Tuesday, May 17, 2011

Randomness at game end

This weekend I played a card game called "You're Bluffing."  It is a game where the players take turns auctioning off cards and trying to buy cards from each other with secret bids.  The goal is to end up with full sets of cards from a deck with 10 different sets of 4.  The game is a lot of fun and has some real strategy to it - it is clear that better players will win the vast majority of the time but there is definitely some luck involved.  The strange part about the game design is that it is very predictable in the early game and there is a lot of strategy but the endgame almost always comes down to one person putting a stack of cards facedown on the table and one other person having to guess the value of the hidden cards in the stack; the result of this single guess will determine who wins the game.  The trouble is that by the time the endgame arrives it is very likely that there is no useful information anyone can bring to the table in terms of making the guess and both players are basically picking numbers randomly to determine the winner of the game.  Of course if you play very badly throughout the game you can be in a position where no amount of luck will give you victory at the end but it is abundantly clear that most games are decided in the closing moments and not before.

Argument 1:

This is a good thing.  There is nothing worse than being in a game for the long haul where you know who is going to win but much time yet needs to be spent confirming that fact.  If you are a competitive gamer who thrives on pushing the envelope of skill and strategy to claim victory it is terrible to play a game where you know you have lost in the early going.  Maintaining that ray of hope for every player involved (even if for some the ray is faint indeed) keeps the game enjoyable because everyone is intent on playing well and angling for victory.

Argument 2:

This is a terrible thing.  During the game you will make dozens of tiny, tricky decisions and there is nothing worse than getting to the end of a game where you feel you played spectacularly and then are forced to make a completely uninformed guess to determine the winner of the game.  Even if you are in an excellent position having outplayed your opponents for the whole game you still stand to lose it all based on a roll of the dice.  This leads to people hardly caring about most of their interim decisions because they know that someone will be crowned at the end and it might well be the person who screwed up all game.

So which argument is more compelling?  Of course it isn't as if everyone will agree either way but I wonder what most game players prefer.  Do they like having all the decisions in the game slowly add up to inexorable success for the person playing well or do they like to keep the game wide open even to the end?  Obviously most people dislike both of the extreme examples - playing a 6 player game for hours and then simply rolling a die to see who wins would feel entirely unsatisfying and playing out a game where the winner is determined halfway through and then much busywork must be completed before the game is 'done' is boring as hell.  Some games (Puerto Rico, for example) solve this by keeping the scores private so even though the game is completely decided the players don't yet know it.  This allows people to be involved right to the end even if they have no chance... though as playskill increases it becomes harder to obfuscate the scores like this.

What do you think?  Is it better for a game to remain an open question right to the end or is it better for it to be more deterministic from actions that took place well before the denouement?


  1. Something in the middle seems best, to me. It's no fun for the losing player if victory is assured for one player after the other makes a mistake, but it'd be awfully unsatisfying to have all one's work be for naught when the end is determined on what is essentially random chance. A good game, I feel, is one in which it is possible for a player to build momentum by making right decisions, and to noticeably be losing when making poor ones, but in which proper play can still upset the match.

  2. The length of the game is a big factor in this discussion. A short game like Can't Stop is still fun even if a player can be far ahead but lose to their opponent having a lucky turn. But a long game where a player screwed up for the first 75% of the game gets lucky and wins is frustrating and makes everyone think the first half of the game was pointless to play.

  3. I don't think PR solves the problem at all. Public private points aren't private at all, it just makes it take more mental strain to properly track. If you want the outcome to not be determined a couple actions before the end of the game you need actual private points like what they did to a Brief History of the World or Louis XIV.

    Personally I like games which are deterministic assuming the other players don't have too much control over how your game will progress later on. PR is a very good example of this. I think a game played with very good players should come pretty close to a tie every time because you too often get to choose who scores points. Once you progress beyond 'is craftsman good for me' to 'what does craftsman do to everyone if I call it compared to if the next guy calls it compared to if no one calls it' the game essentially becomes an exercise in who the other players want to have win.

    Power grid is the same way. Often on the last couple turns it will be obvious that someone is going to win for sure. Unless people collaborate to ensure they can't buy the raw resources they need to do so. But most of the time the person buying all this extra junk won't win either so they're just choosing who wins.

    Contrast to something like Le Havre which is deterministic to the extreme but other people just have a hard time stopping you from playing your game. They can squat in a building, sure, but beyond being very bad for them you still have outs with the whole final turn going anywhere thing. The actions the other people take certainly impact you, and they can potentially swing a game, but not to nearly the extent you can in PG or PR.

    Agricola is the same way with the advantage of having actual true hidden information in the cards you have yet to play. Some are worth big points and can be played quite late so you actually can't work out who will win for sure until the very end of the game. You also can't screw people nearly as much since you have to be able to use the actions you take. Sometimes there aren't enough renovations to go around and people do get screwed, of course, but you can generally avoid them if you plan far enough ahead.

  4. I think you are right Ziggyny in that PR doesn't solve the problem once the players are very good. Perhaps I should have phrased it as "tries to solve the problem" instead. For the vast majority of players the secret points system works pretty well but it breaks down pretty hard for the real professionals.

    The trouble with agricola is that it has the opposite problem to being deterministic - the winner can be very random. You cannot know if your opponents have cards in their cards that will net them a huge amount of points or not so you must simply guess and then see who wins. Clearly better players win the vast majority of the time but it does feel terrible when you make the best possible plays and lose to information you could not have obtained.

    I am biased because I love PR so much but certainly can see the advantages in other systems. The best solution as I see it is for everyone to just play FMB.

  5. I don't think losing to information you couldn't have had makes a game worse - think of poker. Of course you don't have the entire Agricola deck memorized, so it's a little different, but you could memorize it. That being said, I think that lack of information encourages you to play to do the best for yourself rather than to play to beat any other particular person, and I think that's a real strong point of Agricola.

    I'm not sure that it is preferable for a game to be determined by the end or the beginning inherently. I think the most important thing is that the you feel like the play of the game is consistent, and meaningfully linked to the outcome. If you know the game involves a strong chance element then losing to chance in the end seems fine, as long as that chance was consistent with the flow of the game (lose a big battle in which you are favoured to end a game of Titan due to poor dice rolls). If you know that the early decisions have a huge impact on the end of the game then that feels okay because you can think about what you should have done better next time (as long as the early decisions aren't simple enough to make the game trivial).

    I think the worst case scenarios are when the majority of the game is extremely tactical and then it comes down to luck in the end or when early luck can give an insurmountable advantage. Though I don't know the game you are discussing, it sounds like it might have this problem. Despite the fact that the gameplay remains the same, it gives you the impression of being very strategic early on and very random at the end.

  6. Yeah, it isn't that losing at game end to a big reveal is bad... it is that if a game isn't focused around that it is bad. Poker is entirely about trying to puzzle out what your opponent has based on social (rather than game) cues. You never know for sure what he holds but that is the essence of the game. Agricola is about resource management, predicting other player's strategies and counting but has a 'big reveal' moment in the final couple turns that can completely change everything. I like the game overall but I dislike the way the cards work for this and other reasons.

    "Your're Bluffing" is mostly tactical with a bit of randomness and guesswork throughout. By playing very well you can certainly set yourself up to be in an excellent position at the end and it would be very unlikely for you to lose... but even an excellent player can be flat out ruined by a random bad guess on the last turn. The trouble is that that random bad guess can occur during an interaction between two other players! I actually have some ideas for how to fix it but understanding them would require full game knowledge. I will probably pick the game up since it is cheap, small and quick to learn and teach people so I can figure out if they agree with my ideas.