FIFI25 vs. murasawa, October 2021

From HexWiki
Revision as of 15:17, 22 November 2021 by Selinger (Talk | contribs)

Jump to: navigation, search

Game information

  • Size: 11x11
  • Red: FIFI25
  • Blue: murasawa
  • Result: 1-0 (Red won)
  • Comments: Eric Demer
  • Location: Board Game Arena, table 212776246

Coordinate conversion and directions

This game was played with BGA's old hex implementation, so its coordinates and board orientation are different from what's shown on this page. ​ I apply a small rotation to show move coordinates and positions on this page:

(11;11) -> a11 ​ ​ ​ ​ (1;11) -> a1 ​ ​ ​ ​ (1;1) -> k1 ​ ​ ​ ​ (11;1) -> k11


References to the "left" and "top" etc. work for this page's orientation and the default hexworld.org/board orientation, but not for the orientation shown at the boardgamearena.com link by Location under Game Information. ​ (In particular, the left edge goes from a1 to a11 inclusive, and the top edge goes from a1 to k1 inclusive.)




Moves and comments

You can see the diagram for each move by following along at ​ this hexworld link .


FIFI25 plays ​ 1. a11

murasawa does not swap, and plays ​ 2. f6


abcdefghijk123456789101121




3. e6 ​ ​ ​ ​ good ​ ​ This works with the swap piece, since normally play towards an obtuse corner would favor whoever is on the short diagonal, but a11 means it would favor Red here.


4. e4


5. g7 ​ ​ ​ ​ h7 would normally be better (joseki), but g7 is good here, since if Blue connects to the right with ​ ​ 6. h5 ​ , ​ ​ then ​ 7. g5 ​ works well with f6, due to the edge template that f5+g5 would form.


6. f5


abcdefghijk1234567891011109486735


7. h5 ​ ​ ​ ​ 8. h4


9. c4 ​ ​ ​ ​ good (joseki) ​ ​ Blue is fairly strong towards the left, so Blue will probably connect to the left edge. ​ That means Red will probably need to ladder along the top, and c4 will be quite helpful with such a ladder.


10. c3 ​ ​ joseki again


11. d3 ​ ​ 12. c5 ​ ​ ​ ​ ​ ​ This ends the joseki.


13. e5 ​ ​ This seemed fine to me until I saw Blue's response :-). ​ Due to that response, Red should've tried j3 instead. ​ (j3 would create a ladder along the top towards the left and a ladder along the right towards the bottom.)


14. e8 ​ ​ ​ ​ This wins for Blue, though Blue did not keep the win: ​ ​ Due to c3, Blue can keep c5 connected and serving as an escape for height-3 ladders (Red can't foil), so Blue connects left via f4 or c9. ​ For the right, h4 gives Blue either a height-2 ladder plus a switchback threat or a height-3 ladder, and either of those is enough to climb to e8.


15. j3


abcdefghijk123456789101118201116151917121314


16. h3 ​ ​ ​This is locally good — Blue is being tidy before laddering along the right — and keeps Blue's win, but was not necessary: ​ i4 wins more efficiently than h3. ​ ​ See the variations for that.


17. f4 ​ ​ h3 reconnected Blue to the right, and f4 threatens Blue's connection left, so Blue should keep Blue's win by defending Blue's connection left, with c9.


18. j2 ​ ​ Instead, Blue wastes a substantial fraction of a move, since Blue was already connected right. ​ (See the variations for that.) ​ Now, not only am I no longer able to prove that Blue wins, but I in fact prefer Red.


19. b4

b5 was better (minimaxing): ​ ​ ​ ​ It gives Red a ladder escape fork at b2, so the only advantage b4 might have over b5 is avoiding a potential peep at e2. ​ ​ However, that is almost-always much less important than the extra strength b5 gives along the left. ​ ​ Furthermore, in this case, since Blue has e4 and e3,f2,g1,h1 are all empty, that peep provably achieves nothing: ​ Red can just answer with e2 with d2, and f2+g1 will still be red-captured.

However, even b5 is almost-certainly not enough here: ​ If Blue responds in the cell Blue plays for 22, then the situation becomes the same as just after Blue plays 22 in the game. ​ ​ I would've played b9 instead of either of ​ b4,b5 , ​ due to Blue's strength towards the right. ​ ​ See my explanation of the situation on top.


20. b3 ​ ​ ​ ​ If Blue was playing this as a timesuji [1], then I think Blue should've played g2 instead, so Blue would have more of them left. ​ Otherwise, this was bad because redD2 captures e1+d1, and thereby kills b3. ​ Red might play b7 to try punishing b3, but I would probably instead neutralize b3 with a move near the top.


21. g2 ​ ​ ​ ​ simple and fine, since this makes ​ 20. b3 ​ completely useless to Blue


22. d8 ​ ​ ​ ​ Very Good: ​ ​ Before I saw Blue play this, the position looked good for Red. ​ With this move, I am almost-certain that Blue wins. ​ (Blue does not keep this almost-certain win.)


23. c8 ​ ​ ​ ​ Red's only chance ​ ​ c8 has a connection to the top, and c5 is an escape for ladders under c8, so the upward direction is settled.


24. b9 ​ ​ ​ ​ blunder; now Red wins ​ ​ Red keeps this win for the short rest of the game.

Blue needed to defend along the bottom, either immediately with b10 (simpler), or after threatening to cut through near the middle of the left edge (b8 is faster).


abcdefghijk123456789101121272623222425


25. c9 ​ ​ forced


26. d7 ​ ​ Blue has a height-4 ladder. ​ If Red just pushes, then Blue will connect to c5 which will then ladder to b9.


27. b7 ​ ​ ​ ​ Instead, Red yields, winning. ​ ​ I also give more explanation regarding this move.


28. ​ Blue resigns ​ ​ ​ ​ b7 connects to the top via c6 or a6.


abcdefghijk1234567891011

Variations and more explanation

Blue's connection right

​As stated in the main comments, ​ ​ 16. i4 ​ ​ was winning for Blue,

abcdefghijk12345678910111617

and after ​ ​ 16. h3 ​ , ​ ​ i4 meant Blue was connected right.

The above diagram shows the area Blue uses to connect right without h3.


Blue defends

both bridges.


If Red pushes three times, then Blue just pivots and climbs,

12345978610

and Red yielding instead of Red's last push doesn't help Red.

12374659108

Blue doesn't need the 6,7 exchange here; that is simply shown to make clear why Red does not respond to the diagram's 8 further left than the diagram's 9.


If Red instead yields immediately, then

321654

Blue gets the same thing as if Red pushed twice.


That leaves Red pushing exactly once and then yielding, in which case Blue's 1 gives Blue a switchback threat, which lets Blue still climb.

15243768

Again, Blue doesn't need the 4,5 exchange here; that is simply shown to make clear why Red does not respond to the diagram's 6 further left than the diagram's 7.

The situation on top

Between ​ ​ 18. j2 ​ ​ and ​ ​ 19. b4 ​ , ​ ​ but Blue cutting Red off there would just give Blue a height-3 ladder along the left.

abcdefghijk1234567891011


abcdefghi12345

The above diagram shows the relevant area, and the below diagram shows the sequence.

abcdefghi1234513524

f2 is locally better than f3, but that doesn't matter here, since Red would still be connected in the upper-left due to the ladder escape fork at "*".

The ladder along the bottom

abcdefghijk1234567891011

Here, Red played b9, which loses because c8 is still connected to the top. ​ The below diagram shows the start of the ladder that should have occurred instead.

abcdefghijk1234567891011x214635

Blue played ​ 18. j2 ​ when Blue was already connected right, thereby giving up a guaranteed win. ​ ​ ​ ​ Furthermore, if Red had responded with ​ ​ 19. b9 ​ , ​ ​ then I would've preferred Red. ​ ​ ​ ​ However, j2 does mean Blue's center group connects right without using any of the lower-right, so Red has no useful forcing moves. ​ ​ ​ ​ As a result, I am almost sure that Blue wins here.

The final move

abcdefghijk1234567891011O


Against "O", "*" is the same defense as if a height-4 ladder is pushed too close to an a3 swap piece: ​ In that case, the attacker might be anticipating a foldback underneath, but if the defender yields just before the attacker's push makes a bridge to the swap piece, then instead the the original defender gets a height-2 ladder.


abcd123454563278191011

If Red pushes with c3, then Blue gets a foldback underneath. ​ Red should instead yield with b3, so that instead Red gets a ladder.

abcd12345262237154

To avoid this, Blue should pivot sooner.