The Betrayers Banquet

Here is something for you to try out tomorrow with the family, well some families. The Betrayers’ Banquet is a dinner party/event that ingeniously combines the iterated prisoner’s dilemma with good food, bad food and entertainment. Here is their description:

The event works as follows:

A banqueting table is set with 48 chairs, 24 on each side, at which players are seated at random. For a period of two hours, the food is served in small portions every fifteen minutes, and varies in quality; at the top end of the table, it is exquisite – food you could expect at a fancy restaurant. At the bottom end, the food is charitably described as unpalatable. In between, it is a spectrum between these two extremes.

At regular intervals, pairs of opposing diners are invited to play a round of the prisoner’s dilemma with each other; They are each provided with a small wooden coin with symbols on each side representing cooperation and betrayal, which they place on the table concealed under their palms, and then simultaneously reveal:

  •  If they both cooperate, then they are both moved up five seats towards the good food.
  •  If they both betray, they are both moved five seats down towards the worse food.
  •  If one betrays and one cooperates, the betrayer moves up ten seats, and other down ten seats.

The event is presented as an initiation ritual of a freemason–esque secret society; service is run by servers in hooded robes and the game is arbitrated by a dour, unsympathetic master of ceremony, who punctuate the courses with grave speeches describing the discovery of the game in the court of Charlemagne in the eighth century.

From the participant’s point of view, aside from getting to play a game and try a variety of different foods, the main attraction is that they get to move around the table and talk to a variety of people throughout dinner. The iterated prisoner’s dilemma is famous for creating very complex social dynamics, which keeps conversation lively and generates a high eagerness to continue playing.


What happens to people who are fewer than ten seats away from the top or bottom? Or do they not play?

I assumed the pair chosen to play is decided at random and is not in a seat 10 away from either end. . .

. . . but that leaves only ~4 pairs left. I guess that could work.

Another angle is that if you both cooperate or both betray, you stick with each other; otherwise you are dispersed and will probably never encounter each other again (assuming there are not thousands of games in a night).

Hmmm .. what happens to the pair occupying the tables that this pair is being moved to? Do they move up or down? I assume the only way up is through playing the game so, if someone is displaced, it would have to be to the seat of the one who just played.

I assumed everyone in-between the new seat and the old slide one seat over to compensate.

Game designer speaking here: If you'd move off the end of the table, you move as far as you can.

Do the guests have to come dressed as their World of Warcraft characters or is that optional?

And if there's attractive women and an orgy involved, count me in! In like Flynn (effect)!

Average is truly over - but is it desirable that the servers are from the underclass, as compared to cadet branch family members?

And truly, the banquet scene in Delany's 'Stars In Pocket Like Grains Of Sand' is unbeatable - this sounds like a child's game for both the Sygn or the Family.

As of right now there are 34 tickets available with a price of £99 ($161.50) each. Price includes an eight course banquet, alcoholic drinks and a show.

You can bid whatever you want on a ticket, but you have to pay your bid, and only the top 34 bidders get a seat.

I'm not if I understand how conservation of rank is enforced. If every pair cooperates how can everyone move up five seats? Plus since you only move in multiples of five it might as well be five tables of ten.

Nice game with many interesing features.


1. I predict a boring game with just about 100% cooperation.
After all, all guests see the action of all the others. So there is a lot of reputation at stake:
Betraying might give me some advantage vs. one of the 48 players.
But then ALL 48 players know I am inclined to betray and will punish me.
So betraying has a 1/48 upside but 48/48 downside.
(This is different from "normal" turnbased prisoners games where there is only 1 other player. And even in "normal" turnbased prisoner games
cooperation would be my default strategy, unless the other sides keeps cheating).

2. the lively conversation part makes the game much more complex:
the guests aren't just playing for the quality of the food, but also for a specific company.
Assume for instance, that there is a nice lady I want to talk to 10 seats down the table.

24 slots on each sides, moving restricted to +/- 5 or 10.
I don't understand that numerical logic.
First of all, most players dont have a chance to be on top or bottom position.
Second: What happens near the end of the table? If I a am only 4 slots from the bottom, i dont have any downside risk, as i cant lose 5 ranks.
Third: The fact, that its only steps of 5 and 10 means, that the same players meet frequently. This might be a feature, but I would consider it a bug.
This can easily be solved by for instance 2/5 instead of 5/10.

if I am only 4 slots from the bottom, i dont have any downside risk, as i cant lose 5 ranks.

If people at the edges do play, I wonder if this would tend to keep them at the edges. If you are 2nd from last, you may be more likely to betray because the penalty is so small.

The game designer speaking here;
1. You'd be amazed. People are cutthroat, especially when they're drunk and hungry.
2. It's quite hard to aim that precisely. Between each turn, 23 other pairs get to have a go, so the table shifts quite a lot.
3. If you would go off the end of the table, you go as far as you can. Consider also - if you get 10 up and 10 down, 20 people change partners. There's a lot of mixing.

The business about seats is unnecessary and confusing.

You could just assign diners points, which they gain or lose in the game. More points gets you better food.

The point is to move people around so they can socialize with each other, not score points.

You don't win if you eat better food, you win if you maneuver your way to sitting next to the most popular people in the room.

Why the most popular?

And what if they are lousy players?

And of course you can always have people change seats randomly after every round. Might make it harder for the waiters, I guess.

Are the results any different if the participants are limited to old white male doctors and lawyers?

At the end of the table they serve treakle.

So, If you do not perform well, and move to the bad end of the table

It is known as

The Treakle Down Theory.

This game would be far more interesting using modulo arithmetic. For example, if a player was at position 2 and wanted to be closer to the top, he could plan to lose 10 spots, putting him at 16. In addition to interesting individual strategy, this would promote more movement at the top and the bottom.

With all this talk about the nonexistence of Knockout Game, it's time for a new holiday: Fistivus!

Why should the articulate be privileged in the Airing of Grievances? Simply merge the Airing of Grievances with the Feats of Strength and we have a 21st Century non-discriminatory holiday: Fistivus!

My knowledge of game theory is a little sketchy but isn't that rule set pretty much guaranteed to ensure people always betray.? If you betray you risk losing five places for the chance to gain 10 whereas co-operating stands to lose ten to gain only five. Why would you ever trust? I think the game might be more interesting with a third possible vote ("Doubt" perhaps) that trumps betrayal. You could get a Rock Paper Scissors thing going on with no obviously optimal strategy.

Have a read of this:'s_dilemma

Maybe the exercise was designed to focus the participants on changing the rules of the game, that is, they could all agree on a new rule for distribution and be better off.

I also think this illustrates this is not an evolutionarily stable game, and one that can only be played by people who ate before they came to the banquet.

Imagine this game being played by a bunch of hunter gathers who had to coordinate their hunting in order to catch a prey. One group acts cooperatively; another plays the banquet game later.

Which group survives and evolves.

I got annoyed about a third of the way through the post. As much fun as a hole in the head. Then again, I hate board games.

boy you economists really know how to have fun

Comments for this post are closed