By the way compare H. N. Pillsbury's results vs. Frank Marshall and vs. Carl Schlechter and then compare their results vs. Emmanuel Lasker!... Compare also Pillsbury's results against Em. Lasker!!

<According to Reuben Fine in "The World's Great Chess Games" (1951), Irving Chernev thought this was the second greatest game ever played.>

That is what Chernev said in his introduction to the Dover reprint of Pillsbury's Chess Career by Sergeant and Watts.

"It is, in my opinion, the 2nd greatest game ever played..."

(before you ask it is Alekhine - Bogoljubov, also at Hastings, 1922.)

On material values, I developed a rather complicated scheme for LACHEX. I couldn't easily capture the relative values of Q vs RR or Q vs NNB or NBB or R vs BB or Rvs NN, etc. So I tried Q=9, R=5, P=1, B=3.1875, and N=3.125. Then I had the bonus: BB gets +.25. Other matters were handled by quasi-material values and positional values that depended on other things. By using several hash tables: entire position, K+Ps, Ps alone, etc. Some positional scoring didn't need computation, just a table look-up.

There was also some change in positional evaluation based on opening-middlegame vs ending, etc. It was things like Q+R is never an ending, QBB gets +1 ("attack point") QR gets +1, RR or QRR still +1, QN gets +2, Q gets +1, QNN +1, etc.

Moving pieces to the Center or towards the King got bonuses (B, N, R, B in "Kings Field" get +1, Q get +1, the sum is cubed.)

This was for positional evaluation. Also +.25 for having the "move" (Zugszwang handled by search). Reversed positions differed by .5 so tempi sort of got counted.

The root position got an extensive evaluation assigned to moves, not resulting positions. This allowed ranking of moves before the search. Searching good moves first vastly improves the speed of alpha-beta.

Most of this was "tuned" from several thousand tactical and strategical puzzle and game positions. These also had things like 1.e4 e5, 2.Nf3 Nc3, 3.Bb4 a3 having a preference for 4. Ba4 rather than Bxc6 whereas 1.d4 Nf6, 2.c4 e6, 3.Nc3 Bb4, 4.a3 Bxc3 was preferred.

Probably the most famous reference to the Pillsbury Doughboy marketing character was in the original film Ghostbusters. However, no doubt for trademark reasons, the reference wasn't clear, so unless one was familiar with him from ads one wouldn't have made the connection.

I actually incorrectly recalled that character as being the actual Pillsbury Doughboy before web search cleared that up for me.

Here is a good example why that is. Say white plays 45 Kh2 instead, below.

click for larger view

And it's a good thing there were mistakes. Accurate chess is so boring.

In judging the greatness of a game, accuracy from both players is an important factor. Few like games where one player just Rolls Over And Dies. (What we we used o call "Hitting the ROAD".) But when neither player takes a chance, the result will likely be a sterile draw. If that's the best chess has to offer, I'm off the to Tic-Tac-Toe site.

Chess doesn't become interesting until the players are willing to take a few well-considered risks to upset the balance. Unfortunately, computer analysis us implanting the idea that such actions are "mistakes" and take away from the quality of the game.

I eagerly await the day when some super mega engine completely solves chess, and presents us with The Perfect Game. carefuly calculated down to the last gadzillionth of a centipawn.Hopefully, computers will then abandon chess as being too simple a problem to warrant their attention, and we can have our game back.

What impact will The Perfect Game have on our chess? It will become just another Cheap Opening Trap. The knowledgable will avoid it, the unwary will fall into it. But even the victims need not worry. The supporting analysis will be too vast to be written down, the labyrinthine variations beyond the capcity of the human mind.

In other words, the situation in the AD (After Databases) years will be no different than in the BC (Before Computers) years. If we couldn't comprehend or remember the analysis of human GMs, how will we be able to use the even more complicated computer analysis?

In my mind, I like to imagine The Perfect Game has already been played. We just didn't recognize it.

Imagine two patzers named George and Fred, enjoying a game while sitting on a park bench during their lunch hour. At the end, George said "That was a good game, Fred", and Fred said, "Yes, it was fun. Got time for another one?"

<Chess doesn't become interesting until the players are willing to take a few well-considered risks to upset the balance. Unfortunately, computer analysis us implanting the idea that such actions are "mistakes" and take away from the quality of the game.>

Pillsbury plays a really interesting and quite modern and dynamic Queen's Gambit Declined and goes for a kingside attack while Tarrasch seeks his chances on the other side. <28.Ne2?> is bad because now Black would have been close to winning after <28...c3! 29.bxc3 bxc3 30.Rc1 c2> (D).

click for larger view

Stockfish says it's about -3, and to me that open b-file looks very inviting for the black rooks to come down to b1 (they're not of much use double on the e-file anyway).

Stockfish doesn't approve much of the following moves by both players, so Pillsbury stays on the brink of losing after <31.Nc1?> (<31.Nf4> would have equalized).

After <35.g4>, however, his attack finally gains some momentum, but the Great Siegbert unwisely chooses to ignore it, and although Fine's annotations show some fanboyism, I do agree to him that from here on, Pillsbury plays admirably. You really need nerves of steel to play <38.Ng4!> and especially <39.Rg2!> with your lone knight against a mass of enemy advanced pawns and pieces.

After <38...Bxb3?>, Stockfish gives Pillsbury a slight advantage, but <40...gxf6??> is completely losing. Fine says <40...Nxf6 41.Ne5> is "no improvement" -- but boy is he wrong, because now Pillsbury could have won immediately with the insane <41...Ne5!! Nxe5 42.dxe5> (D).

click for larger view

Now White, in addition to <Kh1> and <Rfg1> as in the game, threatens <e6> followed by <Qxf6#>, and there's just no way whatsoever Black can stop this. (No, not even the Great Siegbert!)

But the text move is also forcing a win, and Pillsbury's finish is really neat, especially <44.Qg3+!>. Most annotators rave about <45.Kh1> ("!!!") but personally I find that one quite obvious; and, I mean, what else *could* you play?