Dots Championship 41 Dots and Boxes
6 replies. Last post: 2016-09-05Reply to this topic Return to forum
Loony at 2016-08-19
Now that the current Dots championship is almost finished, I’d like to share some thoughts about my games.
#1780093: In the beginning shinnosuke let me build those two small independent areas near the edge. The rest is just one big area that is big enough to have at least one chain and at the same time bounded enough to prevent more chains. In an ideal game we would see something like this. However, there was a twist when he played 11.j3. I still win nim, but when he creates the small chain on the right and joins it to the center I should sacrifice the two boxes on the bottom ; definitely too soon for that, hence my 14.e4. After splitting the big region the plan was to create a chain on the left and play a2 at some point. If the chain forms too fast he can sacrifice on e2, so I had to maintain some threats to build loops.
Interestingly the “big” region got split again. I played a2 nevertheless, because the center looked like two of those chains will probably become one later. After 19.e8 the most obvious move is 20.f11, but I thought it fails, too much confusion after potential 21.f9. The remaining areas can both become one or two/zero chains (with h5). In hindsight it was winning too, but 20.f7 was much more safe. It ensures two chains in the center join each other. After 21.e10 I can sacrifice the top left, and the only other option for p1 is shown by the actual game.
#1780094: As p2, trying to build a small chain near the edge in the very beginning is almost never good, because p1 can extend that chain by playing edge moves and at the same time reducing the remaining region until there’s space for only one chain. After 11.c4 I already had a very good position. If Marius played 12.d5 I could have answered with 13.g6 and the area on the left is big enough to cover a chain. After 12.d3 I don’t even have a non-sacrifice losing move. The sacrifice in 14th move was necessary, otherwise the chain would grow too much, but with two sacrificed boxes that early it’s hard to do anything wrong.
#1780095: Same idea as for #1780093 in the opening, threatening to create small outer areas without actually building them. Bernhard falls for my trap and helps me to remove all of them, then in move 8 I already begin to clear the edge too. Splitting the big region with moves like g8 doesn’t work, because the left can easily become a loop. The game is basically over at this point and in the end one super long chain forms indeed.
#1780096: Well, I didn’t have to think at all for this game. Why? Because it’s the opening (and also midgame) that has been played the most often in Dots. I’m sure if someone checks the database of LG he will find out the position at move 9 or 11 is the most frequent by far (considering symmetries). I wonder why this opening is that attractive to many players, even top players like Geoff Cameron used to play it. The main idea for p2 is to attack the center directly and try to split the board into 3 parts. While that might sound logical (because p2 wants to have 3 chains) it’s not the best approach, because p1 can use loop threats to join those areas pretty easily. Just pay attention to moves 3 and 5 of p1, they’re really passive and don’t even try to fight against the 3-parts split. Then 7 threatens to join the two top areas while extending the chain on the right at the same time. It’s kinda indicative when in a 3-region position p1 can have the confidence to actually extend given chains already!
So p1 could play 9.c6 and use the loop threat to get 2 big chains, but the game could continue like this. Unfortunately p2 has the nim advantage in those positions and some years ago we thought it destroys the chain on the bottom. Hence the 9.f1. It gives the possibility to use a loop threat with g2 (so the top areas don’t necessarily have to join) and when p2 plays e2 I can play delayed c6. The two additional moves on the bottom ensure a chain there as shown by our actual game.
Note that according to some more recent analysis 9.c6 actually wins too. I’d recommend to take a look at these two games that I played with Geoff: #1637600 and #1708227. The former was played very well by Geoff who pulled off a win against me in that hopeless opening by perfectly timing a pre-emptive quad sacrifice. However, 17.c2 would have won for me, just endgame mistakes. The latter was one of the easiest game I’ve ever played and now you know why.
#1780097: I thought I had a good position all the time as we just had 3 chains and the position didn’t seem special. I believe 14.a2 was my first mistake. The 2x2 area on the bottom left can be sacrificed at the cost of 2 boxes and I was somehow afraid. Then 16.b3 was braindead, firstly because it’s somewhat inconsistent to my previous move and secondly because Tobias wins nim with 17.j3. If he sacrificed the two boxes I would have played j5 and with the quad and close endgame I would have won 12-13. Maybe I should have played j5 first, I will analyse that line and post here later. Anyway, congratulations. Among all the games I played this championship I didn’t expect to lose exactly this one.
#1780099: I took the advantage pretty soon (big area to cover two chains and the one on the top right that could become a chain anytime). Moves 14 to 17 were optimal. I considered playing 18.b3, because it’s the most obvious move (makes a chain on the bottom so the big region will be two chains, sacrificing the top right is too costly with the quad), but in the last moment I changed my mind and that lead me to a losing game, nicely exploited by Roel with 19.b5. Endgames with 2 quads are advantageous for p1 most of the time, no matter how many chains. At this point I didn’t know for sure I was losing yet and ran out of time so I played 20.i10 to influence the short chains in my favor. 21.a6 keeps the advantage for p1 and from there it’s pretty fast to check I had a losing position. There were two possible traps for my opponent, 22.c10 (only winning move would be 23.b9) and 22.c8 (only winning move would be 23.e8). I used a lot of intuition to decide my move, knowing that if Roel just puts a bit effort into brute-force analyzing the game he would win either way. I’m happy it worked in the end and I’m really curious to hear if my other trap would have worked too or not!
#1780100: For me this was the most interesting one, also many mistakes by both of us. When I saw the first three moves by Recalcitrant I assumed he was either a secret pro or copying the moves from somewhere. It’s a good opening for p2 and I remember 5.e6 was the way to refute it, but I forgot how it works after some years of not playing that opening.. 7.h7 was bad and 8.g8 looks already advantageous for p2. According to Recalcitrant he played that move to set up for a possible quad later. The move was good, but not for that reason. Remember what I said about three parallel chains? I was going for it, but the center and bottom one were too close to each other, so p2 has the nim advantage when playing c6. I saw all of that coming, so my 9.g10 had the only purpose to split the board a bit more to set up for possible chaos play later. 11.b11 was just another trap: Assuming p2 plays a useless move I could connect on the right with h11 and if c8 comes then I could create a second chain by a8. Even if he played c10 it would allow me to force him to sacrifice boxes there even sooner. I can’t play d11 instead, because he still has the nim advantage and one huge chain in the center. Unfortunately he found 12.b9 which makes my previous move useless, keeps the nim advantage and prevents me from getting sacrificed boxes out of that area. Nextly, 13.a6, just a desperation move, trying to build a small chain on the left, with luck baiting him to to sacrifice the b6 box, knowing that even with that it was probably lost. 14.b3, the first questionable move by p2. According to my post computer analysis I could have won with 15.d11 (only move) there, but it’s kind of complex and I didn’t want an uncertain position, so I went for 15.j7, because now we have two small according areas on the top and a big area with possibly 2 chains in the bottom. Turns out it was actually losing, but Recalcitrant didn’t see either that he could have won by 16.k6 which turns the big area into one chain again. 16.d3 again dropped the ball (and again d11 was the only winning move for me, and again I didn’t see it). I was really convinced of 17.j5, because now the bottom really is 2 chains, f4 can’t be sacrificed because of f1 (note how that particular trick to build a chain near the edge already appeared in two of my games this ch, compare #1780096). Now p2 should have played 18.h11 to be able to use the quad on the right later in the close endgame, instead 18.i4 which has no real purpose in my opinion and gives me back the victory once again. Now my original plan was to play 19.f1, but all of a sudden I feared this line, which would have been quite sad. I had like one hour left with no vacation days, so I didn’t see that I could have actually won. I found my 19.d11 in the last minute and as I was kind of stressed I don’t remember my thought process, but anyway, turns out it’s a winning move, looks like nim. After both of the top areas are set chains I could savely play f1.
So yeah, thanks for the games, and even though this is not exhaustive analysis I hope my thought process behind the games and the mistakes I pointed out are helpful. :)
Tobias Lang at 2016-08-20
thx for sharing and congrats!
Loony at 2016-08-30
So for game 1780097 against Tobias:
At move 12/13 it was a win for me, but there was only one winning move for me: 14.a6. But 15.i8 was a mistake too, I have many optimal moves then, for example 16.c6 which wins the chain fight and in spite of a sacrificed box + quad also the game. I have no clue how I couldn’t see that, pretty much an obvious position without traps. Move 17 onwards is played correctly by Tobias.
Tobias Lang at 2016-08-31
Can u tell me which move instead 15.i8 would have won?
Loony at 2016-08-31
Sure, that’s 15.b1. You threaten to create that quad on the bottom left and if I sacrifice with c2, you force a quad on the right with j5, followed by a6.
Recalcitrant at 2016-09-05
Very interesting, thanks for the analysis!