My first game against Wakaba

[tekesta]

I just played the Gobot "Wakaba" on KGS. It is a Gobt powered by GnuGo 3.8 and rated around 6k. It did not give me too much of a challenge, but I ended up losing by 4½ points. Even though I have been working on my reading in the opening and early middle game, it is still suboptimal and I think that falling behind at this stage of the game is what often causes me to lose against stronger opponents, whether human or robot. (In any case, I'll concede that my skills in every category need improvement )

Below is the game. Much obliged in advance for your kind commentary.

[Boidhre]

is reason enough not to play this bot.

[RBerenguel]

is reason enough not to play this bot.

I liked 17, though.

[Mike Novack]

is reason enough not to play this bot.

A question though. Do you consider that a mistake in direction atypical of a 6 kyu human?

By and large the "expert system" AI programs (those using some go knowledge based evaluator to pick the best move from a set of candidates) tend to play too tight and defensively. But that is not atypical of some humans at that level.

Meanwhile the programs using MCTS for the evaluator tend to play a "loose" style depending on accumulated aji all over the board to support a later counterattack. Again, some humans play that way. In particular, the imbalance between "strategy" and "tactics" of the MCTS programs isn't atypical of "strong at a very early age" humans (look at childhood games of people who later became giants of the go world).

So what can we conclude from this? While playing against the computer programs can be very useful (my opinion) they do not prepare you well for:
1) At the lower levels (not MCTS) playing against a human of your own level or stronger whose style is loose, aggressive.
2) At the higher levels (MCTS) playing against a human of your own level or stronger whose style is tight.

You still need to play against humans. The problem is, not always easy to find a human 2-3 stones above you willing to give you those games.

[Boidhre]

is reason enough not to play this bot.

I liked 17, though.

Yeah, what I don't like is is saying that needs a reply on that board. It doesn't, at all.

[Boidhre]

is reason enough not to play this bot.

A question though. Do you consider that a mistake in direction atypical of a 6 kyu human?

By and large the "expert system" AI programs (those using some go knowledge based evaluator to pick the best move from a set of candidates) tend to play too tight and defensively. But that is not atypical of some humans at that level.

Meanwhile the programs using MCTS for the evaluator tend to play a "loose" style depending on accumulated aji all over the board to support a later counterattack. Again, some humans play that way. In particular, the imbalance between "strategy" and "tactics" of the MCTS programs isn't atypical of "strong at a very early age" humans (look at childhood games of people who later became giants of the go world).

So what can we conclude from this? While playing against the computer programs can be very useful (my opinion) they do not prepare you well for:
1) At the lower levels (not MCTS) playing against a human of your own level or stronger whose style is loose, aggressive.
2) At the higher levels (MCTS) playing against a human of your own level or stronger whose style is tight.

You still need to play against humans. The problem is, not always easy to find a human 2-3 stones above you willing to give you those games.

A 6 kyu human? I don't know, I haven't seen a mistake like that since I was a ddk playing other ddks and perhaps very weak sdks. Most of my regular sdk opponents (around 5-2k) would never make that move. I would believe that it's a mistake a KGS 6k could make though, I just wouldn't consider it typical.

[Shaddy]

I didn't even think 11 was much of a mistake.

[Boidhre]

I didn't even think 11 was much of a mistake.

*shrugs*

Enclosing looks a lot bigger to me and with the corner open I'd not be worried about the invasion so much.

[Bill Spight]

It is no easy matter to reply correctly to Wakaba's bad moves.

[Bill Spight]

I didn't even think 11 was much of a mistake.

*shrugs*

Enclosing looks a lot bigger to me and with the corner open I'd not be worried about the invasion so much.

Just think of as honte.

Seriously, it threatens to invade or reduce the White moyo on the left side.

[Bill Spight]

is reason enough not to play this bot.

I liked 17, though.

sucks, though.

Black must, repeat, must crosscut. No reading required. If he can't crosscut he should not play .