Posts:
2,233
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente - Understand vs Memorize
Posted:
Oct 27, 2010, 10:25 AM
im just wondering if every one disregarded my suggestion with out even looking at it for the sake of proving they could find their own way... im referring to the move suggestions for the puzzle. i have yet to see any one talk about those moves at all. interesting.
and sure jason, i have no problem with you showing your moves, and i will even go so far as to comment on them.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Re: Pente - Understand vs Memorize
Posted:
Oct 28, 2010, 4:48 AM
OK, here are 2 possible ways it could play out for black. Both start with up2ng's suggested opening moves ("G12, G11, then G8 and J8"); except, I went G10 instead and forgot to try J8 till the time of posting this message. The 2nd one adds nosovs' suggested modification to move 27.
Zoey, I tried your suggested moves first and spent hours looking for a win for black. I was convinced black could not win in this puzzle until I explored up2ng's suggestion. Now I'm not so sure.
Posts:
2,233
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente - Understand vs Memorize
Posted:
Oct 28, 2010, 11:03 AM
in option 1; seems good as long as white doesnt play M11 on the 27th.
option 2; this is the more interesting path to look at. im not sure that K14 on the 28th is whites best play here. this should be examined more perhaps.
M14, Q10, Q10, P11, S14*
make this suggestion on a board if you would, so others can see the white solution to your second option.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Pente - Understand vs Memorize
Posted:
Oct 28, 2010, 7:58 PM
First of all, I think it's clear that black cannot just block after white makes that move of 27. M11* -- black must capture, probably to O9*. I was thinking that even before Zoey's new path for white which I didn't notice myself, but I was looking at another little wrinkle for white which involves playing to J8 and then H7, looking for a win by captures. In fact, if you play what I'm suggesting here immediately after Zoey's sequence, white wins easily.
Try some more scenarios after black instead recaptures at 27 ... O9*
I have not looked in much detail and I'm not sure that black has a win -- initiative has certainly been stalled if white has any VCF paths from this position.
Since Zoey mentioned it, I'm going to take a quick look at his suggested moves. But let's not lose the point of this thread. The suggestion to this puzzle is not the important part of the discussion -- it was simply an illustration of a complex position and how a player might proceed with some thought process for which is the "best" move to play next. I was assuming that his solution worked since he designed the puzzle. But the point is to find a few moves that "might" work well, try to think ahead in your mind to see the most likely responses by your opponent, and then at some point you need to decide on a "best" move, sometimes with incomplete information (not being able to see far enough ahead).
It should be clear from this example that as the complexity of the game increases, humans have more trouble identifying a correct solution without making an error. If it turns out that the path I was looking at first does not work and if I was unable to see Zoey's suggested path during the pressure of a timed game, then I might have gone ahead with my idea, which may turn out to be a "mistake", a deviation from perfect play, due to high complexity combined with time pressure. Successful strategies for black will often look to create messy situations like this to increase the chance for white to make a fatal mistake.
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Pente - Understand vs Memorize
Posted:
Oct 28, 2010, 9:04 PM
Ok, I took a look at Zoey's suggested moves and I currently believe that they fail. Hopefully I'm wrong and then I'll look a bit more. Full description below:
------------------------------------
The suggested moves, black to L10 and then to M6, is a clever choice. White appears to have strong initiative, creating a tria which cannot be captured across. Most of black's possibilities to respond to this tria with consecutive tesseras all appear to lack continuation (still looking into black to H13?).
However, black finds a way to use a soft block combined with a counter-tria which renders white's tria a false threat. Meaning, white cannot make an open ended tessera for an immediate win -- the soft block prevents this on one side, and on the other side the tessera would play into a capture that would yield a winning tessera for black! So, for now, white's threat has been countered and black steals the initiative.
When black makes this tria, it is a threat that must be dealt with. There are only a few ways to do this. Play a sequence of tesseras for a VCF, capture across the tria, block the tria, or look for something clever and outside the box as black has just done.
Based on this, the only possible moves for white are: M10 Q10 R10 J8 K8 N5 L5 P9
Eliminate R10 and J8 which lead to clear and immediate losses.
M10: This looks like a great move at first glance. You are forcing black to play into a keystone capture, then setting up another keystone capture at S14, creating a 4x3 for victory. Even if the stones in that corner were not available, there might be continuation at N13 for a 3rd consecutive keystone capture and then looking to win by captures. The problem is that because black played M6, forcing black to now block with a keystone at Q10 accidentally allows black to create it's own gapped tessera as a counter-attack. Now, instead of white continuing forward with consecutive force moves, it is now forced to respond and block at P9. Furthermore, when M10 was first played, black likely captures to L11, creating a second tria -- the resulting 4x3 is a victory for black.
Q10: This is the open tessera that is countered by black's M6. It is now a false threat because if Q10 is played, a capture to P9 creates a win for black.
N5: This is a very interesting outside-the-box move. It is also the reason why the author of the puzzle added an extra black stone at M5. Without the stone at M5, white to N5 creates an open tria which also turns black's tria into a false threat -- because if black continues, white captures across it, creating a 4x3 for the win. With the M5 stone in place, black has the time to simply defend this attack by playing K8 and white has no continuation.
K8: This is the other direction to try to attack the pair. An added bonus is that this creates an immediate threat that black must block at J8. By blocking at J8, white now essentially has the equivalent of two unblocked trias, which is often a win condition unless white can respond with a VCF (tesseras). Again, because of the black stone added at M5, white cannot capture because it is too slow. This would seem like a good time to come back to M10 and eventually set up the keystone capture that we cannot currently execute, but M10 still does not work here for the reasons described above.
The only moves remaining are the highly defensive blocking moves at L5 or P9. Incidentally, this is often how you should look for moves -- look for the most aggressive counter-attacks first, and then "settle" for blocking the tria if there is no other option -- unless the block itself leads to something very strong for you.
L5: This is the less aggressive / more defensive block. It seems like a necessary block to prevent even more continuation on the low end. However, this block creates no additional counter-attack possibilities for white and black can proceed with several other available trias such as J8.
P9: This is the more aggressive / less defensive block. You are gambling that you can build something from here that is faster than what you are leaving wide open for your opponent on the other side. Black clearly has plenty of continuation on the low side. But is it fast enough? P9 now creates several serious threats at tessera speed for white, including M10, Q10, Q8, R9 and so on. Either black must play something tricky that counters ALL such threats at once, or black must win out right now with tesseras. So far, I'm not seeing it -- this is a win for white.
Posts:
2,233
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente - Understand vs Memorize
Posted:
Oct 29, 2010, 12:27 AM
and yes the point was not so much who wins but how to determine the best next move in such a complex situation. its not so easy, and i think that point has been proven here. how ever it seems this puzzle is taking on a life of its own as people are now interested to solve it lol.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Pente - Understand vs Memorize
Posted:
Oct 29, 2010, 4:57 AM
Ah, yep you're right about that sequence winning after the capture. 21 ... H13 is dead.
There is no solution. Black is in a losing position at move 21 -- white wins with perfect play.
---------------------
In losing positions like this one, the next best thing to do is to increase the complexity and hope that your opponent makes a mistake. Which move has the best chance of causing your opponent to blunder? In this case, I would go with 21 ... H13 and hope that you are not playing against nosovs or that your opponent cannot find his winning move.
Re: Pente - Understand vs Memorize
Posted:
Oct 29, 2010, 5:07 AM
Zoey said: "in option 1; seems good as long as white doesnt play M11 on the 27th.
option 2; this is the more interesting path to look at. im not sure that K14 on the 28th is whites best play here. this should be examined more perhaps.
M14, Q10, Q10, P11, S14*"
Jason: Zoey, Here is your suggested modification; with one exception, white never gets a chance to go S14 because black caps at move 31. I don't see how this is a better option for white. With this in mind, I don't see what is wrong with Option 1 since black can still win if white plays M11 on the 27th.
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Pente - Understand vs Memorize
Posted:
Oct 29, 2010, 5:39 PM
Nice catch Jason! I did not see that capture either to ruin the S14 line.
Try a couple of ways of combining the idea I was looking at for white with this path that Zoey suggested. Here is an example:
28. J8 29. H7 30. H11*
At this point, white did not play a forcing move, but that is 3 captures, only two more needed to win and there is a hanging pair at J10 and then L12, which should thwart any line black might try along the K-line (or the diagonal such as L10). If instead black plays H10*, white's response of J10* counters black's possible tessera at H12 -- it's now a false threat due to playing into a 5th capture.
In addition, if black cannot win from there very quickly, there should be a way to force one more capture by white playing:
Q10 If black responds with P9* then, R10
This combo could also be used for the 4th cap because of the hanging pair at S14 for the 5th cap. However, a 3rd cap must be forced somewhere first...
(Ok, I just realized my J8 -> H7 -> H11* line probably doesn't work because a response at H10* actually creates a 2nd diagonal broken tessera... I wonder if the order of some moves can be switched to make it work...)
In the interest of just trying to find that 3rd capture, I wonder if a passive move such as 28. F11 is enough to force black back onto the defensive keeping in mind that the Q10 -> R10 combo is looming to force two more captures...
Someone who already has this set up to move stones around please try a couple of these ideas!
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: Pente - Understand vs Memorize
Posted:
Oct 30, 2010, 3:09 PM
Great diagrams Jason,
I was thinking at first that white could block instead of cap on option 5, but that doesn't quite work because of black to K15.
The trouble here is that black is threatening two win conditions at the same time -- the first is the capture to K14 and continuing with K15, the second is to extend to G8, G10 and G13 to force a LOT of captures. The quick idea of F11 was an attempt to get in the way of both of these, but it doesn't work. Other passive ideas for white such as M15 and then L13 (and a few others) just are unable to address both of these threats at the same time. This is starting to look like a solution for black!
One more idea, since you're already using Mark's WPente program with this puzzle, try starting after move 27, black's move to M10 (and try starting at some earlier points also), then set BOTH players to level 10 or maybe even level 12 if you have an extremely powerful computer. If there is a definite forced win inside of 5 or 6 moves for either player from this position, the computer will find it.
As of right now, I'm ready to say that this path is THE solution for black, and that the originally suggested moves for black actually fail (as I described above). I hope that Zoey will chime in and verify both statements.