AI Tournament #1 in action

All the AI has been randomly grouped in pairs. And the first round will be as follows:



Every pair will play in ALL the 5 maps selected for this contest. As you can see, we have something unexpected: we will have 7 winners from this first round and we will need always an even number of contestants. In order to fix that, all the 7 winners will be put on a 7 slots map to play FAA, and the player with the best score will pass directly to the third round.

Now with the other 6 AIs, we'll begin the second round, playing in the same way as we'll do in the first round. After that, we will get the 4 best AIs, which will play a third round to define the first, second, third and fourth position in this Tournament.

I'll be posting the replays with the respective maps.


Downloads:

• Introductory map #1 (the first 7 AI in a FFA game)
• Introductory map #2 (the last 7 AI in a FFA game)
• AI files.
  
  
• Preliminary thread

Wohoo! lol ^^ Got to see those fights right now. ^^

14 contestants... not bad for the 1st A.I tourney.

lol, i destroyed kill em all in tests

@moyack:
One thing that needs accounting for. What happens if certain ai's draw by either both getting stuck or running out or both running out of rescources. I would guess its best to just take the highest score of the game to determine the winner.

Also, some AI's do better against some strategies and races than others...

You are right SM, in the case that one match get stuck, I'll use the scores to decide the winner.

It's for that reason that you had one month to do the AI, that time was for hard testing against any environment condition with any race.

So when can we see first replays of the contest?

Oy.. After testing my new AI versus an old .wai file, it was kind of dissapointing to see it lose. Hopefully I'll win one round in the tourny, but if anything, Ill be looking forward to AI Tournament #2.

I think it would be more fair if you didn't have elimination, but rather a each-ai-plays-against-each. As it is now, an AI that might otherwise take second place could be eliminated by the first place AI (or some really lame AI that just happens to be good against it) in the first round already.

QFT

Oh, and are you starting the AIs up all at once or with a short wait in between them? Without the wait, sometimes AIs refuse to budge and just sit there and make a few buildings and then stop. For some reason, adding like a 1 sec wait in between each one helps.

Yea, that was a wimpy move to beat on other AIs while not posting your own. Of course some of us posted our AIs for the public, so it's fine to test against those, but you really don't need to brag about having the best AI.

I do agree with Anitarf's suggestion, but I have no idea how much longer it would take to match everyone against everyone. If it's not much of a problem to do, then please consider doing it that way.

There's "only" 91 possible combinations (14*13/2), that would mean 455 games played. It's a lot but if you had more than just one person doing the games, it would be doable in a week or two.

Yes, it will be the most fair way to do the things, but unfortunately I will do the evaluation quite long, and it's time consuming like hell.

I think the elimination to define the 6 contestants for the second round is fair enough because each of this AI pair will play in 5 different maps, so the best of the best will be selected.

Anyways, if I can get more help from the moderators to release the replays, it will be great. Could you give me a hand Anitarf??

doesn't 91 combinations == 91 games?