Tic Tac Toe Coding Challenge

[chilly_charly]

Since players with random components are precious for evalution, here is an another one. It's only an optimized version of the CCref player, with a tuned evaluation function. Should be as fast (means faster than Altenbach player...;)), but I probably made some mistake while converting it to LV7.1. I'll check that later.

On average, it scores 162 over 200 against the original CCref. On Bruce scale, I think it should rank first.

I felt it would not be fair to hide it from the community. So now you have some time left before the 12th to improve your own player.

Enjoy !

[Kevin_Price]

---

@Bruce Ammons wrote:

I will say that for a majority of the matches, it was 100% draw (or win/loss).  Only the players with random components made a difference.  If I narrowed it down to the top players and reran the scoring, it would most likely be 100% draw and I would have to go with a tie breaker.  I prefer the current scoring since I don't have to do a tie breaker.  I feel the ability to play against the players with random components is an adequate score differentiator.

---

Since I was the one who suggested the re-run among the top players, I'll just chime in and say that I agree with Bruce's comments above.  If they're pretty much 100% draws against one another, the ability to dominate weak/random play already IS a good differentiator.

-Kevin P.

[ien_fr]

Hi all,

I send here a software wich you can use to challenge other player algorithms

Have Fun

[BruceAmmons]

Everybody,

Remember there are only a couple of days left to submit your solutions.  The deadline is midnight Friday!  You can email them to me (tictactoe@ammonsengineering.com) or post them here.

Thanks,

Bruce

[altenbach]

Ahh these stupid deadlines! Why does it always have to be Friday night, robbing us of a nice relaxed weekend of tinkering? 😉

I don't have time to do anything beyond of what I already have, so I just need to package it up and send it to you. Basically the core of "Deepest Toe" with a few tweaks will probably be my entry.

[BruceAmmons]

I would be fine extending the deadline to Sunday night.  Everybody gets one last weekend to tinker!

Bruce

[Daklu]

I agree Altenbach, deadlines are a pain. (Who invented calendars anyway...?) Unfortunately, my deadline isn't imposed by Bruce, but by the fact that my demo license ends today.

Toe Jam is stable, but like all software projects, remains incomplete. I'll pack up my final tweak and send it off.

('Tis a pity. I had such high hopes for 'Toe Nail.' And the icons would have been sooo cooool.)

[Kevin_Price]

Darn, I didn't find any time to tinker AND I forgot about the deadline.  Oh well, here's the KP2b I would have submitted.  On each play it looks at all remaining cells and computes a score for each one.  The score is based solely on how that play affects possible 4-in-a-rows.  Putting myself closer is bad, blocking an opponent who's close is slightly more bad.  And the closer, the badder.  Also, affecting multiple 4-in-a-rows is worse than just one.  Finally, pick the "least bad" play, or randomly from all those that tied for least bad.

Ok, here it is...

-Kevin P.

Message Edited by Kevin Price on 07-18-2006 01:11 AM

[shoneill]

Kevin,

Sounds almost exactly like what I would have implemented if I had gotten around to it.  Unfortunately my involvement didn't get past my Ultraconservative VI which was more a proof of concept than anything else.  Still, I did think about it, and your approach was one which stayed with me.  Good luck.

  I'm looking forward (as a neutral observer) to seeing the results and the strategies used.

Shane.

[chilly_charly]

Kevin,

That's exactly what my ref players do. No wonder that they were on a par when playing against each other. I guess a majority of submissions have adopted a similar strategy. I'm waiting eagerly to see how Christian's player, whith it's fully deterministic approach, is doing against all the "advanced" random players.