| George John 2005-04-11, 5:59 pm |
| David Kane wrote:
David,
[SNIP]
quote:
> Of course, it is quite
> possible that more active players do in fact improve faster
> than less active players, so *should* be earning more bonus
> points.
As a general trend, I think this is very much true for younger and/or
lower rated players. For example, if you take ten, 600-rated students
and have them each play 100 games, especially if they keep score and go
over their games with a coach or teacher, they will very likely improve
quite a bit; although, under the prior rating system their ratings
would not change at all. With the current system that has variable K
and bonus points, their average ratings can go up due to the phenomenon
you mentioned above.
The key question here is it better to do this with a bonus system, an
activity system, or a combination of both? The Ratings Committee chose
a bonus system, which overall I think makes more sense than an activity
point system. The bonus system at least attempts to selectively reward
'unexpectedly' good results, often giving points to those who most need
them. By contrast, activity point systems (at least simple ones) give
points regardless of actual outcomes.
quote:
> I suspect that the main objective of the rating committee
> is to pump points into the system at a rate that keeps the
> adults happy, with other consequences being mainly
> ignored. I'm not complaining about that but was only
> raising a theoretical issue that has puzzled me.
The objective of the 'accelerated' system (which we now have) is to
correct about 50 points (IIRC) of rating deflation that occurred while
waiting for the new system (first proposed in the 1994-96 timeframe) to
go online (which happened in 2001). After that 50 (?) points have been
corrected, the aforementioned constant of 10 was to be changed to a new
value (16 was the originally suggested number) to produce just enough
bonus points to counteract the natural deflation in the system, and
keep the overall system stable.
Best regards,
George John
|