David Richerby wrote:

Wlodzimierz Holsztynski (Wlod) wrote:

In addition to the new players there is also the issue of inactive
players (vide Kamsky). I have addressed these problems, as well as
inflation/deflation--no such thing in the case of the Ockham rating
lists! In particular, to qualify for the next higher rating list, a
player has to be in the top 1/3 of the players from that higher
list, for say 12 consecutive games, when compared on her/his current
highest rating list (see the details in one of my earlier postings).

That doesn't stop inflation. The primary cause of inflation is, as I
understand it, new players coming in, giving a bunch of rating points
to the established players and then giving up. Your system does
nothing to stop this, as far as I can see, though the idea of having
stratified lists should slow the propagation of these inflationary
points.

Dave, what's wrong with you or rather
with your comprehension??!!!

There are, in the increasing order of the strength,
rating lists L0 L1 L2 ...

Only the lowest list L0 is not fully under control.
But this does NOT cause any inflation of the higher
lists. The newcomers to the higher lists will be
STRONGER than the average active players of
those stronger lists due to the very criterium
of admission. If anything, one would worry
about deflation!

Let me recall for you the details of the
conditions under which a player from
list L_k can advance to the list L_(k+1)
(s/he will still belong to all lower lists).

Such a player is compared to the players
from the list L_(k+1) on the list L_k.

But only to the players whose rating on L_(k+1)
is at least 1000, which means that they are
at least average for the list L_(k+1), i.e. they
are in the top 50% of L_(k+1).

Now our candidate for 12 consecutive games
has to be, within the list L_k, in the top 2/3
of the active players from L_(k+1) who already
are at least average or better on list L_(k+1).
This means that our candidate has to be,
for 12 consecutive games, in the top 1/3
of the active players of the L_(k+1) list
(in the top 2/3 of the top 1/2, i.e. about in
the top 1/3).

I hope that this makes it clear. I am ready
o answer questions but please try to refrain
from fast and unfounded judgements.

******

The parameters like 12 games or top 2/3
can be adjusted, even repeatedly--no harm,
to keep inflation/deflation in check. One
would follow the rating of the newcomers
of each of the list L_k. This way only
minimal fluctuations would be allowed, and
they would be easily corrected.

Formally, inflation/deflation in my Ockham
system is not possible at all, because the
average rating on each list is always 1000.
Thus in this formal sense the situation is
ideal.

It is however important that the newcomers
to the lists L1 L2 ... (never mind now L0)
will be coming at a sensible rate, and that
they will be doing better but not much better
than the older members. The taking over by
the new generation should take place at
a reasonable pace. It should of course, and
it will, depend on the coming new talent.

The proposed Ockham system should do
it quite well. It is a robust concept, not overly
sensitive to minute changes of parameters.

Regards,

Wlod

PS. One could control L0 strictly too
but it's not too important, it's not worth it.
The weak players from L0 list will have
almost no influence on the rest of the
more advanced players. Some vague
social pressures and the cost should
already keep L0 reasonable, so that
the weakest players will be at least
at the USCF 1400-1500 level (with very
minor exceptions, especially among the
active players).

If you want a control already at the L0
level then a practical (and historically
tested) way would be to have certified
(trusted) chessplayers-examiners. A
candidate would have to score, say,
7.5 out of 12 against an examiner
rated 1000 or 6.5 out of 15 against
an examiner rated 2000 (it's a multiplpicative
scale)..., which would mean that the
candidate during the match is on a level
clearly ABOVE the average 1000 L0-points.

Perhaps it's too much.