Register
to add
comments.
Game number: Main Page
White: David J Bush    Black: upendra
twixt.DEFAULT.mc.2022.sep.1.2 (LG) | This game (LG) | Download JTwixt file
On 2022-09-18 at 14:01, Peyrol (info) said:
When I say "you" or "your" I refer to my opponent, upendra. I don't mean to exclude anyone from reading this.

Instead of swapping, the bot prefers |2.n18. It also considered m16 or m17. I believe Maciej Celuch also likes to play in this region of the board, but I'm not sure, maybe he would swap. This is way over my head. I didn't remember exactly what the bot response was. I just played in the same area.

The bot chooses |4.o13 instead of o12. In general it seems to prefer a 2-0 block instead of 3-0, especially in the early stages. It has also been known to stage a 2-1 block such as |4.p13. I still have no clue at this stage of analysis.

I use a crude method to guide my choices in the opening. Of course I look for blocking moves, but I also try to coordinate my moves with my existing pegs on the board. That means I count links. I chose j11 because it is just 3 linking moves away from o15. In retrospect this does not seem to pay sufficient attention to my prospects for connecting to the top.

The bot has an option to activate a real time evaluation display, where the top three move candidates are shown in a histogram. Running at full strength takes about 10 to 12 minutes per move on my machine. It's sort of like watching a slow motion horse race. At first, the race looked like a toss-up between |5.q12 and |5.r12. But in the last couple of minutes, |5.k11 got more and more visits, until it finally pulled ahead and was chosen by the bot, apparently in the very last second of analysis.

The bot definitely prefers |6.K7 instead of K6. No other move came close in the horse race. But your K6 was also strong.

The bot prefers |7.h6 with no close alternative, but it still evaluates the position as somewhat better for black. So early in the game, that doesn't mean much, as the bot frequently changes its mind. I set the bot against itself here, just to see some justification for white h6. I got the line 8.i7 9.f7 10.e8 11.i4 12.h9 13.e9 14.f13 15.m8 16.L8 17.p9 and I stopped the bot here. The battle has moved to the right, but could return to the left. White H6 is justified because white might be able to play H12 later on. The bot evaluation has changed to roughly equal.

My |7.R7 was supposed to make a double threat, either to win along the right side, or to suppport white L8. This probably gives black too much leeway in the middle of the top region. If I were going to make a threat towards the right, I probably should have played in a more conservative fashion, such as the bot's suggested M8 for white. Here the bot likes |8.n9 and says black has a strong advantage. One line is 9.s12 10.r9 11.r10 12.r5 13.m6 14.o6 Check out the 4 congruent triangles defined by 6 black pegs. In general, that triangle is a robust shape which is difficult to attack. 15.k5 16.j8 17.q8 18.r15 19.r14 20.p14 21.s16 22.r20 23.p19 24.s17 25.t18 26.t21 27.r17 28.o21 29.m20 30.p18 31.t22 32.s23 33.u20 34.u22 35.o17 36.r19 37.s19 38.q18 39.q16 40.q17 and black is crushing.

Your N10 also seems to win. The bot agrees with me about S12. The line continues |10.r10 11.j8 12.i5 13.d6 14.d7 15.c8 16.c9 17.d11 18.a8 19.l7 20.k11 21.l10 22.j13 23.p9 24.p14 25.d15 26.i11 27.v13 28.q12 and black wins.

You asked about |10.R14. The bot says 11.t15 12.q12 13.u13. I did not like that move for white. One thing the bot is not always good at is hustling for the losing side. On rare occasion it will just lay down and die. I took over the white side and played 13.q11 14.p10 15.p9 16.u15 17.j8 Here the evaluation changed from significant black advantage to small black advantage. 18.i5 19.l7 20.l9 21.m5 22.i12 ugh. So 10.R14 also would have won for black I guess.

Your R15 also works. Again the bot agrees with my move R14, but it's not enough to save me. I'm caught in a trap after |12.s10 13.s9 14.q13 15.s16 16.u9 17.r11 18.r12.

|15.d6 16.g4 17.d11 18.e8 19.n7 20.l15 black still wins according to the bot. I thought I could have taken over the analysis by now, but nooooo. You are still holding on to a winning position that you had since early in the game.

I thought I could have won with |23.h12 but the bot says 24.j16. For that matter, 24.i7 probably also works.

Your only mistake was P19. Instead, O18 would have kept the win. The winning plan for black is to conduct a pincer attack on the right side. My U12 was strong enough to make a real threat to connect to the top, and it makes a threat to connect to the bottom also. My problem is, if I am forced to defend the right side on both top and bottom, my defense is flawed. One line is |30.o18 31.u17 32.m13 33.t8 34.w13 35.v12 36.t14 37.w14 38.v17 39.w18 40.w15 41.v15 42.v13 But after 30.P19 I am able to reach the bottom in the middle, so you don't have a pincer attack.

Please ask if you have a question or wish to examine any variation in greater depth.

On 2022-09-19 at 09:14, Peyrol (info) said:
In that last variation, a simpler line was |30.o18 31.u17 32.m13 33.t8 34.w13 35.v12 36.t16

On 2022-09-21 at 21:22, upendra_ (info) said:
In this line: 15.d6 16.g4 17.d11 18.e8 19.n7 20.l15 ... I would have responded to the 15.d6 with the blunt object 16.d7. But I suppose 16.g4 is superior because black still cannot be blocked to the east, and now has a stronger platform to the west? Is that the logic?

And a related question: if 15.d6 has no hope of stopping black to the east... why would the white bot bother?

On 2022-09-21 at 21:22, upendra_ (info) said:
p.s. thanks for doing this analysis, David!

On 2022-09-22 at 04:56, Peyrol (info) said:
|15.d6 is essentially the Medcalf defense. The point is that white can later play D11 or near to that, creating a tempo threat that improves white's connection to the bottom, assuming white reaches the top somehow and can bring that connection over to the left. These hopeless edge attacks the bot does are all about trying to improve its position by minuscule amounts because that might be enough to turn a loss into a win. The bot prefers 16.g4 because, as you said, it keeps the top left, and it improves the connection to K6, which in turn improves black's defense in the top middle as well.After 16.d7 17.c12 18.g4 19.c8 20.c4 21.g3 22.e5 23.r10 24.r5 25.m6 26.o6 27.k5 28.l15 Black threatens to connect right with O18 or P10. 29.n8 30.j8 31.m10 32.l11 33.k9 34.j16 35.c16 36.h12 37.i14 38.k13 39.i13 40.f16 41.d18 42.g14 Apparently your blunt instrument still works, but the win is arguably more difficult for black than the 16.g4 line.

On 2022-09-22 at 14:02, upendra_ (info) said:
(I realize I mixed up my east and west in the above comment!)

Medcalf defense... googling... https://boardgamegeek.com/thread/1234635/interesting-corner-battle-one-instance-medcalf-def and http://twixt-commentator.duckdns.org/game/469338 will dig into it, thanks!

I assume there are other named strategies beyond the Medcalf. Is there a list / guide somewhere? Thanks David!

On 2022-09-22 at 16:32, Peyrol (info) said:
Not much. Perhaps there should be. The Achilles defense just refers to a 4-1 block, for example white E8 versus black F4, which is an important corner battle. There are names for other blocks, but I just use two numbers. That's all I can think of offhand. Running the bots for this game, in the |8.n9 variation, I was impressed by the attack in the bottom right corner (from my first post above) that began with 9.s12 10.r9 11.r10 12.r5 13.m6 14.o6 15.k5 16.j8 17.q8 18.r15 19.r14 20.p14 In order for this attack to work, O12 must have a connection to the left, and there must be a mostly open board on the bottom left area, so black can make threats to connect directly to the left. Maybe that attack deserves a name. The bot P14 attack? I repeat the line from above: 21.s16 22.r20 23.p19 24.s17 25.t18 26.t21 27.r17 28.o21 29.m20 30.p18 31.t22 32.s23 33.u20 34.u22 35.o17 36.r19 37.s19 38.q18 39.q16 40.q17

On 2022-09-23 at 15:37, upendra_ (info) said:
I'm not used to seeing the game with such a wide lens, so if you don't mind, let me summarize the bot P14 attack to make sure I understand it.

So with 20.p14, black creates a platform from P14-R15.

White is forced to do 21.s16. Black does the achilles defense. The achilles defense is successful because the play can now wrap back up to the P14-R15 platform (without which, the P18-R19 platform would be useless)

Do I have the gist right?

On 2022-09-23 at 16:03, Peyrol (info) said:
You did not repeat the variation that led to the P14 attack, which started on move 8. As a result, when I click on your move 20, I do not get the intended position. It is also helpful to place a vertical bar | before a move which is the first deviation from the game. That way, you can look at more than one variation in a single post. I also recommend you use the preview button and click on the last move of each variation, to verify you get the intended positions, before you commit to posting the comment.

|8.n9 9.s12 10.r9 11.r10 12.r5 13.m6 14.o6 15.k5 16.j8 17.q8 18.r15 19.r14 20.p14

Your summary was essentially correct, but I would add that white can prevent black from reaching the P14 group, but that would allow black to connect directly to the left. For example 21.s16 22.r20 23.q17 24.p19 25.n18 26.m20 27.m19 28.o19 29.o18 30.k19 31.p20 32.p21 33.i18 34.j17 35.j16 36.k15 37.m14 38.i14

On 2022-09-23 at 18:58, upendra_ (info) said:
Re: twixt-commentator notation/form/etiquette, understood! Thanks!

You must be logged in to add comments.