championship 42 Dots and Boxes
3 replies. Last post: 2016-11-24Reply to this topic Return to forum
Tobias Lang at 2016-09-09
even if I will go dowb to league 2 I appreatiate to c JJ10 and Astrid back in championship. I will enjoy to follow the top games.
I hope to 2 c wccanard back in action too - he is my idol ;)
Loony at 2016-11-22
Now that ch 42 is finished I will again share some of my thoughts. Also, feel free to ask any game-related question about my ch games, your ch games, other ch games or any Dots game. I planned to add a bit more general stuff about strategy and mindset of the players, but the recently published Dots guide by purgency is excellent and covers a lot of it. I will assume you are familiar with it.
I appreciated seeing a strong player like Astrid back in the championship. The opening was not too interesting though. Last championship I talked about#1780096 in this thread and pretty much everything applies. After 12.j9 I had to be careful. I was scared of f9 because then two loops and another chain could potentially form in the top. Two small loops make the endgame close in many cases and I didn't want the game to be reduced to an analyzing-endgame-positions-faster-battle when I had a clear advantage in midgame and opening. I can't play 13.e8 because then 14.i8 makes three chains. I can't play something like 13.c6 because 14.g10 wins nim for p2. Finishing the quad directly runs into 14.f9 too. I guess I was scared of this line, but it seems to win for p1. Anyway, so I wanted to play a move in the bottom that strengthens the chain and prevents it from being sacrificed. 13.f1 or 13.a4 are both fine. I chose a4 because it ensure there will be a very long chain of length 8 and a 2-chain. f1 can run into a situation like this where an additional 1-chain can be forced by p2. Here it didn't matter but it's one of those little subtleties that can decide a match.
JJ10 eventually dropped out of the tournament and timed out, too bad. 2.i8 is a good move for the purpose of playing balance, see 7ics - Loony, where I lost but had lots of opportunities and a hard time defending for 7ics. To avoid that I played 3.g6. Now i6 is not as strong because the shape of the game tends more towards three parallel chains. 4.h7 is a really bad mistake, now the board is cut into two halves. Basically game over. I think the position after 3rd move favors p1 a lot so there won't be great moves for p2, but intuitively I would play something like 4.d9, 4.c8 or even 4.c6. The latter looks quite interesting actually.
1.j9 is the only first move known to be winning but I don't like it for human play. First of all it allows p2 to build some fast chains at the edge if p1 isn't careful, secondly those “2nd-row” moves are strong in conjunction with balance (“3rd-row”) moves because p2 wants to keep the outer regions as small as possible to make balance work (if they become too big the chains might join). Normally I'd play balance against this (2.c4 for example) but here I decided my opponent doesn't know how to defend against just building a fast chain on the top right. It's a bit dangerous, consider the position after 7th move. f10 could possibly be sacrificed and the chain grows and joins the center. After 13th move I had a problem again: If I play something like g4 or h5 (making the bottom right a chain, not a loop) p1 can still always sacrifice the left with d9. If the other chains are already too long the sacrificed boxes might not suffice to win. Hence 14.b5. The top left is a certain chain now and I was prepared to see a move like 15.h3 or 15.e4 with potential quad and forcing me to sacrifice some boxes. His 15.h5 solved all my problems, it was exactly the move I would have played next.
Moves at the very corner (2.a10) are a bit useless in the beginning. Edge moves are usually good choices for p1 to prepare joining regions at the edge but the corner is too far away to have any influence. I consider this kind of move as neither good nor bad for either player. Here it just gave me time to build a favorable center structure. 4.h9 is a good balance move. I wanted to prevent i8 because it's exhausting to play against quads. I decided against 5.h7 because it makes a connection between the top left and top right mroe difficult and I wanted to stay flexible where to build my chains. 6.d9 again a good balance move! I should have played 7.d7 threatening g4 (board is split into two parts then) and 8.e4 wouldn't help as I have enough loop-join threats like i4, f1, a8. But I paniced seeing a line like 8.f9 (9.g10 10.b9) 9.b11 10.e10 11.i10 which results in a position where p2 could sacrifice the whole bottom at some point and the center chain being long enough to make the sacrifices worth. I don't think it would have worked in hindsight but I'm not sure either. 7.c10 however was a mistake (I played it too prepare d7 after) and Tobias found the correct response 8.f9. His third balance move in the game, I'm impressed. Now there will either form a chain threat in the top with f11 or i10 or the center chain joins the bottom with c6. I decided a chain threat at the edge would be too dangerous and hoped for the line that actually happened. After 10.c6 I don't have to sacrifice d6 directly but can play 11.b5: Now 12.c8 doesn't work because 13.a8 makes a second chain. I calculated some lines like this or this and thought I have good chances even if it might still be lost objectively. 16.h3 finally dropped the ball and it's worth to look at the following moves to see how the bottom becomes one chain against the best attack.
This is one of the games I already let my pc analyze and it turns out that the position after 10th move is indeed won for p2. In 12th move p2 has only two moves that win (c4 and i6). The winning lines in 16th move go as follows: var1, var2, var3, var4.
Here kind of a trade-off happened. Recalcitrant let me play balance and he built an early quad in turn. That probably put me in a winning position but also made the game annoying and endgame-dependent. The early quad suppressed all kinds of sacrifices so beginning with move 8 it was all about forcing a favorable loony endgame with the correct number of small chains. The chain rule was basically rendered obsolete especially after 11.i8 because it makes a second quad and I have no simple rule to decide who has the advantage. My idea was to go for 3 chains and hope the chain on the left becomes big enough or joins one of the quads (generally, chains joining loops, called “dippers”, favor the winner of the chain rule) or start brute-forcing one move before my opponent does. :) 18.d9 was the last move played completely blindly without any reflection other than my general idea explained above. 19.d3 wins. I made a list of all the possible endgames that can happen now and who wins them: (numbers are chains, Q is a Quad, 6L a 6-loop and - indicates a dipper, two loops joining each other, *r and *b are red resp. blue wins)
3+3+Q+6L+Q 1+2+2 *r
3+3+Q+6L+Q 1+1+1+2 *r
3+3+Q+6L+Q+3 2 *r
3+3+Q+6L+Q+3 1+1 *b
3+3+Q+6L+Q-3 2 *r
3+3+Q+6L+Q-3 1+1 *b
6-Q+Q+3+3+3 2 *b
6-Q+Q+3+3+3 1+1 *r
6-Q+Q+3+3 1+2+2 *r
6-Q+Q+3+3 1+1+1+2 *r
6-Q-3+Q+3+3 2 *b
6-Q-3+Q+3+3 1+1 *r
7-Q+Q+3+3 2+2 *b
7-Q+Q+3+3 1+1+2 *r
8-Q+Q+3+3 1+2 *b
8-Q+Q+3+3 1+1+1 *b
9-Q+Q+3+3 2 *b
9-Q+Q+3+3 1+1 *b
9Q+Q+3+3 2 *r
9Q+Q+3+3 1+1 *b
That comes down to “if e4 happens, I need a chain in the bottom and two 1-chains in the top” and “if c4 happens, I either need a huge chain on the left or a 2-chain in the top”, assuming no sacrifices. That means he can play c2 against almost all of my moves to win easily, hence my 20.d1. Here 21.b3 wins for p1 (the only move) and I think it could have been found with a few hours of analysis, but both Recalcitrant and I were low on time. The position was interesting because depending on if c4 or e4 are played, I needed a different small chain situation in the top and vice versa. Whoever commits oneself first would lose the game and it turned out p1 didn't play the correct waiting moves.
I also have the computer analysis beginning with move 11 but after what I explained about this type of position I say it's not useful to look at it. At least I don't learn anything useable by doing so. But if someone is interested in certain moves after 11th, just ask.
6.i6 is a cool move. It can have the idea of 7.h5 8.f9 to make p1 sacrifice a box later or to extend the chain in the top and sacrifice f4 and f2 at the right moment to either build 2 chains in the bottom or sacrifice all of it. After 12.k8 I was scared of j5, I don't want to sacrifice 2 boxes that early. So I give up on a box immediately and hoped the sacrifice of f4 wouldn't work just then, but it's difficult. Here accurate play leads to a p1 win though. 14.d9 however solves all my problems, the move has no real purpose and now allows me to ensure a chain in the bottom with 15.j3.
Nothing special, in the beginning I played normal balance moves but quickly noticed I can create a chain in the bottom while letting the rest be big enough to cover two chains. After 17th move I just prevented any attempts of other outer chains. With g8 and b7 a 4th chain could possibly happen, but there was no nim problem that forced me to either let the 4th chain happen or the other regions joins, I always had the sacrifice ob d8 at a pinch.
8.c8 looks good because with moves like d9, f9 maybe the top and bottom chains could join, however, I still had h5 or g6 to cut them apart. I guess it was already too late for p2 to try something, opening went very well for me. p2 could have tried 12.f7 and indeed the chain can be destroyed, but it's too costly.
Tobias Lang at 2016-11-24
great work! We all have 2 study :) thx for sharing.