Cover Story / Ken Regan Figure 5
rich set of data with which to create all sorts of chess-based applications. In 2012, FIDE sold the marketing and
Move Number Figure based on 3.6 million* moves from all Category 11 and higher tournament games between 1971 and 2011. Figure 6*
licensing rights of professional chess to AGON, a company run by Andrew Paulson. According to the New York Times, “[Paulson] wants to turn chess into the next mass- market spectator sport.” Paulson plans to supplement Internet coverage of major competitions with something he calls ChessCasting, a broadcast of not only moves, commentary, video, and live engine evalua tions, but also biometrics such as a player’s pulse, eye movements, blood pressure, and sweat output. Regan’s work adds many non-invasive statistics to this list. “The greatest immediate impact on the professional chess world that I think I’m going to have, besides my anti-cheating work, is that I’m going to come up with a statistic called ‘Challenge Created,’ which is going to be an objective way to single out the players who create difficult problems for their opponents.” The greatest over- the-board practical problems are not always caused by the objectively best moves, and Regan’s metric can quantify this distinction. Other statistics that emerge from Regan’s IPR calculation include ways to visualize the degradation of move quality during time pressure (in Figure 5, notice how error increases as the move number approaches 40, the standard time control) and a way to normalize the different chess rating systems of the world. Amateur players constantly wonder how, say, their
Chess.com rating compares to their na - tional federation’s rating. IPRs provide a way to standardize this procedure. In some ways, IPRs are even more accurate than traditional ratings, because they’re calcu - lated on a per-move basis rather than on a per-game basis. One bad tourna ment could sink a traditional rating, but if this bad tournament was the effect of only, say, three isolated bad moves, then such bad luck would not detrimentally affect an IPR. Regan does admit, however, that engines bias their evaluations ever so slightly against human-like moves, and this effect “nudges IPRs slightly out of tune.” The exact reason for this tiny bias is unclear and it obsesses Regan during his free time. For the improving player, IPRs can be
Year Figure based on 120,000 positions between players with Elo 2585-2615 played between 1979 and 2014.
used as training metrics for different phases of the game. Say a person wants to obtain an objective measure of how well they play middle games out of the Ruy Lopez versus how well they play middle games out of the Scandinavian. All they would need to do is isolate the particular moves and positions of interest, send them through Regan’s IPR-generator, and they have a performance metric. This method has been used by Regan to rate historical players. For years, statistician Jeff Sonas has been rating historical players, but Regan’s IPR is more objective. Sonas uses
www.uschess.org 29
I
g Players with 2585 - 2615 Elo Average Error (pawns per move)
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76