The iGoogle website offers a game server gadget. Lots of abstracts, from classical to obscure, have been implemented there, including Twixt. This is a great resource for aficionados across the world who like to play but don't want to commit to a turn-based game. See www.googlegamecenter.com for more info.
You can also view the game at http://twixt.servegame.org/game/806158
There the game is labeled as Tiziu Uzu versus Randy J. You can examine any variation you like. Click on the moves listed in the comment section to make the moves on the board.
Maciek Celuch, known to his friends as Maciej, is the top rated player on the turn based server Little Golem. We played around 9 or 10 p.m. my time, which was about 3 or 4 a.m. for Maciej in Poland. He probably had to get up early the next morning to go to school. I made the even numbered moves. His handle is nie_wiesz and mine is twixter.
E4 looked too strong to pass up, so I swapped it. That means I took the yellow side with border rows at the top and bottom, and Maciej played the rest of the game as purple (left and right borders.)
3.O15 4.J15 5.J13 6.G13 7.i15*
The asterisk indicates a link automatically added to the i15 peg.
It might look like yellow has a decent setup on the left side. The natural looking move here would be 8.F15*, which looks likely to force a connection to the bottom row. F15* is probably not a bad move, in fact. But I knew that if I continued to play on the left, two things would happen. As he attacked me on the bottom left, Purple would build up a wall of linked pegs which would make many threats to connect to the RIGHT side. Then, at the right moment, he would cut me off on the top half of the board with H9 (the red dot) or something similar. Without any pegs in the top right quadrant of the board to help me out, I would lose against this attack on the top, where the real threat lay.
Against a strong opponent, the opening is a time to spread out your threats in a direction parallel to your own border rows. So I placed what is called a "downstream peg."
Basically, this threatens either R15 or L14*. R11 is placed to help out L14 in case purple tries a simple ladder defense with L12. In that sense, R11 is downstream from J15.
Purple threatens Q11* which would weaken my lone R11 peg, so I add a link to it.
This is better than 10.P10* because it improves my position near the center of the board. If I had played 10.P10* then after 11.O11* purple would dominate the center and would have a relatively easy win.
The field of battle is taking shape, and as usual against Maciej, this shape is quite complicated. Purple might cut me off along the top, or along the bottom, or might connect his top and bottom threats through the center. Against all these myriad threats, with my time ticking away, I follow a simple principle: try to limit your opponent's reply to something which does not help him more than it helps you. The gap between purple's N17/O15 pair and the right border looked wide enough for me to attack there. If I could win the bottom right corner from P12/R11, then I might be able to reach the top with help from my E4 peg. If I could not win the bottom right, my attack would stretch purple's defenses and might give me an opportunity to attack in the center or on the left. Even if S16 does not help me, at least it does not make my position worse, and gives me time to try to read the board.
12.S16 13.S20 14.F15*
I begin to believe that I can prevail in the bottom right corner, so before the battle shifts to the top, I improve the connection from G13 to the bottom. My overall strategy is this: I believe I can connect both R11 AND G13 to the bottom, so in order to win, all I need to do is connect R11 OR G13 to the top.
15.H17* 16.R18* 17.R14 18.R13* 19.P13* 20.T14** 21.P19 22.S19 23.T18* 24.Q18*
Yellow threatens O19* or R21*. If I had played 24.R21* instead, then I could still win the local battle after 25.Q20 26.P20* 27.R19* 28.P17*, but this looked unnecessarily complicated.
25.Q7* 26.N11* 27.i20
This takes the sting out of my O19* threat, therefore threatening to win the bottom right with Q21**, and at the same time seems to shut down my threat to connect F15 to the bottom. Another choice here for purple was 27.H21, which after more than an hour of study I cannot find any refutation to. I will compare these two branches when we return the the bottom left of the board, assuming all the intervening moves would have been the same.
28.R21* 29.L7 30.J8 31.i6 32.G7 33.i10
3 a.m. notwithstanding, Maciej finds a way to connect his P9/Q7 to J13. There are lots of traps here for yellow. I did not see how to force a connection from N11 to the top, so I have to rely on F15.
34.H9** 35.K9** 36.i11* 37.K11**
First let's look at the variation with a purple peg on H21 instead of i20. In that case 38.D19 loses to 39.D21. For example 40.D23 31.E19* 32.F18* 33.G18* and purple wins.
So it looks like 27.i20 was the losing move for Maciej. Returning to the game:
The crucial variation looks like 39.E19:
Here the mechanical 40.F20* loses to 41.D17* 42.D16* 43.G20* 44.C18* 45.E21* 46.C21* 47.B20 and purple wins:
But instead yellow wins with 40.E17** 41.C20* 42.F19* 43.F21* 44.H20* 45.H22** 46.J19* 47.K19* 48.K17**
The apparently dead peg on J15 turns into a life saver here. Yellow threatens L14* which purple must block. 49.M14* 50.L19* 51.M20* 52.N20* 53.O21* 54.O19* 55.M19* (trying to draw) 56.O18* 57.P18* 58.Q19-O19/Q18**
Yellow finally punches through by removing one of his own links. With about 5 minutes left on my clock, I don't know if I would have found this line or not. But in the actual game, Maciej let me off the hook:
39.D22 40.C21* 41.F21* 42.D23* 43.E17 44.E16
He was hoping I would play 44.E18 which would have lost to 45.C18*. So Maciej resigned at this point. There is still one tricky shot he might have tried. After 44.E16 45.D15* I have to play 46.D14** 47.G18* 48.C16*. I didn't actually see this during the game. I'd like to think I would have found it, but I'll take the win any way I can get it.
Very well played! An exciting game!