============================================
Tournament Journal - ACCA Pan Am, Nov 8-9 2008
There are 17 entrants. Prophet is seeded 10th. Given the field I will be very happy to finish 12th.
Realistically, it's entirely possible to finish last. The only engine in the line up that Prophet
has a consistently winning record with is NoonianChess. But - anything can happen in six rounds.
Round 1: white vs NaumXMP - third seed in the tournament. Virtually no chance.
Actually looks pretty even after 15 moves... matter of time though. Prophet stays
even until around move 26. Naum is just too strong, outsearching Prophet by 4-5 ply
and searching at more than twice the node rate. (Not that Naum couldn't win if
those things were "even.") An eventual king side attack with Naum's queen, rook,
and knight did Prophet in.
Round 2: black vs. Tinker. A winnable game. (also losable) Tinker is out of book after
1. d4 f5 2. c4 Nf6. Prophet's strategy, of course, was to play something unorthodox to
finally get a time advantage (not). After move 4 I'm a little concerned Prophet is going
to get into problems with space/mobility. Move 14, Prophet destroys pawn shelter around
king in an attempt to keep the g5 pawn. It's "up" for the moment but I fear with as much
material is on the board a king side attack might overwhelm Prophet. Prophet goes +2 after
21. Qd3. Pieces seem to be coordinated a bit better now. Still concerned over a couple
major pawn weaknesses. (Though Prophet can afford to give one pawn back if it needs to --
it usually wrecks its position to try to hold on.) 26. ... Qc2. Prophet seems to be getting
aggressive. Tinker now has pawn problems of its own. And, Tinker is starting to get low
on time (about 13 minutes to Prophet's 25).
Tinker is getting about the same depth and NPS as Prophet, but is running on just one CPU.
Prophet is only getting about 2.2x speedup on 4 cores right now, so Tinker is getting about
2x the node rate of Prophet. Kind of sucks to have a parallel search, but having an engine
so slow that the benefit is negated! Brian reports that Tinker is in "deep trouble extra time
mode." Makes sense - no point in having lots of time on the clock as you lose the game.
31. Bh5. I wish Prophet would take one of Tinker's bishops. Too much firepower still on the
board - time to trade down. 32. ... Rgd8. Prophet seems to be unsure of how to proceed. Looked
like a typical "wait for the opponent to make a mistake" move. Tinker at < 7 minutes now. Prophet
has over 23 minutes. 35.Bf2 Nxf2. yes! finally some progress. And the double bishop threat
has been removed. Prophet now at +2.7. Tinker says -1.66.
39. Qxe3 : Prophet goes +3. Not sure why Prophet is so happy - materially it's just one pawn up.
43. ... Kg8 44. h3 Kh8. Prophet is shuffling the king because it doesn't know what to do.
Tinker is getting more optimistic... currently at -0.48. Prophet still says +3(ish).
46. Rdc1 Tinker's rooks are connected on the c file. 47. Rc8 Tinker Offers a rook trade.
Tinker offers a draw. Prophet takes with the b rook. Tinker now at 0.00. It's hard to see
how Prophet will make progress. Maybe with the passed e pawn and supporting f pawn. Prophet
failing low on move 50... coming back down to reality finally? This must be a hash table
bug of some kind. 63. ... Rd7 Prophet allows Tinker to take a pawn, giving it two passed
pawns on a and b files. I sense that's not good. 77 ... Bd4 Prophet's optimism is gone.. back
to 0. Tinker is very happy with a very advanced a6 pawn, +1.95. TInker's a and b pawns
too much, Prophet is going down. At least it was a good game.
Round 3: Prophet vs. Romichess (apocalypse1112) Romi must not have a book. Thinking for a very
long time after 1. d4. Missed much of the game... came back in at move 20. Prophet seems to be
a bit up... two bishops vs. a knight and two pawns. Both engines rate Prophet at +1. I
am sensing another pawn ending that Prophet won't be able to handle. 24. Rxc7 Prophet now has
a pig on the 7th. Both engines say Prophet is at +1.6. Romi goes under 10 minutes on the clock
with Prophet having 28 minutes. Rooks come off the board after move 30, leaving Prophet with
a bishop and pawn vs. knight. Tony mentions that Prophet is leading in ch 64, and that he has
his pinkies and fingers crossed for a draw. Can Prophet hang on? Last tournament Prophet was
winning vs. Romi but somehow gave it away in the end. Hope history doesn't repeat! Move 34:
Prophet exchanged the bishop for the knight. I wish he hadn't. Now it's all pawns. Prophet
played the ending well, pushing pawns up slowly... and Prophet wins! Feels good to avenge last
year's loss. Got some cudos from the folks on ch 64. Talked a bit about Prophet's parallel
search.
Round 4: NoonianChess vs. Prophet. Now that I've mentioned NoonianChess is the only computer
in the tournament Prophet consistently wins from, Prophet will lose.
Hopefully not...As usual, Prophet gives up a lot of time by not having a deep book. NoonianChess gets more
than a five minute advantage. Noonian's first move out of book is searched to depth 10.
Prophet should NOT have a problem here. Was out for most of the game. Peaked in a couple
of times and saw Prophet's score indicating a slight advantage. Came back at move 35 to see
+1.5 for Prophet. Noonian says -0.89. Prophet's search advantage finally kicks in and
seals the win. Note: Prophet seemed to take a really long time on move 42, while failing HIGH.
Double check it's not using extra time as it does when failing low.
So, Prophet finishes day one with 2 wins, 2 losses, placed 11th.
Round 5: Amateur vs. Prophet. Of course Prophet is out of book very early. This game is a
write-off, I probably won't watch much of it. Amateur gives up a pawn from book moves:
1. Nf3 d5 2. d4 e6 3. c4 c6 4. Nc3 dxc4. Checking in at move 13: appears Prophet is hanging
in there with a positive score, but has problems with piece development on the queen side.
The queen's bishop is hemmed in by the pawn on e6. Will Prophet be able to develop and
maintain a lead? Or will Amateur choke Prophet to death? Move 17: Prophet still thinks it
has an advantage, but king safety is in shambles. Amateur thinks it has just about pulled
even. I know that Amateur gives much higher scores to positional features than most engines
though. I'd feel better if the queens were removed, reducing the king safety problems.
Move 19: Prophet at about +2. Move 28: Prophet still up. Both engines seem to be at a loss
as to how to proceed. Move 39: queens come off, with Prophet still up by a (doubled) pawn.
Move 49: Prophet goes +3! Amateur at -1.6. Move 52: Will Singleton says "well, this looks
about over, nice game from prophet". I reply that it's not over till it's over. Will asks
me about my multi-CPU approach. Prophet wins!!!
Round 6: Crafty vs. Prophet. This one really will be a write-off. And I can't watch it anyway...
Have a short chat with Bob about parallel search, then I have to leave. Prophet was down about
a pawn when I left. Came home to see Prophet lost.
Prophet finishes the tournament with 3 wins, 3 losses, 8th place.