A won game and a lost game.

[tekesta]

Here I have two games. In my latest one I played Black against mfgo12kyu1 and won by 28.5 pts. Before that, I played Black against GnuGoBot38 and lost by 37.5 pts. More than anything I would like to know if I have any bad play habits to which I ought to pay attention. I am about 12k on KGS at the moment. If I want to get to SDK level I will have to weed out any bad habits in my play.

Here is my latest KGS game against mfgo12kyu1.



And my last game against GnuGoBot38

[EdLee]

[EdLee]

[tekesta]

Thanks for the corrections, EdLee. I often descend to the first line in an attempt to establish territory. Looks like I often let my anxiety get the better of me, especially in games against stronger opponents. Perhaps with more hours with the life & death and tesuji books I can feel a touch more confident. Also, recently I tend to play a rather "slow" development game. (I wonder if I'm trying too hard to be like Shusaku ) I have forgotten what constitutes "fast" and "slow" play in Go.

In a few days I'll post up another KGS game.

[EdLee]

I often descend to the first line in an attempt to establish territory.

You're welcome -- we must decide case by case whether to just connect, descend, make a tiger's mouth (2 ways!),
or even to hane on the first line. Many beginners drop to the first line without thinking,
and that becomes a bad habit that has to be un-learned later.

...my anxiety get the better of me, especially in games against stronger opponents.

Correct. Different people advocate playing the board versus playing the opponent. I suggest play the board. YMMV.

[Mike Novack]

Here I have two games. In my latest one I played Black against mfgo12kyu1 and won by 28.5 pts.

"mfgo12kyu1 [-]: GTP Engine for mfgo12kyu1 (white): Many Faces of Go version 12 (2002, 2008, 2010 world champion), playing at level 12 kyu. Uses Many Faces' knowledge combined with Monte Carlo evaluation."

You didn't say what version. Only the initial release of MFOG 12 was 1 kyu at its highest level. MFOG 12.022 or even somewhat earlier would be 1 dan on "standard hardware". Since I have this program I know that only the 1 dan and 3* kyu levels (the top two levels) use Monte Carlo evaluation with lower levels using the AI. If you have it set to the 12 kyu level that's just the go knowledge based AI both picking plausible moves and evaluating which is the best of the lot, not using MCTS for the evaluation.



* Miscalibrated -- actually probably more like 2 kyu

[oren]

You didn't say what version.

He's playing the account on kgs with that name. The version doesn't matter for the advice he's looking for.