chess statistics scientific article

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
whereagles
Posts: 560
Joined: Thu Nov 13, 2014 11:03 am

chess statistics scientific article

Post by whereagles » Wed Jul 06, 2016 9:45 pm

Anyone heard of this article?

http://www.chessanalysis.ee/Quality%20o ... suring.pdf

Not sure if this was published in a journal, but seems interesting (if not debatable lol)

nimh
Posts: 46
Joined: Sat Nov 29, 2014 11:06 pm
Contact:

Re: chess statistics scientific article

Post by nimh » Thu Jul 07, 2016 5:33 am

I've already posted it here, and, no, it was never intended to be published anywhere. The methodology of such analyses is still raw and a lot of improvement is needed.

whereagles
Posts: 560
Joined: Thu Nov 13, 2014 11:03 am

Re: chess statistics scientific article

Post by whereagles » Thu Jul 07, 2016 8:26 am

Thanks. Cheers :)

CRoberson
Posts: 1994
Joined: Mon Mar 13, 2006 1:31 am
Location: North Carolina, USA
Contact:

Re: chess statistics scientific article

Post by CRoberson » Sun Jul 10, 2016 2:34 pm

I found several points interesting. Particularly the rating deflation ones. Based on the math, a person rating 2000 in the year 2016 has the knowledge
to achieve 2378 in the year1916. Oh, if we had a time machine to test that.

I am not happy with the opponent oriented ratings systems due to that and other types of effects.

It seems to me static entities could be used to improve these issues. If we had programs that would play at a given level even when used on varied HW,
we could use such programs to periodically test people to reduce those effects. Of course, we could use static programs on static HW such as
dedicated Chess systems like Fidelity, Mephisto, SciSys, ...

For programs, it could be programs with set search depths and with Qsearch turned on or off. So, 20 levels are possible with just up to 10 search plies.

MikeB
Posts: 3382
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: chess statistics scientific article

Post by MikeB » Sun Jul 10, 2016 5:40 pm

CRoberson wrote:I found several points interesting. Particularly the rating deflation ones. Based on the math, a person rating 2000 in the year 2016 has the knowledge
to achieve 2378 in the year1916. Oh, if we had a time machine to test that.
...
There is no need to test . Im all aspects of life, people get better what at what we do. Whether it's in sports, academia or chess - whatever. The exception is for the things that we no longer do on regular basis- take hunting for example.

If you take any professional sports star and put him in the 1916 league - he would be the super super star. Chess is no different, Whether it is 2378 or some other number , I have no clue, but I am certain that directionally that it would be a true statement.

yanquis1972
Posts: 1762
Joined: Tue Jun 02, 2009 10:14 pm

Re: chess statistics scientific article

Post by yanquis1972 » Mon Jul 11, 2016 3:52 pm

i cannot think of a more boring test, nor a bigger waste of a time machine, than seeing if an barely-rated expert would be a non-rated, non grandmaster level, but more dominant expert 100 years ago -- when chess was really in its adolescence, & true, organized professional chess its infancy...the only interesting thing i can even imagine is that he'd squeak into some games against masters of the day & get blown off the board in some games that may be recited today...

specialization -- people paid a living to eat, sleep, & breath one single thing, for years, or their entire life -- will naturally lead to advances in ability. i dont find that remotely interesting, but the study itself seems very cool.

i do find it extremely hard to believe rybka 3 on a quad i7 @ 65 seconds/move is only 2930 elo (FIDE 2008, at that!)...but the author did say he was just surmising.

Post Reply