In article , Mike Murray says...

On Wed, 7 Jan 2004 23:07:42 -0500, "StanB"
wrote:

SNIP


Or even people dropping out of chess.

Good point. Few drop out (or die) with a lower rating than that with
which they started.

Bad point. The above is irrelevant. If deflation occurs, it occurs while the
player is improving, NOT when they quit playing. This is a common
misperception.

Kevin L. Bachler

In article , Angelo DePalma says...


I didn't read Elo, but as I recall a discussion on this very topic about a
year ago brought some folks out of the woodwork who claimed that your
statement, Paul, was mathematically incorrect. That adjustments had been
made to counter deflation, and they worked.


You guys are talking around each other. The Elo system is implicitly
deflationary. As a player improves and gains points he will cause opponents of
"static strength" to deflate. This is what Paul is saying, and it is correct.

The new system is designed to re-inflate the system. Based on the way this was
done, players that do not play or who play little will see little benefit from
this. Based on the last measurements I had heard about, the system was, in
fact, re-inflating.

A bigger issue might be that chess, as a whole, consistently gets harder to
maintain the same place. For example, with the extra data and training
available today, it is much more difficult for me to outbook an opponent,
particularly someone who can devote more time to chess, like a high school
student. They are almost always more likely to know the latest lines better
than I.

So I don't try. I use old variations (which they seldom know) or try to
positionally outplay them.

But I see the following with regularity Angelo: young players with ratings
comparable to mine who I am convinced have less chess knowledge than I do. This
means that they have had a recent level of tournament success comparable to
mine. It does not mean that they are "stronger" in an overall sense. And I try
to use that to my advantage.

Kevin L. Bachler

I also believe that kids learn things faster than adults. One example
is how fast kids can pick up a second language. But I'm surprised no
one mentioned another very important reason why kids improve so
rapidly. They have a lot of spare time to spend studying chess.

How much time do we working adults have? I cannot possible imagine
being able to spend 1 hour a day studying or playing serious chess.
But I remember in Jr. High and High school where I would have large
blocks of time where I would screw around or watch sports or something
like that. If I was as interested in chess then as I am now, I could
have sustained over 2 hours of study a day, assumeing I didn't get
"chessed out".

On 8 Jan 2004 05:28:09 -0800, Kevin L. Bachler
wrote:

Or even people dropping out of chess.

Good point. Few drop out (or die) with a lower rating than that with
which they started.

Bad point. The above is irrelevant. If deflation occurs, it occurs while the
player is improving, NOT when they quit playing. This is a common
misperception.

Why is that? If player-A wins points from player-B, the points are
still in the system. How would that cause general rating deflation?
The analogy or model that springs to mind is one of a primitive
monetary system (no credit, no banks, just cash) -- the custom of
burying all a person's money with that person after death would be
inherently deflationary, wouldn't it?

In article , Mike Murray says...

On 8 Jan 2004 05:28:09 -0800, Kevin L. Bachler
wrote:

Or even people dropping out of chess.

Good point. Few drop out (or die) with a lower rating than that with
which they started.

Bad point. The above is irrelevant. If deflation occurs, it occurs while the
player is improving, NOT when they quit playing. This is a common
misperception.

Why is that? If player-A wins points from player-B, the points are
still in the system. How would that cause general rating deflation?

Because deflation is not caused by removing points from the system. It is
caused by players ratings dropping below their actual current strength.

Ex: (Under the "standard" ELO formula with no bonus, feedback, etc.) 4 players
are all rated 1500 and are in a closed pool. 3 stay static in terms of
strength, one improves to what we would call 1700 strength. Over time the 3
static players will see their ratings drop. But if you add up the number of
points there will still be a total of 6,000 points in the system, yet there is
deflation. If the improving player leaves, taking his points out of the system
causes no additional deflation (although it may make it more difficult to
RE-INFLATE the system)


The analogy or model that springs to mind is one of a primitive
monetary system (no credit, no banks, just cash) -- the custom of
burying all a person's money with that person after death would be
inherently deflationary, wouldn't it?

No. Deflation here means that your rating has decreased to a level less than
your "actual strength."

A full numeric example is below taken from an earlier post I made. It was in
response to a question about keeping the average rating of the pool the same, or
the aggregate number of points in the system the same, in order to prevent
deflation. Following the post are one or two additional comments. I hope this
helps clarify that having a player leave the pool is not implicitly
deflationary.

+++++++++

From: Kevin L. Bachler
Subject: Inflation - Deflation
Newsgroups: rec.games.chess.politics
Date: 2001-05-14 11:49:54 PST


Keeping the aggregate points the same DOESN'T stop deflation. Note the
following example, where the aggregate points are the same, but 3 players end up
being deflated:

Note that this example was done under the old system with K=32, but the argument
applies to the general trend in any situation with equal K among the
participants.

***********
A common misconception is that the change in the average rating of a
pool can indicate whether it is deflated.

(Note: if the number of participants doesn't change, this also means that
the aggregate points are the same.)

Let's consider the following example, which is a little contrived to
make it simple. However, the principles remain the same in other pools.

Suppose there are 4 1500's, A, B, C, D. They are all established
players. Their ratings are stable. To make the calculations easy, we
will assume that we will calculate rating changes under the old Elo
formula with K=32. Suppose that they are the only four players in the
rating pool. The average rating of the pool is, therefore, 1500.

Elo recognized that simply having an improving player can cause
deflation.

Let's suppose that A decides to study for awhile. As a result, his
strength increases to a degree that on average he scores 3 out of 4
against B, C, and D. These odds represent roughly a 200 rating point
spread.

What we would want the pool to do is this: Since B, C, and D are the
same strength as before, their ratings should stay at 1500. A should
see his rating go toward 1700.

That is, their performances inidcate a strength relative to where they
started of 1500 for B, C, and D and 1700 for A.

Let's suppose that they play 10 rated games against each opponent (30
total.) B, C, and D score 50% against each other, but only 25% against
A, exactly as outlined above. That means B, C, D and E win 12.5 games
each, and A wins 22.5 games.

What are their ratings (assuming for ease that we rate this as 1 event)
at the end of these encounters? (This example is simplified, but
illustrates the point, and the prinicples hold true even if we treat it
as several events.)


The rating formula is:

(W-We) x 32 + Rating old = Rating new

Since all players started at 1500, we expect them all to score 50%.
The winning expectancy, We, therefore = 15 for all the players.

For B, C, D:

(12.5-15.0) x 32 +1500 = 1420

12.5 points, for B for example, is 5 points against C, 5 against D, and
2.5 against A.

For A:

(22.5 - 15.0) x 32 + 1500 = 1740


What is the average rating of the pool? (1420 + 1420 + 1420 +1740)/4 =
1500.

Hmm...exactly the same as before.

Yet, B, C, D are all rated LOWER than their actual skill level of
1500. And even if A loses his "40 extra" points back to the pool
fairly evenly in another series of games, we would see:

A: 1700
B: 1433
C: 1433
D: 1433

That is, 75% of the players in the pool would be deflated, by 67 points
each, even though the average rating of the entire pool is unchanged.

+++++

Returning to the current email...

So now suppose play A stops playing and leaves the pool. What is the impact on
the ratings of players B, C, and D? Answer: NOTHING. Having a player leave the
pool doesn't change deflation at all. The deflation HAS ALREADY occured.

However, if any of those remaining players began to improve, it would further
deflate the pool without an opportunity to win back points from player A.

To prevent this from happening, additional points, such as bonus points,
feedback points, or having a sliding K for different players can be used to put
points into the system, thereby helping the 3 1500's to remain at their "static
strength".

Kevin L. Bachler

"I can only think of 3 players who were raised here in chess who have made GM -
in the past DECADE. (Shaked, Ashley, Nakamura. If you can add to this list, let
me know.)

John Fernandez
==============
I think most young Americans with real intellectual talent simply find
better things to do than waste their lives on chess.
I'd bet if you could "rate" the 1000 best computer programmers in the USA,
you'd find them to be the intellectual equals to the world's top 1000
grandmasters. I'd take our top 1000 medical researchers over 1000 grandmasters
any day of the week.
The USA doesn't have to be number 1 in chess, or even number 44, as long as
we're number 1 in the things that count.

RSHaas

The USA doesn't have to be number 1 in chess, or even number 44, as long as
we're number 1 in the things that count.
(RSHaas)
==============
Additionally, the USCF shouldn't worry much about USA's world chess ranking.
USCF should worry much more about enhancing and enriching our chess experience
if they intend to maintain the $49 adult membership.

RSHaas

Charles Allen:
How do one day swisses with 3 games work out? I would consider G/75
or G/60 games...

RSHaas:
I love G/60 time control. It feels like real chess. The last two
tournaments I ever ran were both 5 rounds in 1 day at G/60.

I'm pretty certain I played in some "40/60, mumblemumble" tournaments
back in the 70's, which is why I mentioned G/75 or G/60 as possible
reasonable times for me (need to read the delay clock rules though).

I was really asking about the "3SS" part. In other words, do players
mind having less determinacy for the winner?

RSHaas:
Meanwhile, are you at all related to the late preacher, Charles
Allen, of Atlanta?

Nope (well, that I'm aware of).

--
Charles Allen

I think most young Americans with real intellectual talent simply find
better things to do than waste their lives on chess.

You have nailed it.
The challenge is over.
A fundamental change happened in 1998 when Kasparov lost to Deep Blue. This
was the equivalent of losing a weight lifting event against a forklift.
A light went off in the public consciousness.
I cannot speak for the others, but Tal is advancing an academic career and will
end up being so much more productive for society than the endless pushing of
pieces.
Chess is fun, but it is not life.

Richard Peterson

"ASCACHESS" wrote
Chess is fun, but it is not life.

As opposed to college football? (smileyface here)