13x13 Game Against MFOG

[Ryan93277]

Shalom ~

Would someone kindly help me understand how I was able to win this game? I believe I was trying to stay connected. Does anyone have any ideas how I could have been more efficient?

Thanks,
Ryan

[schawipp]

Both players made a lot of strange moves in basic life and death situations:

- could have plainly captured K10
- K7 is more or less dead already. Better make sure that L10/M10 are dead, too.
- As black ignored the situation, white should now kill the k10 stones (just count the liberties!)
- - White finally managed to kill the k10 stones
- - it's hard to believe that sequence. Both players consistently ignore the snapback.
- etc. - white could still resolve the situation by simply finishing off the k10 stones (e.g. by playing L11).
- whut...? This was last chance for L11.

I recommend to practice life and death problems (Tsumego) and to play either a stronger bot level or a human (better imho), who will punish basic life and death mistakes more quickly.

BTW - welcome to Go - if it's one of your first games, many mistakes are quite usual (and are still usual even many years later at least if I speak for me...), don't let the overwhelming possibility of mistakes discourage you

[Ryan93277]

Thank you schawipp for your quick response! I appreciate your criticism.

May I ask a quick follow up question:

In response to move 61; would a better move be at M11 instead of K6?

[EdLee]

Hi Ryan,

The basic shape here is extend up, to E4, not to drop to the second line E2.

Same thing -- still E4 is the local shared vital point: both B and W want it.
Your first line hane, D1, is too early -- it's for the end game, not now.

Same thing. Still E4, not here.

Bad. Find a better local move.

The same mistake as -- Find a better local move.

This is a pass.

In response to move 61; would a better move be at M11 instead of K6?

Yes.

[Ryan93277]

Thank You EdLee for your comments.

[Mike Novack]

I recommend to practice life and death problems (Tsumego) and to play either a stronger bot level or a human (better imho), who will punish basic life and death mistakes more quickly.

If you are using a bot as opponent to improve your play. that is very important. Use a strength setting (for the bot) such that you usually lose (but SOMETIMES win) and as you get better (win as often as one game in three) bump the bot up a level.

Also -- you are suing MFOG, right? Well with this program only the top two levels are using the MCTS evaluator. Below that the program is using just the "go knowledge" AI. The problem with that is its weaknesses will be systematic. Sightly better as a learning tool to specify "random" (don't always make the same move in a given situation).

[Ryan93277]

Mike Novack You are right. I am using MFOG, but unfortunately it's older version. Thank you for your advice. I was wondering how would be best to play against this bot.

[oren]

Mike Novack You are right. I am using MFOG, but unfortunately it's older version. Thank you for your advice. I was wondering how would be best to play against this bot.

Best way is to play people.

[Bill Spight]

If you can set MFOG so that it gives you three stones on the 13x13 and you win about half the time, both of you will play better.



Main focus: Sacrifice a stone by throwing in to take away a liberty.

Even more important. Try to kill this opponent's stones. It is weak.

Good luck!

[Mike Novack]

Mike Novack You are right. I am using MFOG, but unfortunately it's older version. Thank you for your advice. I was wondering how would be best to play against this bot.

When you say "older" what do you mean? If you mean MFOG 11, all of its levels are go knowledge based AI. MFOG didn't use MCTS until MFOG 12.

If you meant an early (not updated) version of MFOG 12, why aren't you updating it. Fotland doesn't charge for that. Free download for that on the site.

If you own MFOG 11 but want to update to MFOG 12, contact him, since he gives (or at least used to) a reduced rate, about half the price on the site.

[Ryan93277]

Just a quick reply in response to my comment on an earlier version. I am using an old Windows XP machine and had an older copy MFOG 12 still and haven't gotten around to update it yet. Thank you everyone for your input in helping me getting stronger.

Shalom!

Ryan

[Mike Novack]

a) That your machine is running XP irrelevant. That your machine older depends on the CPU in it, older being a relative term. With a fairly early multiple core CPU should still be able to do about 1 dan if ......
b) You want to update to the latest version (1/23/2011)

Remember, with MFOG only the top two two levels are using a MCTS evaluator. Also, that you can't necessarily go by the strength indicated, since:
1) for the MCTS levels, that depends on processor power (MFOF uses time management, adjusts the numer of play outs according the processor power and the time control you choose)
2) Not easy to calibrate. Thus if the level marked 1 dan is playing at 1 dan (on your machine) the level marked 3 kyu might be more like 2 kyu (I got that from Fotland)