Moderators: The thread “Proposed ADM: Default delay times. Rule 5F” has drifted off topic. Suggestion: Move four of the posts near the end of that thread to the top of this new thread. (The posts in question begin on May 10, 2011, 8:14 am, and were submitted by Goichberg, Mottershead, Goichberg, and Wiewel.)
We all know that quick ratings are greatly deflated compared to regular ratings. To be sure, almost everybody plays worse at fast time controls than slow ones, but this does not explain the discrepancy. The two systems should be aligned so that, on the average, players’ quick ratings will be about the same as their regular ratings.
To the tiresome argument that “if the two rating systems are supposed to be the same, why have quick ratings at all”, I respond that it is only the average that should be the same. The number of players whose quick ratings exceed their regular ratings should be about equal to the number whose regular exceed their quick.
What happened to create the discrepancy? It is clear that one major cause (perhaps the major cause) is inactivity. A player who has a six-year old quick rating of 1300, but whose regular rating has jumped from 1400 to 2000 during these same six years, will now have two ratings 700 points apart.
No matter what caused the discrepancy, it is self-perpetuating. If a new player entering the rating pool breaks even against five players with regular ratings of 1700 and quick ratings of 1500, this new player too will immediately begin with a regular rating of 1700 and a quick rating of 1500.
So there are two problems: (1) how to bring quick ratings into line, and (2) how to keep them there.
I submit that problem (1) will never be solved by dual ratings, universal (quick-inclusive) ratings, or tinkering with bonus points or K-factors. Why adopt such a slow approach? We need an immediate infusion – something as simple as –
a. Immediately add 130 points (or some other number) to everybody’s quick rating.
– or, something a bit more complicated but probably much better –
b. Immediately add a number of points to everybody’s quick rating, this number to be determined by a formula based on several factors. One such factor could be how big the player’s current discrepancy is. Another could be how stale the player’s quick rating is. There could be other factors as well.
Problem (2), like problem (1), does not lend itself to a good solution through the use of dual ratings or bonus-point/K-factor manipulation. If the problem is inactivity, take the horns by the bull and implement a solution designed to combat the problem precisely where it exists. For example –
c. If a player’s quick rating has been inactive for two years or more, whenever a player plays in a regular-rated event, change his quick rating by the same amount as his regular rating.
– Or –
d. Same idea, but use a sliding scale. As above for four or more years of inactivity, half as big a change for two years, 1/4 as big for one year.
Of course, the office or the ratings committee should do as much testing as it reasonably can, in order to tweak the formulas in b. and d. above so they will resemble reality as closely as possible. Inevitably, somebody is going to complain that accurate testing may not even be possible, due to incomplete data prior to 2004, or for some similar reason. I say, an imperfect solution is better than none! Do the best job possible in interpreting the test results and tweaking the formulas. Then implement the formulas, and keep an eye on the situation for a year or two, at which time the data will be more reliable. Decide, at that time, what further tweaks, if any, are in order.
Above all, adopt a solution directly aimed at the problem. (1) Since quick ratings are deflated NOW, inflate them NOW, in one swell foop, rather than gradually. (2) If inactivity is the problem, attack its effects DIRECTLY, rather than indirectly through various cute manipulations.
Bill Smythe