Page 9 of 9

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 6:50 am
by Boidhre
Annihilist:

Seki (I believe)

Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . 6 X O 3 . . . .
$$ | 5 1 2 X O . O . .
$$ | 4 O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]


Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . 5 X O 3 . . . .
$$ | 6 1 2 X O . O . .
$$ | 4 O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 8:36 am
by xed_over
Boidhre wrote:
Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . . X O 3 . . . .
$$ | . . . X O . @ . .
$$ | . O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]




why does everyone keep wanting to play :w3: ?
that's what the marked stone is for, so you don't have to play :w3:

its not seki. white can kill.

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 8:48 am
by Bill Spight
Click Here To Show Diagram Code
[go]$$B Black to play.
$$ ------------------------
$$ . . . . X . . O X X . . .
$$ . O . O X . . X O . O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


The first play was obvious. ;)

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X C 3 O X X . . .
$$ . O . O X 1 . X O . O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


:b1: springs to mind because it threatens :b3:, making two eyes. We do not even think about branches from here.

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X . 2 O X X . . .
$$ . O . O X 1 . X O . O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


:w2: is forced. This position looks hopeless. Even "logically." ;) Why?

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X . O O X X . . .
$$ . O . O X X 3 X O . O O .
$$ . . . O X W W X O . . . .
$$ . . . O X W X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


The 3 stone White string is worth 1 eye, "logically". If :b3: captures, . . .

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X 5 W W X X . . .
$$ . O . O X X X X O . O O .
$$ . . . O X 4 . X O . . . .
$$ . . . O X . X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


:w4: is nakade for 1 eye. Now the 2 stone White string is worth 0 eye, "logically". If :b5: captures, . . .

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X X . 6 X X . . .
$$ . O . O X X X X O . O O .
$$ . . . O X O . X O . . . .
$$ . . . O X . X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


:w6: throws in for 0 eye. "Logically," 1 + 0 = 1, which is less than 2, so Black is dead.

The road not taken, the line not read:

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X . 1 O X X . . .
$$ . O . O X 2 . X O . O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


:b1: was not even considered, because it leads to this position, which is known to be dead.

That is a strange thing to say. If you do not even consider :b1:, how do you see this position? ;)

Be that as it may, I don't think that I considered :b1:.

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X . 2 O X X . . .
$$ . O . O X . 1 X O . O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


Next, what about this atari? Note that I stick with the same :w2:. That's not a bad heuristic. If :w2: countered other plays, maybe it will counter this one, too. And it does. I am still left with 1 eye plus 0 eye. Black is dead. (I did not actually add 1 + 0. I just know the eye value of each string, and that they do not add up.)

OK, time for the stupid play:

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X 1 2 O X X . . .
$$ . O . O X . . X O . O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


The light dawns! I already know that capturing does not work. 1 + 0 and all that. But I do not have to add. :)

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X 1 2 O X X . . .
$$ . O . O X 3 5 X O 4 O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


:b3: combines the two positions, so that they are not independent. :) That means that the eye values do not just add. :b5: makes a simultaneous capture, yielding this:

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X X . . X X . . .
$$ . O . O X X X X O O O O .
$$ . . . O X . . X O . . . .
$$ . . . O X . X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


The three point eye has an eye value of 1.5, and the two point half-false eye has an eye value of 0.5. 1.5 + 0.5 = 2, logically. Black is alive!

Now I know that :w2: in this line is incorrect. (Without reading? ;)) I have to transpose :b1: and :b3:.

Click Here To Show Diagram Code
[go]$$B
$$ ------------------------
$$ . . . . X 3 2 O X X . . .
$$ . O . O X 1 5 X O 4 O O .
$$ . . . O X O O X O . . . .
$$ . . . O X O X X O . O . .
$$ . . . O X X X . O . . . .
$$ . . . O O O O O . . . . .
$$ . . . . . . . . . . . . .
$$ . . . . . . . . . . . . .[/go]


Bingo! :)

Now I remember running across a simultaneous capture problem long ago. Too bad I had forgotten. The logic of simultaneous capture would have made this problem easy. It's magic. It's as though 1 + 0 = 2. ;)

If I had been a depth first kind of guy, I would have solved this problem sooner. As you can see, I am not a breadth first kind of guy, either. I call what I do sideways search. ;)

Note the value of assuming :w2:. A line of play starting with two incorrect plays led to the solution. ;) You can see why I say that seeing is better than reading. :)

After some thought, I classified myself as a logical player. Even though I play by feel. Still, this problem was not easy for me. That's because of a flaw in my logic. The eye values did not add up. (Not that I added them consciously. ;)) The key was to combine the positions without adding. :) And that I discovered by reading an illogical line of play. ;)

I don't know if there is a moral to this story, but I thought it was worth sharing. :)

Edit: Afterthought. I found the heuristic of considering the same :w2: valuable. But that was once I was into reading mode. Actually, except for the solution and the line I did not consider, :w2: could have been a pass. But by then I had dropped out of evaluation mode. I was in calculation mode. If I had evaluated the positions correctly, I would have had to backtrack, instead of transposing. But it all worked out. ;)

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 9:19 am
by Boidhre
xed_over wrote:
Boidhre wrote:
Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . . X O 3 . . . .
$$ | . . . X O . @ . .
$$ | . O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]




why does everyone keep wanting to play :w3: ?
that's what the marked stone is for, so you don't have to play :w3:

its not seki. white can kill.


I didn't want to play :w3: I was just trying to show the result of his sequence for white moves 1 and 3, i.e. that it wasn't a killing sequence for white.

Edit: I didn't want to show him the answer as it'd spoil the fun for him. I can see how my post was unclear though. :)

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 10:31 am
by xed_over
Boidhre wrote:I was just trying to show the result of his sequence for white moves 1 and 3, i.e. that it wasn't a killing sequence for white.

ah, I misread your response to his response. sorry.

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 10:43 am
by Boidhre
xed_over wrote:
Boidhre wrote:I was just trying to show the result of his sequence for white moves 1 and 3, i.e. that it wasn't a killing sequence for white.

ah, I misread your response to his response. sorry.


No problem. :)

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 3:04 pm
by Annihilist
Boidhre wrote:Annihilist:

Seki (I believe)

Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . 6 X O 3 . . . .
$$ | 5 1 2 X O . O . .
$$ | 4 O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]


Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . 5 X O 3 . . . .
$$ | 6 1 2 X O . O . .
$$ | 4 O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]
:w3: at :w5: instead.

Pole connector through the middle should kill it.


xed_over wrote:
Boidhre wrote:
Click Here To Show Diagram Code
[go]$$W White to play and kill.
$$ +------------------
$$ | . . X O 3 . . . .
$$ | . . . X O . @ . .
$$ | . O . X O . . . .
$$ | X X X X O . . . .
$$ | . O O O O . . . .
$$ | . . . . . . . . .
$$ | . O . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]




why does everyone keep wanting to play :w3: ?
that's what the marked stone is for, so you don't have to play :w3:

its not seki. white can kill.
I honestly don't remember. I think I may have seen a semeai coming so I decided to restrict liberties.

Re: Logical players, intuitive players ..

Posted: Thu Nov 01, 2012 7:19 pm
by Kirby
Wow, what a long thread. I don't know how I missed it. I have many thoughts on this topic, but I'll just comment on Monte Carlo go, which was brought up several posts, ago.

Namely, I don't feel that monte carlo is a detour to go AI at all. In fact, humans often make decisions based on some degree of confidence in what is true. Every fact that you "know" to make a decision still contains uncertainty.

A lot of cool things can be done with computers, banking on this uncertainty, and it'd be arrogant to say that humans don't do the same.

Re: Logical players, intuitive players ..

Posted: Sat Dec 01, 2012 3:19 pm
by DeFlow
Monte Carlo simulation was a part of the finance & risk-courses on my university. As far as I know, with respect to go, Monte Carlo simulating relies on simulating many (e.g. 10^9+) courses of play and choosing the one with the highest expected profit.

Human go decisions differ from person to person, but I think it is reasonable to state the following basic aspects:
1) Positional judgement (Balance of power and balance of territory (as by 'Attack and Defence')
2) Weigh the urgency of different areas against eachother
3) Define candidate moves for every local fight/struggle (based on shape and/or brute-force reading)
4) Weigh merits of candidate local moves against eachother

I am confident to say that all these steps occur in some form of another in every player, but in every player other aspects are more explicit than others, and betweem players the overall explicitness varies a lot. More explicit players can be called 'logical', less explicit players 'intuitive'.

If we try to match Monte Carlo-simulation to any of a humans aspects of go decision-making, all steps can be performed with support of a Monte Carlo approach, but the human mind has many shortcuts over this brute-force approach. These shortcuts will be implemented in an efficient go AI.

Robert is so good at making all these shortcuts and heuristics explicit, maybe he should help and build a go AI?

Re: Logical players, intuitive players ..

Posted: Sat Dec 01, 2012 3:27 pm
by RobertJasiek
If I had the time to do it:)