Dan Percival on Fri, 3 Aug 2012 16:22:55 -0700 (MST)

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [game-lang] Late to the party


That's quite an overview of the computer-Mao problem!

I'm curious if you have an example in mind of a game in which intentional rule-breaking formally causes a loss while unintentional rule-breaking doesn't. In a large part, the rules of games "just are," and enforcing them is outside the scope of the game. (A counter-example I can think of is Diplomacy, which specifically states that rule violations are allowed but must be rolled back once discovered, if possible.) GDL implicitly uses this formulation, such that state transitions that break the rules are impossible. This is reasonable(!), but worth noting.

On Fri, Aug 3, 2012 at 1:48 AM, Simon McGregor <londonien@xxxxxxxxx> wrote:
If we are willing to see this rules-hiding aspect of Mao as part of
the game (which seems reasonable to me), then we should conclude that
an implicit part of other games is that players are given declarative
information about the consequences of their moves before the game

In a sense, GGP does this with the game start "move" that delivers the game description, the player role, and the time parameters. For Mao, I imagine that the description would be limited to the facts of the game system: a deck of cards, a circular arrangement of players, player hands, the goal of emptying your hand, and the actions of putting a card on the play pile, drawing a card, handing a card to another player while citing their offence, returning a misapplied penalty, and privately communicating a new rule with a chosen witness (presumably in GDL or some other DSL) -- does that cover it?

It might be possible to introduce the basic rules of Mao into this framework by example, via two "teacher" bots which agree with each other on the basic rules of Mao (meeting the "penalties must be seconded" rule). Still, the game would be vulnerable to catastrophic revolution, with every player giving penalty cards to every other player at every time-step. Is there something other than deference to authority and conflict avoidance which prevents this from happening in human Mao?

One further interesting aspect of Mao is that, if players never
discuss the rules explicitly, there is a built-in potential for
memetic mutation. If even experienced players only learn the rules by
inductive reasoning; it's possible for them to infer slightly
different rules from the ones their teachers had, and pass these rules
on to newer players. This ties in to work on the origins of language
and social customs.

This is one of the things that fascinates me about Mao and Nomic as AI testbeds. Without making any strong statements about AI here, I think you could see some interesting emergent effects with current machine learning techniques.

game-lang mailing list