| |
BGonline.org Forums
Reducing the draw percentage
Posted By: Nack Ballard In Response To: OT - World Chess Championship 2018 (Stick)
Date: Tuesday, 4 December 2018, at 8:55 a.m.
I, too, was flabbergasted when Carlssen offered a draw in the final (twelfth) game of the classical session. His pieces are better placed, he has a central advanced pawn, and several attacking plans aimed at the queenside, including a number of ways he might soon force the win of an exchange (rook for a minor piece). Against best defense, Carlssen still maintains a safe advantageous position, and with a 38-to-8 minute edge on the clock to boot.
I figured that I must be missing something obvious, but then I checked Stockfish analysis: White's best initial move is assessed as more than half-a-pawn worse and its main variation (starting with Nh3) culminates in a nice methodical queenside break. (For any interested chess players, grandmaster Peter Svidler gives a nice analysis of Game 12. Fast forward to 27:20 if you just want Svidler's comments on the final position.)
It is true that IF Caruana somehow equalizes or nearly equalizes he will most likely turn down a draw offer later and might even win, but I don't believe that Carlssen was an overwhelming enough favorite in the tie-break portion to justify not pressing that strong a position. Based on computer modeling, Carlssen was 85% to win the tie-breaks (being world-ranked #1 in Rapid and Blitz, vs Caruana being #8 and #16), but I think he would have to be at about 97% in the tie-breaks to justify offering a draw when he did in Game 12 (which won't even be accepted if Caruana happens to have analyzed the position so well that he sees he is very unlikely to lose -- it's only a draw in hand if Caruana goes along with it).
Presumably, Carlssen had enough time to rehash the position before the press conference, so I'm surprised that he still defended his decision and even referred to former WCs Kasparov's and Kramnik's play-on opinions "stupid." Admittedly, none of this unequivocally proves he was mistaken.
To my taste, there are way too many draws in competitive Chess. It should be noted that schemes such as the Bilbao scoring system (+3 for a win, +1 for a draw, 0 for a loss) are a great way to motivate sharp non-drawish play in tournaments, but are irrelevant for match play.At a high level in Chess, one needs too large of an advantage to convert a close position to a win. This is why I gave up tournament chess in the late 70s. I could see the writing on the wall: the stronger I got, the more drawish Chess would become when facing equal competition. Chess is a lovely game, but Go is at least as beautiful without the draws. You can win Go by as little as half a point, whereas you must win Chess by what would be the equivalent of 10 or 20 points to change a draw to a win. So, I switched to Go.
If you stick only to fast controls (rapid, blitz or bullet), there will be fewer draws, but for a game like Chess I think the slow classic time control should be the primary test of skill. Too bad it's so drawish when played at a high level!
As I see it, this draw-drawback can be fixed, even in match play, by any of several solutions. A few are listed below in my ascending order of preference:
(4) Broaden the definition of a win/loss result. When there are fewer than "x" (e.g., ten?) pieces on the board, and a player correctly claims that a forced checkmate is no longer possible by the opponent, he can claim a win or "partial win" by material advantage (and then say five more moves are played by each side before the adjudication). Proving this would have been difficult in the past in some positions, but we can now use a chess engine as an arbiter. By "partial win," I mean a result greater than 0.5 but less than 1 (and there might be a scale calibrated to the size of the material difference). An invalid claim is penalized by a loss or greater partial loss.
Unfortunately, this would change the nature of the game (better or worse would be a matter of opinion) and not everyone would like that. There would be less incentive to fully win, but there would be a ton of interesting decisions and virtually no tied results in tournaments or matches.
(3) Black always gets draw odds. As that change alone would give Black too big of an advantage, give White further positional compensation, which can be discovered through experimentation or engine simulation. For example, White gets to play two moves before Black plays his first. If that's too much, White can move a pawn one square plus another move. Or, if it's too little, White gets two moves plus a pawn one square. And/or maybe one of the moves has to be with a rook pawn, whatever it takes to equalize prospects. (If more than one specific equalizer is discovered, different equalizers can even be used each game.) Every game is decisive with this solution.
(2) Every round of a tournament or match has a decisive result (1-0) consisting of a tie-break series. First, a game with classic (slow) time control. If that is a draw, then a rapid game (with the opposite player getting White), and if that is also a draw then an Armaggedon tie-break blitz game (where Black spots time on the clock but gets draw odds). Ignoring which player it might benefit in the specific case, imagine how much more exciting the 2018 World Championship match would have been with this scheme.
Unfortunately, none of the first three possibilities solves the conundrum of players memorizing the first couple dozen moves of opening(s). More on this below.
(1) Shuffle Chess or Fischerandom Chess. Back in my teenage chess days, I had read somewhere that a tournament had been played around the turn of the 19th century where the initial locations of the bishops and knights had been switched, in order to take players out of the "book." There seemed to be some sort of consensus that the games were more original and interesting.
Not long after (mid 70s), I read about Shuffle Chess. In this variant, only pawns start on the board. In alternating turns, each player chooses one of his/her own pieces to put on the first rank square of his choice, until all eight first-rank squares on both sides are filled.
In standard chess, the draw percentage among top players is very high, and keeps getting higher, as players have to guess less and less which moves are best, and memorize deeper and deeper into the opening, maintaining equilibrium until the positions are simplified to where there is not much advantage left for which to play. (In the regular series of the Carlssen-Caruana match, the average error was only 4 or 5 centipawns per move.) By contrast, in Shuffle Chess, specific opening preparation is pointless, as there are just too many possible opening setups to focus on them all. Players have to reinvent the opening pretty much every time.
Fischerandom Chess is similar (and is often confused as being just a renaming of Shuffle Chess). A computer program chooses a random back-rank formation for White and duplicates it as Black, with the condition that bishops are of opposite color (which any decent player would opt for in Shuffle Chess anyway) and another condition that the king is placed somewhere between the rooks (with modified castling on either side being legal).
I slightly prefer the original Shuffle Chess (perhaps with the condition of K between Rs being added) over Fischerandom. It has the benefit that Black has a slight advantage for the first eight turns, as he gets to see each of White's placements before he chooses his own, which partially offsets White's advantage in moving first (starting on the ninth turn). Also, there are even more -- a lot more -- possible opening positions (because symmetry is not a requirement), which means that Chess can probably continue for many more centuries before the openings become can be meaningfully solved and memorized to the degree they are today.
Nack
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.