Game Idea: Just repeat ANTS as is.

[chainti]

Indeed ANTS has such simple but powerful rules, that it is just hard to invent something better.
So I suggest just to repeat AI Ants Contest (with some minor changes) on regular basis until someone find really great new idea.

The only thing in ANTS I suggest to get rid of is food blocking rule (food appear blocks ant move). This rule along with trampling rule (when two ants go to the same tile both died) lead to uncertainty and unpredictable danger when you want to move a line of ants. These two rules just conflict with each other, so I suggest that food appearing should not block move of ant (food just should be automatically gathered or doesn't appear at all at this place).

So I strongly suggest not to invent something strange (with more complex rules), and just repeat ANTS on regular basis (every month champ). I believe that it could take ages to find best algorithm for this game.

Regards,
Anton C.

[SlashV]

Excellent Idea.
Personally, I didn't really have time to finish my bot, so it would be really great to get another shot. Also, I agree that the rules of the game provide sufficient opportunity for improvement of most existing bots, so there should still be a challenge for creators of already successful bots.
If it turns out that creators of already successful bots don't see a challenge in this, I would suggest alteration of the rules, that would require different strategies, but with basically the same game idea.

[simplicity]

I think repeating Ants is a bad idea. Firstly, for newcomers it's extremely bad as you guys have been thinking about Ants way more than me. So a select few have a massive advantage over others. I remember reading the winner of Ants was a beta tester.

I suppose if you make Ants, but it's 3D it could work.

[chainti]

[simplicity]

[NorbertFindig]

Just to make sure everyone knows: It was an open beta, everyone could have downloaded the starter packages and start buildin their tools and upload their bots. It was in no way unfair to anybody. I did it too, played around a bit and didn't automatically win, so xathis must have done something right.

Additionally I guess, everyone did his inventions in multiple iterations seeing what the other bots are doing, this levels out any "head start" over the time of contest in my opionion.

Besides from that: I'd love to finish my Ants bot. I was missing about two weeks (my daughter was too early ) and I really want to see my bot compete again. I had a wonderful discovery and path building algorithm (well, as far as I remember, ...).

Best Regards
Norbert

[dabino]

Both Ants ans Planetwars are equally "underexplored", and can be used for contest. And winners' source code is available, so a lot of newcomers can catch up.
However, as far as I understand, key issue now is not lack of game idea, but some copyright/ownership things, which are unresolved between organizers.

[delt0r]

I am not a organizer but are in contact with them at least a bit. The issues have not been copyright anything but time of the guy sort of "in charge" when it comes to the money or some such things. The good news is that this could be resolved where he will have a quite a bit more time to put into the challenge.

On topic. We don't like repeating games for the reasons already stated above. Also i thought planet wars was a better game than ants. Some liked Tron better of course. But the idea is that each time its new. There are quite a few ai/bot challenges that stick to the same game.

[smiley1983]

If anyone wants an ongoing Ants contest enough to pay for a server, there is TCP server code here:

https://github.com/berak/ants-tcp

Relatively easy to set up (very easy for those familiar with Linux and ssh). I had one running from late March, discontinued around May due to lack of use.

You could also run a contest server with this code:

https://github.com/aichallenge/aichallenge

The setup is harder, though, and it insists that you run a specific obsolete Ubuntu (11.04 from memory).

[chainti]