Semi-cooperative board games and the win/loss binary

In Twilight Struggle, if you cause nuclear war, your opponent wins and you lose. Twilight Struggle is a two-player strategy game that simulates The Cold War. As you know, in the real world, if there is a nuclear war then everyone loses. But in Twilight Struggle, nuclear war leads to one winner and one loser. This speaks to limitations in what a strategy board game can effectively simulate.

Twilight Struggle is simulating a semi-cooperative situation, which means it combines cooperative and competitive elements. A semi-cooperative game is one that allows one player to get ahead of the other, but also allows outcomes which are good for both players or bad for both players.  Note that ties don’t count, because they aren’t good or bad for both players!  A semi-cooperative game requires at least three distinct outcomes, outside of ties. In Twilight Struggle, the three outcomes are USA wins, the Soviet Union wins, or there is nuclear war. This is challenging to adapt to a board game format, because players are accustomed to only two non-tie outcomes: winning or losing.

Under the standard rules of Twilight Struggle, neither player wants to press the big button. If you’re behind in the game, you might think you’re losing anyway, so might as well go out with a bang. However, there’s always a chance that you might catch up if you play on; meanwhile there’s no recovering from nuclear war. So the three outcomes get mapped to win/loss conditions like so:

Ahead -> Likely win, opponent likely loses
Behind -> Likely loss, opponent likely wins
Nuclear war -> Guaranteed loss, opponent guaranteed to win

Nuclear war is clearly the worst choice, successfully simulating the real world.

Now suppose we create a variant of Twilight Struggle where both players lose to nuclear war. So our outcomes are now mapped as follows:

Ahead -> Likely win, opponent likely loses
Behind -> Likely loss, opponent likely wins
Nuclear war -> Guaranteed loss, opponent guaranteed to lose

The problem is that some players may look at that third option and say, actually it looks better than the second option. If I can’t win, then neither can you. Call this “spite”. Not all players will be spiteful, but if one of them is, it completely breaks the game. Nobody would ever be able to win against a spiteful player, because it’s trivial for a player to cause nuclear war if that’s what they set out to do. Another possible result is that players would try to stay away from nuclear war as much as possible, lest they give their opponent a chance to force a loss. But if players do that, it also breaks the game, and fails to simulate the brinksmanship of The Cold War.

So that’s why Twilight Struggle is designed that way, with one player “winning” nuclear war. Essentially, we use probability to simulate 3 distinct outcomes, while remaining within the win/loss binary.

Interestingly, the loser of nuclear war isn’t always the person who caused the nuclear war–technically the player whose turn it is loses. If you play the Olympic Games card, your opponent can sometimes force a nuclear war by boycotting the Olympics. But since it’s your turn not theirs, you lose and they win. This arguably doesn’t simulate the real world, where pinning the blame on your opponent doesn’t exactly help you survive nuclear war. But it’s an interesting little loophole that catches players off-guard, and I suppose that makes the threat of nuclear war more real.

What are other methods can we use to design semi-cooperative strategy games? I will describe a few approaches that I have seen.

1. Hidden identity games

Hidden identity games like Avalon or Battlestar Galactica have traitors who are hidden amongst the players. If you are not a traitor, you want to cooperate with other team members, but be cautious about trusting potential traitors. In effect, this is also making use of probability, because the identity of the traitor is randomly chosen at the beginning of the game.

2. King-making

King-making refers to a situation where two players stand to win, but a third player gets to choose which one. In my circles, we talk about games with “politics”, which primarily refers to king-making mechanics. I generally don’t like board games with a lot of politics, because I don’t particularly enjoy the experience of playing favorites among friends friends, or having friends play favorites with me–but some people like that stuff I guess. (I’m not making a point about whether games are political or not, “politics” is just what we call it!)

King-making may lead to semi-cooperative gameplay. For example, in Carcassonne players can share points by building cities, farms, and roads together. If you cooperate with one player, both of you benefit–you’re using your king-making decisions to boost them, and they’re doing the same for you.  So generally you cooperate with lots of people (although fully prepared to stab them in the back).  But ultimately only one player wins, so if you believe that a certain player is ahead, you might stop cooperating with them.

In effect, this is also making use of probability. Since only one player wins in the end, when all is said and done the cooperation either helped you or hurt you or did nothing at all. But during the game, when the outcome is uncertain, cooperation can be mutually beneficial, because it increases both players’ probability of winning, compared to other players not in on the deal.

Or at least, that’s how I think about it. I recognize that not every board gamer has the same analytical approach that I do. Some players make decisions based on favoritism, trust, grudges, or spite.

3. Iterated board games

In card games like Hearts, it’s common for players to play multiple rounds, and add together the points from each round. During each round, you get a number of points, and you want to minimize the number of points. This creates a hierarchy of outcomes with a clear preference ordering:

26 < 25 < 24 < … < 2 < 1 < 0

Ultimately only one player wins, but getting a lower score is still desirable. The lower the score the better, since it sets you up to win in a future round. In other words, lower scores increase your probability of winning.

Hearts isn’t the best example of a semi-cooperative game, but it does have a tiny bit of cooperation. For instance, if one player tries to “shoot the moon”, other players might cooperate to stop them because if one player shoots the moon everybody else gets the maximum number of points.

4. Speculative approaches

In all the above examples, I argued that what really enables semi-cooperative play is probability of winning/losing. Interestingly, these approaches mostly only work with 3+ players, and Twilight Struggle is a very rare example that manages with 2 players. But is there a non-probabilistic approach? What’s to stop a game from just declaring the players should have a particular preference ordering among 3+ distinct outcomes?

Strategy games already have a few common tools to break away from the win/loss binary. For instance, players may intrinsically value points above and beyond their utility in winning. For example, maybe it’s the last round of the game and I already know I’m winning. But I want to win BIG, so I push around cards and meeples trying to get as many victory points. In short, I have the following preference ordering:

Lose big < Barely lose <<< Barely win < Win big

However, some players might ignore the scores beyond their utility in winning. In other words they’re truly indifferent to winning big vs barely winning–they’re both just winning. I’ve definitely watched players who, when they find a way to win, they get exactly one more point than their opponent and promptly end the game.

Also, many players only care about score differences, not absolute score. After all, reducing your opponents’ score is an equally valid path to victory as increasing your own score. But if players only care about the differences in scores, that’s not sufficient to support semi-cooperative gameplay, because the outcomes are all zero sum.

In games with 3+ players, some players may care about rank placement. For example, even if I’m sure I’m not winning, I might try to fight for 2nd place. This raises a philosophical question. If you have a choice between X% chance of getting 1st place (getting 3rd place if you lose), and having a guaranteed 2nd place spot, which do you prefer? In a competitive tournament where only the 1st place winner matters, it is correct to prefer any probability of getting 1st place, but that’s not necessarily how players play in practice!

Overall, I think there are many ways that board games could break out of the win/loss binary to make a semi-cooperative experience. I’m sure commenters will tell me about other approaches, as well as other games which put those approaches into practice. However, I think a general problem is that board gamers may not respond consistently to the range of possible outcomes. And, of course, it’s not necessarily bad that gamers have different approaches to board games, but if the designer has a specific semi-cooperative experience in mind, they may require more consistency.

Ultimately I don’t think there’s any fundamental barrier to stepping outside of the win/loss binary to create semi-cooperative games. The win/loss binary is just a player expectation, and it’s not even universal across gaming. For instance, I don’t think table top RPGs or video games always have the same norms about the win/loss binary. And maybe that’s because other kinds of games are more open-ended experiences, while strategy games are more rigid. It’s also a reflection of my own rigidity–I’m not sure I would enjoy a strongly semi-cooperative board game if I had it in hand! I just like the idea, really.

(Related article: my discussion of cooperative board games from 2017.)


  1. suttkus says

    Thought one:

    Have you considered all-versus-one as a category of semi-cooperative game? (Fury of Dracula, or Nuns on the Run)

    Thought two:

    An interesting example of a game that isn’t, technically, semi-cooperative, but feels like one, is Mistborn: House War. In the game, players take the part of the various political factions from the books (which I haven’t read) that are part of an empire. As the game progresses, players vie to gain the Emperor’s favor, by attempting to keeping the Empire stable. If the game ends with the Empire intact, the player with the most imperial favor wins. But, if the Empire falls apart, then the player with the least Imperial favor wins. Since player’s favor scores are kept hidden, this feels a bit like a hidden traitor game, but one where players themselves determine when they are going to become traitors rather than having it assigned. I need to play this more to decide if it really works.

    Thought three:

    I’ve often been frustrated with the aspect of semi-cooperative games you bring up above. It’s gotten to the point where hearing that a game is semi-cooperative makes me start disliking it before I’ve even played it. There are a few SC games I’ve enjoyed, but the odds have been low. And I’ve spent a lot of time thinking about two games in particular and why one of them works and one doesn’t.

    Castle Panic is a semi-cooperative tower defense game. There’s a tower at the center of the board, and monsters emerge from the edges to cross to the center and destroy elements of the tower. If the last tower element falls, the player’s lose. If the tower is still standing when the seemingly endless horde finally ends, the player with the most points worth of killed monsters wins. (Note that I only have the first edition of this game, so if the second edition has changed anything, I don’t know about it.)

    Doctor Who, Time of the Daleks is a semi-cooperative dice game using the core mechanic borrowed from Elder Sign. The players take on various incarnations of The Doctor and attempt to stop an attack force of Daleks from reaching Galifrey, solving lots of other problems on the way. If the Daleks reach Galifrey first, all players lose. If one of the Doctors reaches it first, that player wins.

    Castle Panic has been a success in every play group I’ve tried it in. The Doctor Who game has not. (And my wait for a decent Doctor Who game continues, *sigh*)

    But what’s the difference? It’s not the mechanics. As I said, the Doctor Who game borrows a lot from Elder Sign, which has been a favorite game in several of my play groups. It’s the semi-cooperative nature of the Doctor Who game that ruins it. Players can’t get over the “well, if I’m not going to win, NOBODY IS” problem.

    This doesn’t happen in Castle Panic. Most of my play groups don’t even bother counting monster points at the end, content that having saved the tower is a victory for all involved, essentially rendering the game fully cooperative. And even the groups that do like counting up points to get a single winner still tend to feel that losing while having saved the tower is better than losing because the tower fell.

    So, what’s the difference?

    It’s all about presentation, I think. The Doctor Who game has an obvious race track. Players and Daleks move markers along the track and the one who gets to the end first wins. This setup encourages players to be in a competitive mode from from the start. Castle Panic, on the other hand, presents a complete tower and protective walls. The players do not get individual tokens, they’re all represented by the singular set of tower elements. Every time a tower element falls to the monsters, everyone feels like they just lost something, in a way that the Dalek marker taking a step forward simply cannot convey.

    Or that’s been my experience. BGG ratings have the Doctor Who game being better than Castle Panic so I might be in the minority on this. Or there’s just a bunch of Doctor Who fans willing to upscore the closest we’ve ever gotten to a decent DW boardgame.

  2. Ketil Tveiten says

    Some games with win conditions different from «one player wins» that I find interesting:

    New Angeles is a game where players take the role of cyberpunk megacorps running a city, trying to skim profits while not letting their mismanagement lead to a complete collapse. Everyone competes, but have some shared interests. The win condition is that at the start of the game, you draw a card with a player color on, and you win if you beat that player in score, and you don’t care about anyone else’s score. The game itself is so-so, but I wish that win condition would show up elsewhere.

    Nemesis is a survival game inspired by movies like Alien. The players are stock sci-fi characters on a space ship filled with dangerous aliens, and they need to survive by a) making sure the ship is working, b) making sure the ship is headed back to Earth and not out into deep space, and c) safely getting into the hibernation pods without being eaten. The catch is each player has an individual objective they must complete in order to win, such as «get an alien egg» or «send a signal from the transmission room», or even «player 2 can not survive». It is in general quite hard to succeed, so the players are strongly incentivised to help eachother out, but the individual objectives guarantee some friction. Nemesis is an excellent game everyone should try, strong recommendation. The «win if you do X» win condition that allows multiple winners is something I wish I saw more often, as it completely eliminates any of the «I can’t win so I’ll screw someone» kingmaking that tends to happen in one-winner games.

  3. says


    I wasn’t counting all v 1 or competitive team games in the category. Although they technically have competitive and cooperative elements, they don’t really mix. You have a clear cooperative relationship with some players and competitive relationship with others.

    I played Castle Panic like once, and I don’t think I even realized there was a single winner. It’s interesting how it resolves the spite problem by hiding or deemphasizing the competitive component.

    @Ketil Tveiten,
    When I wrote about cooperative games years ago, some people mentioned Dead of Winter as a semi-cooperative game. It’s mostly cooperative, but each player has a secret objective. Alien sounds like a similar idea.

    I’m not sure how these games address the spite problem. Like, if you don’t see a path to achieving your secret objective, do you instead switch to trying to sink the team? Maybe it’s again a matter of presentation, like with Castle Panic.

  4. Ketil Tveiten says

    Spite in Nemesis: it’s not uncommon to hear something like «no, you guys ain’t leaving yet, you will help me get my thing, or we all die». Largely though, the game is hard enough that by the time you realize you probably won’t make it, you’re also not likely to be able to sabotage the ship, so spiteful wrecking is not a real concern.

    Regardless, the game is so good at creating situations that would fit right into a movie like Alien(s) that no one I’ve ever played it with would be particularly upset by a dramatic backstab like that.

  5. says

    I see, so the potential for spite is keeping in the spirit of the game! That’s definitely *not* true of Twilight Struggle, which is why it takes mechanical measures to discourage spite.

  6. Cutty Snark says

    I must confess a great deal of ignorance when it comes to Dead of Winter, so perhaps this is completely wrong, but I seem to have the impression that the character and the player are alittle bit decoupled – for example, you the character might be exiled but you the player draw a new character with new objectives. If that is the case, your spite might end up being directed against future you…

    Note: this is based off of a half-remembered Shut up and Sit Down review, so I could have got it completely wrong…

  7. JM says

    Betrayal at House on the Hill is a variation of options 1. Part way through the game one of the players is made a traitor. It’s made clear who the traitor is but up until that point it’s a fully cooperative game. At that point it becomes everybody against the traitor. People who are not the traitor work together to escape or defeat the traitor. In some cases a player can sacrifice another player to insure they escape but the game is designed such that it is almost always better to help each other, if only to give the traitor more targets. Have not played enough to know how well it works really.

  8. says

    Also, Betrayal at House on the Hill doesn’t always have a traitor.

    Along similar lines, I once played a game (Rune Age) which had a cooperative mode–but there was a chance of a late game event, which would enable players to achieve a lone victory. So that was another form of traitor, but it would work even with two players because it wasn’t guaranteed to come up. Of course, the mere threat of the event would cause competition earlier on, so I think we actually removed the card from the game, lol.

  9. suttkus says

    Dead of Winter is a hidden traitor game at it’s core. Except that there might not be a traitor, and the personal goal cards make everyone act like they might be the traitor at least a little.

    Each player is given a personal goal card, which may or may not require that everyone else loses (making them the traitor), but will always require the player to do something at least a little against the group’s best interests (I will horde food instead of sharing it, for example).

    But each player controls multiple characters, who will act as a group to accomplish that player’s personal goals. Consider them a clique, I guess. Your group succeeds or fails, gets exiled, or whatever, together, so the player is the unit, not the characters. It sort of hurts my immersion in the game, but I do like it.

  10. says

    I don’t enough about Twilight Struggle to comment on that specifically, but if I were in a game where the player who is definitely losing could make us both lose, I would see that as a reason to keep out positions close enough that my opponent felt they had a chance to win.

  11. Callinectes says

    The trick with Castle Panic is that it plays like a cooperative videogame that scores points, where you win or lose as a group, but get to see the scores at the end. Arranged this way points don’t feel like winning, they feel like mere bragging rights, or simple efficiency data to inform future strategies.

Leave a Reply

Your email address will not be published. Required fields are marked *