Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Biased Pandemic

54 Post author: freyley 13 March 2012 11:32PM

Recently, Portland Lesswrong played a game that was a perfect trifecta of: difficult mental exercise; fun; and an opportunity to learn about biases and recognize them in yourself and others. We're still perfecting it, and we'd welcome feedback, especially from people who try it.

The Short Version

The game is a combination of Pandemic, a cooperative board game that is cognitively demanding, and the idea of roleplaying cognitive biases. Our favorite way of playing it (so far), everyone selects a bias at random, and then attempts to exaggerate that bias in their arguments and decisions during the game. Everyone attempts to identify the biases in the other players, and, when a bias is guessed, the guessed player selects a new bias and begins again.

The Pieces

First, Pandemic. Pandemic is a cooperative game with a win condition and three lose conditions that are separate. It provides each player a list of available actions and then allows them 4 moves in which to mix and match those actions. Because of the combined win and lose conditions, players are constantly forced to decide between tactical and strategic objectives, with a strong emphasis on making it easy to choose tactically good moves at the expense of failing to win the game and thus losing by taking too long. Pandemic is a fun game, and it's a great game for people looking to stretch their brains. Obviously you want to be familiar with, preferably experienced at Pandemic before attempting Biased Pandemic. We did have one inexperienced player at Pandemic (out of 4) and that seemed to work okay, though it may have been harder for him than the rest of us.

Enter the biases. Each player selects a bias. We printed out the biases listed here and used them to select our biases. One of our players just made a TOC for selecting biases. There happen to be 104 biases listed in that document, so a deck of cards combined with a coin flip allowed for bias selection. A computer's random number generator, dice, or any other random method should suffice. Some biases may seem unplayable to some players -- certainly, the monetary biases seem unplayable to most of us -- but other players may find a way to play it, so we've refused to cross off biases and just allowed players to re-roll if they get a bias they're sure they can't play.

Examples

This can be a little difficult to wrap your brain around, so let me give a couple of examples. One player, playing the Negativity Bias, went around the board treating cities which had outbroken earlier in the game and ignoring other issues. Another player with Hyperbolic Discounting went further: he treated cities, any city near him, while carrying 5 red city cards in his hand and pointing out, in response to entreaties to cure red, that red wasn't much of an issue right now. A player with Reactance had the winning yellow card and simply refused to be told to go somewhere to give it to the player with the other four. He even went so far as to refuse a half a dozen offers of an airlift so he could give up that card. A player with Hindsight Bias claimed that he had predicted that the player with 1 red card would get two more on his next draw, and was upset that he'd let the other players argue him otherwise. A player with The Ultimate Attribution Error suggested that if we weren't doing well because no rationalist could ever win this game because we were terrible at it. A player with the Authority Bias attempted to suggest that we should do things because it's what Eliezer would want us to do. A player with Illusion of Control declared that his next draw, he simply would not draw an epidemic. There were many others.

Recommended Rules Of Play

We played it somewhat haphazardly the first game, but at the end we agreed on a structure for the next game that we think is better. In our next game we plan to have the order of play go like this: during each player's turn, all players can discuss what the player should do for a timed interval, perhaps 1-2 minutes. The player then declares their intended move. Now each other player gets an opportunity to make a single bias guess. If a guess is correct, the player stops playing the bias, and begins the round again. At the end of their turn, if their bias was guessed, they select a new bias. We considered a bias to be guessed correctly if the player guessing fully described the bias, not just the biased behavior. Bias names, however, were not required.

Notes

One way that you can not do well is by falling into the trap of making the same biased statements repeatedly. After a few rounds of this, the biased statements were pretty obvious. The guesses are an indicator of what the other players are seeing, and we went out of our way to look for ways to respond to the guesses by playing up the aspects of the bias that the other players weren't seeing. For example, a lot of the different biases look like simple overconfidence. One player was playing The Illusion of Control in such a way that the rest of us thought he was overconfident. His response was to start declaring that he simply wasn't going to draw an epidemic card, and when he drew one, he declared that it was my fault for making him draw the card. This was obviously not simply overconfidence.

Before playing, you should figure out how familiar you are with the biases. Players who are incredibly familiar with all of the biases may want to play a game where everyone plays as subtly as possible and your goal is to prevent other people from noticing your bias. For us, our goal was to learn the biases better and identify them in other people, so we tried to ham it up and play them as obviously as possible at first. It was incredibly difficult to specifically identify the biased thinking behind obviously biased statements, even with that, so I'd suggest at least trying it with obviousness first.

One of the most difficult things to remember is that your goal is not to win the Pandemic game. Sure, that's nice, but your real goal is to familiarize yourself with biases, and to have fun roieplaying and identifying biases. Losing Pandemic, especially because the players are following their biased thinking, is a totally acceptable outcome. We won, and do not credit our thinking for it.

We're looking forward to trying the game again, and maybe you'll have suggestions for improving it.

 

Comments (35)

Comment author: DanielLC 12 March 2012 05:55:41PM 33 points [-]

What you need to do is get really good at it so you can use biases subtly, and them make it so there's a certain chance that you get "No bias". That way, you'll find out what your real biases are.

Comment author: freyley 12 March 2012 07:01:19PM 2 points [-]

We hope to get there. It's going to take a while, I suspect.

Comment author: handoflixue 14 March 2012 08:00:58PM 0 points [-]

Very nice suggestion. Upvoted :)

Comment author: xv15 14 March 2012 05:12:53AM 11 points [-]

This sounds awesome. It would be really cool if you could configure it so that identifying biases actually helps you to win by some tangible measure. For example, if figuring out a bias just meant that person stopped playing with bias (instead of drawing a new bias), figuring out biases would be instrumental in winning. The parameters could be tweaked of course (if people typically figure out the biases quickly, you could make it so they redraw biases several times). Or you could link drawing additional biases with the drawing of epidemic cards?

I have this terrifying vision of a version where it is biases -- not diseases -- which spread throughout the world, and whenever a player's piece is in a city infected with a certain bias, they have to play with it...

Comment author: freyley 15 March 2012 12:34:26AM 3 points [-]

I think your terrifying vision sounds like a lot of fun.

Comment author: chaosmosis 12 April 2012 03:10:43PM 0 points [-]

A version of this game where you identify the biases of others but don't announce it would result in this type of competition. You can manipulate the other players more easily when you're aware of their biases.

Comment author: Kaj_Sotala 21 March 2012 11:02:11AM *  0 points [-]

I have this terrifying vision of a version where it is biases -- not diseases -- which spread throughout the world, and whenever a player's piece is in a city infected with a certain bias, they have to play with it..

Possible Arkham Horror variant: each increase of the Terror Track or Doom Counters infects a player with a bias. Closing a gate gives each player a single opportunity to guess the bias of a freely chosen other player, with a correct guess removing the bias. Sealing a gate additionally allows for the removal of a single player’s bias, even if nobody guesses it right.

Alternatively, just let everyone make a single guess in the Movement Phase, as per the normal Biased Pandemic rules.

Comment author: Vaniver 12 March 2012 05:09:50PM 5 points [-]

Hm. I'm curious how well this extends to other cooperative games like Arkham Horror.

(I have a low opinion of Pandemic as a game because it's a one-player game which four people play- which can work out okay if everyone has similar skill levels and tactical aptitudes, but poorly if you have players whose turns are just directed for them. It seems like a good substrate for this, though, since you've inserted a conflict that gets rid of the "play their turn for them" effect.)

Comment author: Kaj_Sotala 13 March 2012 06:59:42AM *  4 points [-]

That's a common issue with any co-operative game, Arkham Horror included. The only ways of avoiding it that I've seen are either having traitors and secret information (e.g. Battlestar Galactica), or introducing time limits and such a cognitive overload that you're lucky if you manage to play your own turn in time, let alone anyone else's (Space Alert).

Comment author: wedrifid 14 March 2012 04:02:28AM 0 points [-]

or introducing time limits and such a cognitive overload that you're lucky if you manage to play your own turn in time, let alone anyone else's (Space Alert).

Is this not likely to amplify the differences between players and so make it far more rational, given the seeking of a cooperative goal, to spend some time aiding allies? It is one thing to suggest tweaks for the allies to improve their play but another when the allies aren't even able to pick up the low hanging fruit!

Comment author: Kaj_Sotala 14 March 2012 06:36:11AM 5 points [-]

Yes, if some of the players are beginners, it's often a good idea for the more experienced player to take some time out of her turn to help them. But this means that she can't pay as much attention to her own actions.

The game is also built so that coordination between the players is of crucial importance. The way the game works is that there's ten minutes during which the players have to plan out all of their actions. The planning is done with placing cards from your hand on the turn track in front of you, e.g. I might play a card saying "move left" to the "turn 4" slot in front of me, and then my character will move left on the fourth turn. The information necessary for making those plans is parcelled out gradually during the ten-minute period. E.g. after four minutes of planning, it might be announced that an enemy vessel will appear on the left side of the players' space ship on the sixth turn. The players must then adjust their plans to accomodate for this, and make sure that around the sixth turn, somebody's character will be on the correct side of the ship to shoot down the enemy. After the ten minutes is over, the players' plans are executed and everyone gets to see what actually happened.

Now if I'm manning the guns on the right side of the space ship, say, and I tell another player to make sure that the guns have enough energy available so that I can shoot with them, then the other player must be up to his task. Otherwise I'll end up wasting half of my actions trying to shoot guns that don't have the power for it, and I won't find out about this until the execution phase, when it's too late to change any plans. On all difficulties but the easiest ones, the action planning cards are placed face down on the track, so I can't look at the other player's actions to make sure that he's doing what I told him to - he has to make sure of that himself. It's possible to have complicated chains of actions involving many players - e.g. "on turn 3, player A charges the main reactor, player B draws the energy from the main reactor to the auxiliary reactor on the left, player C uses that energy to fire the guns on the left, and player D uses that energy to recharge the shields on the left". If player A or player B fails to make his action just at the specified time, the whole chain of actions is ruined, as are any later plans that were presuming the success of that particular chain. Frequently, this is exactly what happens, with whole games having been lost because one player failed to commit a specific action on one single turn. (You can charge the reactors ahead of time, so that there won't be such split-second timing required... but then you still have to keep mental track of when they'll need to be recharged the next time. You can also schedule in redundant actions to make sure that something really is done, but this wastes time and possibly also energy, which is a limited resource.) Fortunately a single game doesn't take long, so there's always time for just one more rematch.

Ultimately, at least on the higher difficulties, each player has to learn to work as part of the team, do their share of the planning, and take care of their own actions. Otherwise the players are bound to lose.

Comment author: wedrifid 14 March 2012 08:38:04AM 1 point [-]

That game sounds like it has a lot of potential for FUN!

Comment author: Kaj_Sotala 14 March 2012 10:20:05AM 1 point [-]

I expect that if Dwarf Fortress players hadn't invented that phrase first, eventually Space Alert players would have. :-)

Comment author: wedrifid 14 March 2012 12:41:53PM 5 points [-]

I think I'd enjoy the challenge of trying to make plans that were robust with respect to handling some errors on the part of comrades. It sounds rather difficult!

Similarly, when I dallied in D&D I used to hold my character largely in reserve, keeping memorized spells in hand and holding back in the battles. A failsafe for when my companions failed - be that through distinctly suboptimal character creation or poor tactical decisions. While sacrificing some glory and the appearance of personally dominating the battles it maximised both safety and chance-of-saving-the-world. This also meant that the fight looked and felt more challenging to the players and DM while keeping the actual risk to the party lower than it may seem. This counters the pesky DM tendency to try to ramp up difficulty and do so incompetently, dooming the party to inevitable defeat.

It helps to choose a character that, when given a few rounds to prepare, could singlehandedly handle more than the rest of the party combined and then proceed to heal up all the wounded to satisfactory levels. I love my perfectly designed 3.5 druid!

I imagine with Space Alert at the highest difficulties winning despite allied incompetence may actually end up physically impossible. When it gets to the stage that even with perfect play every character is necessary. The best I could do would be to make sure I handle the most confusing or error prone tasks.

It would be perfect if the difficult tasks also happened to be the most bland - fiddling with power supplies and ammo rather than personally wreaking havoc upon the enemy at every turn. One of the most difficult elements of maximising winning is managing the egos of compatriots. It isn't uncommon for ego to go hand in hand with incompetence. If I can manage to maneuver the meat-heads of my party into situations that give them glory but are hard to screw up it makes my task of making the team win that much easier!

Comment author: wedrifid 14 March 2012 12:47:07PM *  3 points [-]

I expect that if Dwarf Fortress players hadn't invented that phrase first, eventually Space Alert players would have. :-)

On the other hand it sounds like Space Alert has an actual win condition, not just more and more dangerous ways to lose. The closest I've seen to someone 'winning' is when they went ahead and dug down to Hell itself and colonized it - building their residential areas in a location that is designed such that merely breaching it is supposed to spell inevitable doom. All to give their dwarves the chance to say "Tonight we dine in Hell!" - and proceed to do so for days on end.

Comment author: Will_Sawin 14 March 2012 12:37:04AM 0 points [-]

Barring communication usually works, though the cure is generally considered to be worse than the disease.

Comment author: roystgnr 14 March 2012 01:51:41AM 1 point [-]

What about just barring unsolicited communication? Forbidden Island works for my local mix of adults, old, and young children if it's given that restriction. If it isn't then the older and bossier players (yes, me) try to be too controlling.

Comment author: Eugine_Nier 14 March 2012 03:13:50AM 1 point [-]

It's still a problem if the one player is clearly smarter and the others are smart enough to know who it is.

Comment author: Will_Sawin 14 March 2012 06:55:58AM 0 points [-]

It seems to me that it is more about a certain personality type than intelligence. I know highly intelligent people who would not ask the smartest or most knowledgeable person what to do because they damn well want to figure it out for themselves.

Comment author: handoflixue 14 March 2012 07:48:05PM 0 points [-]

I think my local gaming group might need to steal this idea, hard as it will be to remain silent when others are making suboptimal decisions :)

Comment author: wedrifid 14 March 2012 04:09:43AM 0 points [-]

What about just barring unsolicited communication? Forbidden Island works for my local mix of adults, old, and young children if it's given that restriction. If it isn't then the older and bossier players (yes, me) try to be too controlling.

I like this idea. It doesn't eliminate the problem of making the game potentially too easy if there is one smart player. But it does eliminate the social problem. Asking for advice from others is an entirely different social interaction than having another give you orders!

Comment author: Darmani 13 March 2012 04:49:29AM 4 points [-]

With a sufficiently strong player, Arkham Horror is a one player game which seven people play.

Comment author: Vaniver 13 March 2012 03:10:24PM 1 point [-]

This is true- really it's true of any purely cooperative board game- and I've played it single player with several characters several times. For some reason the effect seems stronger with Pandemic.

Comment author: shokwave 13 March 2012 05:54:31AM 0 points [-]

I have experienced exactly this.

Comment author: freyley 14 March 2012 03:30:56AM 0 points [-]

I would imagine you can play it with any cooperative game. Another great one that wouldn't quite fall prey to the problem you describe is Scotland Yard, which has a group against a single player. The group could play with biases, while the single player plays without and tries to guess the biases. People have also suggested competitive games, such as Munchkin, but I'm skeptical so far. If anyone does play it with competitive games, I'd love to hear about it of course.

Comment author: Vaniver 14 March 2012 03:46:54AM 1 point [-]

I would imagine you can play it with any cooperative game.

The types of decisions players make, and the amount of information players have about each other matters. In Arkham Horror, for example, everything is public- in Pandemic, hands don't have to be (although I imagine you might want them to be for this, to make biases easier to spot). In Pandemic, there seems to be closer connection between actions and the endgame, whereas in Arkham there's an element of building up your character that can be independent to winning (for player motivation, at least).

Without some experience, I'm not willing to speculate on what features make a cooperative game work better or worse for roleplaying or noticing biases, but I am confident that some features will be better or worse.

Comment author: mapnoterritory 14 March 2012 08:47:59PM 2 points [-]

I'd recommend to post this variant to boardgamegeek.com to reach a wider audience. The "variants" forum for Pandemic is here. Kudos on the bias list, it's really well done!

Comment author: Voltairina 14 March 2012 10:09:07AM 2 points [-]

Implementing strategies based on incorporating biases sounds really interesting. Its a little bit as if there were a game of charades where a bunch of people sat around implementing a line of c code in assembly, and then people try to interpret what code is being implemented (only, to me, a bit more exciting). It really gets at the nuts and bolts, I think, of how everyone feels these biases work on the behavioral level, higher level descriptions aside.

Also, I'm a Portlander as well, I just joined the mailing list but didn't make it down to the lucky lab last time. I'm hoping to make it in the future at some point.

Comment author: Alicorn 12 March 2012 08:05:58AM 6 points [-]

This sounds awesome. I want to do this right now.

Comment author: freyley 12 March 2012 06:59:16PM 0 points [-]

Yay!

Comment author: handoflixue 14 March 2012 07:41:27PM *  0 points [-]

EDIT: Oh hey, found the meetup information myself. I've been gone too long >.>

Comment author: Peter_de_Blanc 13 March 2012 07:40:36PM 0 points [-]

Another player with Hyperbolic Discounting went further: he treated cities, any city near him, while carrying 5 red city cards in his hand and pointing out, in response to entreaties to cure red, that red wasn't much of an issue right now.

How does this demonstrate hyperbolic discounting?

Comment author: bentarm 14 March 2012 12:59:13AM *  4 points [-]

I don't know if you've played the game. There are 4 disease, red, blue, yellow and black. "Curing red" doesn't automatically eliminate the disease - it just makes it easier to deal with, and possible to eliminate in the future (and also is part of the win condition).

Treating people who have a disease right now helps them right now. Curing red has only future benefits.

I now realise you might be asking "how does this demonstrate hyperbolic, as opposed to exponential, discounting", which might be a valid point, but hyperbolic discounting does lead to discounting the future too heavily, so the player's choices do sort of make sense.

Comment author: Peter_de_Blanc 14 March 2012 09:19:21AM *  0 points [-]

I now realise you might be asking "how does this demonstrate hyperbolic, as opposed to exponential, discounting", which might be a valid point, but hyperbolic discounting does lead to discounting the future too heavily, so the player's choices do sort of make sense.

That is what I was wondering. Actually, exponential discounting values the (sufficiently distant) future less than hyperbolic discounting. Whether this is too heavy depends on the your parameter (unless you think that any discounting is bad).

Comment author: Rhwawn 13 March 2012 09:43:52PM 0 points [-]

Discounting with distance, I assume. Nearby people are extremely important, while it takes 100 African cities to get his attention, etc.