Rock Paper Scissors Relationships in Strategy (Board) Games.

A look at how hard counters shape player strategy and predictions.

Okay so I should be writing an assignment now which clearly means that I’m actually going to write a little essay on game design — particularly on the ‘simultaneous reveal’ mechanic, and the idea of rock-paper-scissors relationships between player actions. To demonstrate my points I will be referencing Sirlin Games’ “Yomi” and Level99Games’ “BattleCON” series of games. I will also refer to a game that I was previously making, but have since stopped for various reasons, and despite that it never had a title I will refer to it as “My Little Shootout: Cowboys are Tragic” (also yes I will probably name my game as such if I ever do work on it again).

Okay so what are the two mechanics?

‘Simultaneous Reveal’ is what I call a mechanic where players need to select an action/multiple actions (e.g. playing a card), and then once both have selected their action, reveal what actions they decided to use simultaneously. The simplest example of this, of course, would be “Rock-Paper-Scissors”. There is a countdown, and then each player reveals which gesture they chose. Both Yomi and BattleCON use this mechanism (in Yomi you play one card, BattleCON you choose a pair of cards to play). The creator of “Yomi” calls this a “double blind reveal” which also makes sense.

Simultaneous Reveal is a fun mechanic because it can simulate reality (life rarely involves taking turns) — and it greatly enhances predictive gameplay. Instead of other games where you largely react to an opponent’s last turn and perhaps spare some thought for their future turns, when you have a game in which each round comes down to which cards you reveal in that round, it leads to a lot of prediction and mindgames. I really like this style of gameplay and I feel it could be explored further.

Rock-Paper-Scissors” relationships should be self-explanatory, but I’ll still explain them. They are relationships in which different actions hard counter/are hard countered by each other in a circular relationship. As you know, Rock crushes Scissors, Scissors cut Paper, Paper “covers”/suffocates/hugs Rock. If you play Scissors you will beat Paper, but lose to Rock. Yomi and BattleCON both utilise these ideas — though BattleCON does it very implicitly, while Yomi explicitly has a “Block/Dodge > Attack > Throw” relationship.

I’ll quickly explain BattleCON’s move relationships — each character in BattleCON shares 6 bases (named Shot, Burst, Dash, Drive, Strike, Grasp) and has their own unique Character base. These are paired with a character’s unique ‘Styles’ that offer stats and effects (often with adjective names like “Paralysing”) to make an ‘Attack Pair’. That pair is the move you reveal for that turn. So you may play “Paralysing Grasp” — and those two cards combine to make a single move. Now — you can play BattleCON without ever seeing a rock-paper-scissors relationship (there is never a hard counter), but the way you usually pair styles and bases leads to attacks that can often be categorised. That is, most effective combinations usually create an ‘archetypal’ kind of attack. In this way you will sometimes see a relationship where “Counter-attacking Moves” (often with the ‘Strike’ base) > “Advancing Moves” (Drive) > “Projectile Moves” (Shot), which, in turn, usually beat Counter-Attacks. The other bases can often fit into one or more of these categories (Burst can Counter-Attack, but is also a Projectile).

With that said however, there is so much variation in characters, styles, etc. that the “rock paper scissors” relationship is never direct. As mentioned, there is never a “hard counter” relationship. Some Advancing moves will beat some Countering Moves, in some situations.

In the game’s divergence on the “rock-paper-scissors” relationships (hereby referred to as “RPS”) — I find that BattleCON pulls ahead as a game that I enjoy more than Yomi. Now don’t get me wrong, both games are great — but I feel that BattleCON has superior mechanics, and superior strategic gameplay in this regard.

Factors that Weigh into Strategic Predictions.

Now, at a round-by-round level, both Yomi and BattleCON operate in a similar way. You have a hand of cards — you choose the cards you want to use (in Yomi that is a single card, in BattleCON that is your ‘attack pair’), the cards are then revealed and you “work out what happens from there”. Both games have very different processes for what happens after the reveal. In BattleCON, you work out which character was quicker to act (initiative), and then they carry out their actions. If the opponent is not stunned during the round, they try to carry out their actions.

In Yomi, you reveal the cards, and depending on the RPS relationship, you work out who is the aggressor, and who is the defender for that round (though if they both play block/dodge they are both defenders and nothing happens). Because the defender really has little to do that round, you could call them the “passive” player, compared to the “active” player.

There is also a difference in how the game is framed — situationally, and this is where I feel BattleCON makes better use of its prediction mechanics.

While Yomi has some external factors to the game that can sway you to choose one kind of attack over another, such as hand size, BattleCON has very obvious ‘prediction cues’ that can inform your decisions. BattleCON has a 7-square board (arranged in a single row) on which your characters stand. This represents how close/far away they are from each other in 2D space (imagine Street Fighter). This relative position greatly influences predictions - as if you are far away, you are more likely to use projectile moves, and if you’re close, you are much less likely. If your opponent is a medium range but they are playing a close range character, you can predict that they will try to advance on you. This robust external system allows you to punish mistakes (why is the close range character so far away?) and make use of advantageous situations when they arise. The persistent nature of the character’s position also means that ‘game situations’ such as advantages don’t appear and disappear round by round, but must eventually be overcome.

BattleCON also notably has multiple ‘discard piles’ where your attacks go once you use them — at the end of each round, you move Discard Pile 2 to Discard Pile 1's position, and put the attack pair previously in Discard Pile 1 into your hand — so effectively, when you’ve used an attack in BattleCON, you typically need to wait 2 turns to use those 2 cards again.

This greatly affects how you can read your opponent — if their primary movement cards are in their discard piles for example, you know that they’re likely to stand still. If their faster attacks are in the discard piles, you can know that you will have a speed advantage this turn. It also cuts down the options significantly just by removing those cards — you know that your opponent only has 3 Styles and 5 Bases to choose from, AND you know exactly which cards are available to them.

When you compare this to Yomi, Yomi has a static discard pile where used cards are placed, and because each character has a 54 card deck of fairly similar actions (and also that have 2 potential actions per card), you can’t really predict meaningfully what could be coming. Much less can you predict what the opponent will choose from in their hand that can range in size from 1 to 12 cards.

In Yomi, you can try to predict what the opponent is doing partially on their hand-size. Blocking is good for increasing hand sizes, and so is more likely to be undertaken when one’s hand is low. When your hand is high however you are more able to string together combos of attacks, so you are more likely to attack.

Ultimately though, these have a small sway but little impact, in my experience. As previously mentioned, Yomi has a hard counter system, in which you don’t want to be the “passive” player who loses a card from the round’s exchange, and potentially even loses a boatload of life points. Even if you would love to block, if you think that your opponent will throw, you shouldn’t. Your agency and short term goals are overpowered by the hard counter relationship.

“Yomi Levels”

You may be thinking now “Ah ha! But anticipating that your opponent will throw to counter your block is prediction!” — sure! But what is to stop the opponent from reading my attack to counter his throw, and then use a block card? But if I read that and counter with my throw, then my opponent should attack — which takes us full circle, and I should block again.

Yomi was the term David Sirlin, the creator of the game “Yomi” coined to describe these levels of thinking, where “if my opponent does A, I should do B” is the first level, then the opponent countering B with C is the second, and on and on it goes until you are countering action AAAAAA with AAAAAB.

The tricky thing with the notion of Yomi is that, in circular relationships, no-one knows where to stop. When you have a that relationship, and when B counters A completely, there is no bounds to say “okay that’s enough”. If you consider that Yomi-level n+1 always beats Yomi-level n, then you can do more weird things with it. For an RPS system, assume Yomi(n) = Yomi(n+3) = Yomi(n-3).

If your opponent loses a game of RPS, you can assume they were operating at Yomi(n-1), to your Yomi(n). Now how do they react to a loss? They may subtract one to get Yomi(n-2) which would beat your Yomi(n) but, you you are a mindful opponent who also reacts. You drop to Yomi(n-1) to counter them. They anticipate this, dropping to Y(n-3) = Y(n). In this way you could assume subtracting 1 from your Yomi layer could be considered Meta-Yomi(n), and on and on it goes.

Failed Solutions

What I’m getting at here is, there’s really no strategic value to rock-paper-scissors systems between 2 mindful opponents. And I’m not bagging on David Sirlin here, he realised this just as I did when I was making Cowboys Are Tragic. And I think we both went about trying to resolve the innate issues of RPS-based games the same way, and unfortunately I’m not convinced that the solution we picked was enough.

I think we both tried mitigating this infinitely circular reasoning through trying to add enough additional factors that we considered, would at least add some alternate direction to the game. I had positionally advantageous/disadvantageous moves, and special character abilities. Yomi has decks unique to each character with their own special moves, knockdown mechanics, etc. We tried to incentivise certain moves, and disincentivise others, regardless of their position in the RPS cycle.

Ultimately though, when playing Yomi and my own game, none of those factors really matter so much as just trying deperately to be the player who wins the RPS. I feel that because of that hard-counter nature, you rarely want to play your best moves as they could be completely countered by an opponent — and particularly in Yomi I feel you cannot reliably get a good idea of what your opponent may play until you’ve worn down a large part of their deck (and even then… seriously, it’s a 54 card deck, with each card having 2 possible applications).

Here is a diagram:

So as you can see, you’re not really likely to be able to predict attacks, even towards the end. “Hold on!” you say, “you don’t need to predict each individual card though, only the nature of attacks!” True.

Here’s the “action” layout for one of the base characters, Jaina Stormborne. This considers every possible action she can take. As you can see, she’s guided more towards attacking.

So against her you should block you say? Sure — maybe — but again, the cards are dual purpose. If you try to apply a strategy based on her deck layout and trying to play the averages, a player can turn the deck around on you and make it look like so:


So a character’s deck layout doesn’t really give you all that much help when trying to predict their moves.

But what about over-all game strategy? I’ve read guides saying that because you have few cards early, Block is favourable early to bolster your hand. As such, its counter, Throw, is also popular. In the mid-game, apparently quick Attacks and Dodges are popular, as you have a large hand already and they can set up some nice damage. The same can be said for later game, but at that point the guides said that there’s a large emphasis on the higher power cards and stopping them from wiping you out of the game in one fell swoop.

Now, that information basically told me this:

But again, does this help me much? Does this help me know when to dodge and when to attack late? Does it tell me when to throw early round by round? No. Will I be able to see an opponent use an ability that seems like she’s going to ‘Throw’ next turn, and help me to make up my mind as to whether she’s bluffing or not? Again — no. There’s not enough information.

Yomi, as the term coined by Sirlin to discuss levels of prediction is the japanese word for “reading” — and is apparently a reference to “reading your opponent’s mind” — and at this point I’m basically thinking that yeah, it might as well be. I find it hard to believe that unless your opponent makes visible/audible signals as they play different kinds of cards, that you could successfully predict their actions with any accuracy, assuming that they are mindful of not repeating the same card sequences over and over. I would be very interested to see the percentages of rounds a highly skilled player becomes the “active player” when playing a moderately skilled player. It should be a larger percentage of the time due to their skill difference, right? But I can’t help to think they would still trade 50/50 (not counting null rounds).

In this way I would argue that “Yomi” is more about playing a game of RPS, and then playing the game based around who won that round, rather than, trying to win the game of RPS with prediction. In the end it seems largely about hand management, and not really about predictive strategy, despite the very name of the game.

Now, BattleCON.


Okay so I did a calculation of how many combinations there would be if EVERY round you had ALL of your cards. But then took it down to how many you have with the 2 discard piles (standard play mode). Note that to even get to BattleCON’s generic worst prediction scenario, in Yomi 20 cards need to have been discarded. But the thing is with BattleCON, most of those attack pairs make very little sense or are easy to discount as they are unfavourable for the current situation. I put in the “Favourable Category” — as you will generally find that of all the moves your opponent can do, only 2–4 will suit the current situation unless they want to throw you for a real doozy (and if they want to take that risk — all the best to them!). This 2–4 figure is based mainly on my own perception of the game. In Yomi, by comparison, every card in your hand is likely to be a valid option to play in a round, purely because each card could possible counter your opponent’s card. In BattleCON, you can categorise their moves and work out which options you have that are favourable or unfavourable.

When you choose a pair, there is still an element of chance. Maybe you didn’t consider an option. Maybe your opponent saw through your plan. Maybe you countered them fantasically… but the cognition behind the process not only allows for REAL predictions of what your opponent may do, but also every decision feels like your own. Every win feels like a win, unlike my Shootout game, and how I feel about Yomi. Similarly — even if you lose a beat, chances are some of your card effects will still happen, maybe you get to pick up a Token at the end of the round. You can play conservatively if you want, or you can aggressively trade and both strategies will work out with various characters.

It feels like strategy, rather than just making the Paper gesture and hoping your opponent didn’t choose Scissors.

Summing Up and Potential Solutions

Well, I feel that starting a game from a Rock Paper Scissors relationship is an extremely hard task if you want to make a game in which reading your opponent is part of the gameplay.

The nature of hard counters, and circular relationships feels like it completely kills strategy and prediction, rather than bringing it about.

As such, I would start prediction games from a different set of ideas — try to work out what factors can make players actually anticipate an opponent’s actions. I think the biggest part of this is actually making the actions fewer in number, but greater in how they differ from one another.

If you have too many actions, players cannot even analyse or plan around the vast possibilities. If the options are too similar, you approach “hard counter” territory as moves become “one size fits all”.

Hard counters should be avoided because they seem to dominate every other possible train of thought, and instead turn the game into “how will I not be countered?”. Either adding lots of interconnected ‘counter’ relationships and then building additional “prediction cues” on top of that system, or just softening counters such that players get net benefits/losses rather than “all-or-nothing” rounds would help.

Anyway. That’s that. And yes, in the end I did finish my essay. It was half the length of this post.