#51 · Arek Kulczycki
5 May 2016
oh, about 1b* I was silly, black indeed connects
Public forum discussion.
5 May 2016
oh, about 1b* I was silly, black indeed connects
6 May 2016
Arek: In wccanard's f14 diagram, the black edge was supposed to be blocked by white in the upper right, like this. This position should be easier to solve, but I guess the solver would still take years.
Do you think black has a chance to connect f14 to the bottom after white f16, where we ignore the black edge at the top? Do you think it would help if the board was wider?
6 May 2016
I mean: Wider towards the right side, only.
6 May 2016
Yes, apologies, I meant to block off the top with white pieces as urmaul said. The solver doesn't seem close to solving it. It wants to consider many moves though including random moves like o2. Perhaps the solver would have more of a chance if I put far more white pieces on the board, to stop it being distracted up there? But I don't know how much space people think black needs. Probably pieces on the 7th row, but what about the 8th? The 10th? The 12th?
6 May 2016
A piece on the 5th row short diagonal can't connect to the edge:
The highlighted move stops it. The main line is e10 d10 c12 g8 e11 h9 h10 g10 h8:
and black maybe runs out of space because the board is too small :-( Never goes above 6th row though.
Is it known that a stone on the 5th row short diagonal is connected to the bottom on a large board?
6 May 2016
6 May 2016
I remember calculating that 5th row template by hand a while back, as part of my work on the 7th row template problem.
6 May 2016
urmaul is right -- I tried a board of size 13 and it's just too small; black connects with a board of size 14.
My feeling is that this sort of problem is just like the general edge template problem -- people can do the first few examples, but then there's no systematic way to continue and the problem is too hard for computers.
Here is a related question. We all know by strategy-stealing that with no swap, player 1 wins a game of hex. But what if player 1 were playing shalev? Then it's not so clear that player 1 can win any more in practice. But if I were playing shalev on a 5x5 board I would be OK because I play in the middle and then I can template my way to both edges. How many moves do I need to be able to make without reply to make sure I can beat shalev on a board of size n **just using known templates**? This is another way of measuring how little we know about templates. For example on 7x7 I would need to play two moves before shalev moved, because if I played in the centre then I don't know two disjoint templates connecting me to the two edges (and in fact probably no such templates exist). Maybe this isn't such an interesting question but it's a way of measuring something. For example on a 19x19 board there is an answer, but maybe if someone discovered a 7th row template we could get a smaller answer.
6 May 2016
I like to think of that template as a doubled Tom's move. I think I used it in a 19x19 game once.
7 May 2016
Here's a triple Tom's move, I'm not sure if this could have been defended or not.
7 May 2016
@f14: Oh, then no way, black cannot make it. White wins no matter how big a board. Proof? Can't formulate a mathematically correct one, but there are just few ways to go: 1) ... 2) ... 3)... 4 here placing of move 83 is irrelevant)... To everything else white f15 works, no?
7 May 2016
Arek: Well by now I guess that's right. Above lazyplayer also gave this refutation for 4). If black really can't connect, I think we can't prove it until we can prove thinks like 1b. But I understand that lazyplayer (years ago?) and Maciej (in 2009) - no offense, really - both thought they had found a way to connect.
7 May 2016
Now Arek has given us some lines to look at, this will make the computer analysis easier.
7 May 2016
lazyplayer and Maciej must have been wrong :)
8 May 2016
I agree f14 doesn't seem to connect, but I don't think it's as simple as Arek says. For example, this line by Arek seems wrong. The correct refutation is white f17, not f15. (Or am I missing something?) I do think most stones are refuted by either f15 or f17 in a straight forward way (but of course I can't prove anything).
8 May 2016
ok, indeed f15 is not so good, I picked it for simplicity. G17 is usually the strongest, but more complex. In any case white has a fairly easy win here.