Road to my 100 first go loss: I ask for some commentary

[globulon]

I would like to say that I have spent some time with this today and to let you know it may be more challenging than I at first thought. I still think it is a good example of some basic ideas and can give you some good reading practice if you don't get obsessed with finding the "right" answer. I would like to show some hints.

I will try to put them in spoiler tags so you can look at them when you are ready.

The first hint:

The main key that I wanted to convey was that white is already surrounded, so black can make moves that threaten the white group without risking much damage to his own stones. I still think that you should be able to see a couple of these kinds of moves with the hints I gave before (think of closing off your own lower left corner with a threat, and also keeping white divided on the lower side with a threat.

2nd hint:

My most basic variation. This is not meant as a variation that shows best play for white, but thinking up this kind of line can give you a starting point to work from.

small.jpg (20.36 KiB) Viewed 8384 times

[Mike Novack]

Playing against human opponents may be a good idea. But remember, at your current level (and theirs) a great many mistakes will be being made, lots of bad habits.

Considering the last game you showed us, I would say that of you wanted to continue learning using the computer as opponent, you need to put the program up a level of two. You should have the program playing strongly enough that you are losing say 2 out of three or even 3 pout of four. In other words, not so strong that you always lose (then you can't see when you aren't making some usual mistake, but not so weak that your usual mistakes aren't being punished.

Remember, take the "rank" of the computer program as shown with a grain of salt. By and large, the "go knowledge" based AI programs aren't aggressive enough, but play solidly. But some humans play like that too, especially at your level.

[Fllecha]

I would like to say that I have spent some time with this today and to let you know it may be more challenging than I at first thought. I still think it is a good example of some basic ideas and can give you some good reading practice if you don't get obsessed with finding the "right" answer.

I worked on that too and I found that actually white group can be killed. Maybe I missed some move too but I think there's no way out for white.

Skip at variation at for more details

Main problem was timing that forbid me to analyze deeply... From this moment I'd play with 10 minutes and 30 sec byo yomi
VARIATION AT

(the group un the end is killed because if at M2 then N3 and then the two hane at O1 an K1 kills definitively the group because L2 eye remains false.

[globulon]

Hey Fllecha,

Good try!

I have loaded some variations that I looked at and I am curious what you think.



I think the situation is complicated, I don't think black can necessarily kill cleanly. In a case like this, I think it is good to try to make moves that seal white in and break up the bottom.

I think I overestimated blacks chances here myself when I first reviewed it.

On the other hand. I think black can make some valuable moves harassing the white group anyways.

So I think what I want to say is just that the original 53 feels a little too defensive and I wanted to correct the mindset of your comment to his 54. I think those points still stand. I'm don't think you can flat out kill white (though I don't feel confident in my assessment, maybe one of the strong players here could). But you can still make some moves that try to create something with a spirit like the diagram I showed before. You might want to lead with the kick from the corner and make a placement, and then block on the left, then just focus on sealing white in. These will be forcing moves that white will probably need to reply to, so you will be able to make those moves and then go about your business on the rest of the board.

Anyway, I hope my flopping around is some help to think about a situation like this.

[Fllecha]

I'll analize globulon spot tomorrow, I found some interesting lines, and I admit that I didn't analized some interesting lines...

Meanwhile I moved to panda net server and I started to play humans (just 4 games) and I won 3 (easily with 30-50 points difference against 16+k) and lost 1 to a 6k that gave me 3 stones.

- Human plays VERY different from humans, you already said me that but now I understand how

- Humans have some bad habits on behavoir: I was winning by over 70 points and he started to randomly invade my secure territory even when it was clear that ther was no hope... at least he played quicly, I called myself lucky about that.

- I found that there is reasonable traffic on that site for western players

- People play more contact go: anytime they attach and try to combine some random stuff... Pretty important to improve on that, but usually games are more interesting.

So for this moment I'll publish only lost game played with humans: it takes only another two wins to get 17k and start the road to SDK Pretty interesting the ranking system imo on that site.