provided the appropriate feedback that I’m playing poorly.

    Pay-to-win is another instructive example. PTW isn’t a problem because it’s external to the game system or b/c it warps feedback on player choices. PTW is a problem because it’s a dominant strategy. If you want to win, it’s the only option. And the fact that the game-maker charges you to use that strategy makes them no different than a corrupt referee.

    My larger feedback is that despite the magic circle, players bring a lot of external differences into the game system. One player is smarter, the other has some direct experience that informs gameplay, another has a short attention span, a quicker trigger, a greater tolerance for risk, etc. All of these impact play, often much more deeply than opponent research would. We both agree that we want games to be a fair contest, and that at some level of outcome disparity, you’re not really playing a game as much as engaging in a game-like activity. But I don’t think that pruning every sources of potential inequality between the players is a good approach.

  • prester_john

    Interesting article. Made me think of something cool: a potential area for a read (or.. well… something like a read?) is when two unfamiliar players are playing multiple rounds of a game which combines mechanics and strategy.
    Let’s say our game is rock paper scissors but each requires different mechanics to successfully execute. Now the game is still mostly random, but if you discover a relative strength or weakness in yourself compared to your opponent, you benefit by adjusting your strategy to compensate.
    In rock paper scissors this is fairly simple, but I’d imagine this could do tons of crazy things to more strategically deep games.

  • David B Usry Jr.

    This talk renders the most important events in baseball, fencing, tennis, volleyball, etc completely invalid. Good luck with that.

  • Thanks. But I would argue that the “reading” are not the most important events in these things. I think what’s most important in them is that they’re social events. But also, even without the reading, it’s not like there’s no skill to these things – you have tons of execution skill which totally is a real thing, plus a very tiny amount of strategy.

  • David B Usry Jr.

    It feels like you’re arguing that reads are nearly impossible and that makes them invalid. I would argue that reads are at the edge of what’s humanly possible and that makes them a challenging and useful game element. One that high speed games with vast outcomes like tennis and fencing are completely reliant on. I don’t think that these games can be systematized as efficiently as poker and rock paper scissors can and that’s by design.

  • David B Usry Jr.

    Anyway, I like your show and I think you should interview Mike Z.