Tic Tac Toe Coding Challenge

[tst]

---

@Ben wrote:

Ask about the tall one in the back.

---

O.K.

Is the tall one in the back Mirco?
And who is Mirco, actually? A son? A nephew? A brother?

P.S. I like what I heard so far.

P.P.S. Bruce, sorry for hijacking the thread, but I haven't had time to examine the VIs and I can't really understand anything from the posts, so it's really rather boring.

[altenbach]

---

@tst wrote:
And who is Mirco, actually? A son? A nephew? A brother?

---

(He's my youngest son. :))

Now, to stay on topic here, I have to convince them to write a LabVIEW and tic toc toe based song. 😄

Refrain:
tic.. the FOR loop spins
tac... the data flows
toe.. you loose!

[tst]

---

altenbach wrote:

(He's my youngest son. :))

---

Yes, the hair is a dead giveaway .

[chilly_charly]

Seems that the excitement is decreasing. Are the thread and challenge already dead ?

Just to entertain you, here is a new random player. Bruce's was pretty dumb ;). I mean it never hesitated to align 4 spots. This one avoid suicide until forced to. I learned a lot from it.

[Ray.R]

From my end it is always the same thing.... lack of time...  😐

---work--work--work--work--work--work---  ---sigh--- 😞

I even read the postings @ every 2nd line..  YIKES!!  (ok, it shows!)

[Kevin_Price]

Re: Shane's "Ultraconservative", never-lose-as-O player.

I was doing just a little bit of playing around with a test program that allowed me to change players in the middle of a game, or start the board off in a pre-set pattern.  My basic goal was to detect scenarios that led to my own player (KP2a) getting forced into 4-in-a-Row, and to trace through the play history to try to figure out which turn led me astray.  Then I'd also put the other players in that scenario and observe them trying to avoid a forced 4-in-a-Row.

So anyway, there were some fairly early board scenarios that made Shane's "Ultracon" player do very poorly as O.   So poorly that it was forced into 4-in-a-row EVERY TIME, which was even worse than the Random player.  On the other hand, when Ultracon played the entire match, I never saw it forced into 4-in-a-row.  Strategic lesson: staking out a good early position can be crucial to an algorithm's success.

BTW, I noticed that I made a mistake implementing KP2a.  The actual results of my evaluation function weren't what I had intended on paper.  Turns out that what I intended performs considerably worse!   A bunch of other little tweaks I've tried that seemed intuitively like good ideas have also turned out worse.  So much for Murphy's Law...

-Kevin P.

[chilly_charly]

Hi Kevin,

So you found that problem solving is a good way to compare the players ?.. Have you tried the Ring, the Checker board and the others initial configurations ? I feel that's the best way to make a distinction between dumb players (the ones that never lose and never win) and smart players (the ones that take every opportunity to win a game).

The CC ref player was designed to be very aggressive. Actually it's the only player I have developped that is able to win some runs against KP-2a ! I feel that if you play to win, then you have to take some risk : To win you have to avoid adding a piece on a row already occupied by the opponent. So in turn, you are obliged to add one to one of your own rows... So it's difficult to be both a good defender and a cute assaillant.

[Ben]

I have been listeining in as y'all have been talking.

I have been holding back for a couple of reasons.

1) My boss said he may take on this challenge. I will be helping him (cheat) if he asks.

2) If he does not do it, I have a plan that I will submit (provided my schedule permits) that depends on being able to "practice" against experts.

3) We still have two more months!

Ben

[altenbach]

---

@chilly charly wrote:

Seems that the excitement is decreasing. Are the thread and challenge already dead ?

---

It is very well alive. I have a couple of very interesting things in the oven but at the moment I am busy with real work.

Still, I am very proud to say that I have fully analyzed the 0x0, 1x1 and 2x2 boards. 😄

[shoneill]

Kevin,

I'm not surprised that my "UC" version falls completely flat on its face when it hasn't made all the moves itself.

If you look at the board during a match, surely you must be able to notice the strategy, it's really quite simple.

I agree with CC, it's hard to optimize the attack AND the defence.  Hence the coding challenge.  My O-player version doesn't bother optimizing anything as "O", it produces a draw each and every time, guaranteed, irrespective of the opponent.  That is, as long as it is allowed to start from a board with only one "X".

I also don't have much time to code a proper entry.  It would be a shame for this to be my only entry, but maybe I'll find Time somewhere (Maybe I should look behind the sofa!)

Shane.