Suggestion: randomized starting positions

[brazzy]

I propose that the starting position of both bots should be random.

Currently, multiple matches between bots with deterministic behaviour proceed identically (except for who gets the first move) and are thus a waste of CPU time. The intention of making results more meaningful by increasing the sample size is not achieved; one bot may still win 100% of the matches (because it's really just a repetition of 2 matches) even though it would lose 90% with randomized starting positions. It's even possible to optimize bot behaviour to specifically handling the fixed starting positions, which I don't think is what the game should be about.

The only downside I can see is that randomized starting positions could favor one bot - but this effect would even out over multiple matched, which is their point.

[adum]

i've thought a lot about this over time. i think in the end, you're probably right: the game would be more interesting with randomized positions. although, i will say that it's still very much unsolved for fixed positions.

by the way, i think that the 4-player worm game is the most unsolved of all three bot games on hacker. there's a ton of room there for interesting development.

[Simoff]

I guess you should set a fixed number of moves before the bots have to use random values;
I think 3 moves would be fine, so that the bot programmer can choice accordingly to his bot personality, thus making simulations more unpredictable, but still letting the best win.

[Nick-Aotmzgin]

the permanent starting spots are quite fine and i think its better to leave it at is default..

[gfoot]

I got the impression that one of the players has an advantage (one starts closer to the centre). Is that right?

[adum]

one starts closer to the center, yes. but whether that is an advantage or not, i don't think we can say yet.