JoachimSchipper comments on SotW: Check Consequentialism - Less Wrong

38 Post author: Eliezer_Yudkowsky 29 March 2012 01:35AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (311)

You are viewing a single comment's thread. Show more comments above.

Comment author: Matt_Simpson 24 March 2012 01:40:48AM 2 points [-]

It occurs to me that games with some significant strategic component might be useful for priming the "but what consequences does it have?" response. I'm thinking of games like Magic: the Gathering, Settlers of Catan, Risk, etc. (I'm sure the board game aficionados will have better examples than I). I say this because of personal experience with Magic players - as they get better at magic, they tend to get better at life. Well, some of them do. The others perhaps compartmentalize too much, so maybe this won't help with everyone.

In any case, my model for what would work is a relatively easy social game that allows a non-trivial number of actions with unclear consequences... unless you stop to think about them. Magic would be perfect... if it wasn't so complicated and if fantasy tropes didn't turn off a large segment of the population. Ideally the game would be something you create instead of something your subjects/clients may have played before.

I have no ideas for the actual game, but maybe this sparks someone else's imagination.

Comment author: JoachimSchipper 29 March 2012 07:23:21AM *  0 points [-]

Careful, there: some vindictiveness ("if you attack me in Africa despite our pact, I will go totally apeshit on you for the rest of the game") is an essential part of playing e.g. Risk well (in our group) - naive consequentialism ("looks like I lost Africa, taking Australia from (unrelated player) seems best now") does not work very well on intelligent and adversarial agents.

Of course, most of the world is not an intelligent and adversarial agent - pre-committing to going totally apeshit on an unthinking animal is just stupid. The easiest and biggest wins for consequentialism are there, not in games of Risk.

(Non-naive consequentialism works fine. Naive consequentialism probably works fine in many games, e.g. two-player games like Magic.)

Comment author: wedrifid 29 March 2012 12:18:07PM 1 point [-]

Careful, there: some vindictiveness ("if you attack me in Africa despite our pact, I will go totally apeshit on you for the rest of the game") is an essential part of playing e.g. Risk well (in our group) - naive consequentialism ("looks like I lost Africa, taking Australia from (unrelated player) seems best now") does not work very well on intelligent and adversarial agents.

Totally agree. I'm ruthlessly vindictive but perfectly trustworthy (meaning I refrain from making promises I do not keep) when it comes to strategic situations like that. It looks superficially like being completely unsophisticated but it works.