| Jeremy Spinrad 2005-01-28, 12:32 pm |
| This is a different, more serious situation than a game being voided. If an
entire club gets inflated, the effects could last for a much longer period of
time; I believe that this happened in Nevada in the Vaughan era. The problem is,
many players have most of their games internally within the inflated group; new
players come and get an inflated rating because of playing in a club tournament,
and the effect persists. You can't get that from a single game being misrecorded.
Jerry Spinrad
In article <1106854410.973559.13140@z14g2000cwz.googlegroups.com>, petrelet@sbcglobal.net writes:
|>
|> Mike Murray wrote:
|> > However, given that the ratings in question probably influenced the
|> > computation of other ratings, i.e., of players not involved in the
|> > controversy, cascading onward, AND that the players in the
|> controversy
|> > played in subsequent events, it appears unlikely the impact of these
|> > events could ever be fully expunged.
|>
|> Well, actually, if the players affected continued to be active in rated
|> play, the effects would be "fully expunged" with the passage of time,
|> and more quickly than most people would guess. I worked out some
|> simulations a couple years ago for the purpose of dealing with rare
|> situations on the ICC where a game result ought to have been voided. A
|> player might want 32 rating points back for an unjustified loss, say,
|> but if five or ten games had been played in the interim, the numerical
|> effect of the problem would already be almost completely washed out.
|> The point is that if your rating is "too low", then you gain more from
|> your wins and lose less from your losses, and if it is "too high", then
|> you gain less from your wins and lose more from your losses, so the
|> system heals itself, and the effects of any initial disturbance become
|> lost in the random variations that effect everyone's rating.
|> Of course, rating floors are an exception to this.
|>
|> petrel
|>
|