Quote:
Originally Posted by Das Boot
This piece seems poorly-conceived. It doesn't make much sense to use ELO (which incorporates actual draw rates) to generate a points-per-game prediction (which includes draw rates), then hold those two as fixed quantities while manipulating draw rates as your variable.
I linked your response in the
chess sub's '14 WCC thread. It's gonna be bumping come game day(s).
I hadn't thought about whether the piece's statistical underpinnings were sound or not, but good points. What I liked: I felt the chess-specific language was refreshingly accurate. Most mainstream chess articles are rife with niggling misuses of jargon and terminology, combinations of words few reasonably serious chess players would consider using (q.v. Nate Silver's sloppy use of "threaten" upthread).
This patch is a little bumpy:
Quote:
Research shows that great players like Carlsen and Anand draw far more often than rank amateurs — about half of all top players’ games end in draws. Grandmasters rarely make obvious errors, so decisive, winning breakthroughs are hard to come by.
The first sentence (post em dash) is the wrong stat -- the draw rate in 2750+ vs. 2750+ games is even higher than 50% afaik. The half figure includes games played against sub-elite competition.
The graph he linked* cuts off at 2600 anyway. Additionally, "breakthrough" has a
specific meaning in chess; you don't want to use it so casually. Not all wins are achieved through "breakthroughs".
But apart from that it was pleasant.
I can't imagine this sort of thing bothering most of a
mainstream chess article's target audience. But if it's not obvious why these errors or a
sentence like this can be annoying, compare it to
this howler.
* Incidentally, it made two mistakes of the sort I've been discussing. If you have to use the word
"expert", the ratings on your graph had better fall within 1900–2100 FIDE or 2000–2200 USCF. And we all know that "match" isn't a synonym for "game".