| |
Main
Date: 16 Nov 2007 12:00:10
From: Chess One
Subject: STRENGTH, and the GM flip
|
Now and again I chat with the geniuses who administer chess in this country.
One problem in conducting conversations is agreeing what words mean, and
yesterday we stumbled over this one: STRENGTH - [please ignore the computer
reference below, computer people get very emotional and upset if 'their'
subject is even mentioned by 'others']
So when you say 'strength' what do you mean?
>>
>> Some people say that ratings don't equal strength, but I never understand
>> what that means - at least as a general statement ratings indicate
>> playing
>> strength, but no computer program has a real, honest-to-goodness rating.
>>
>> PI
>>
>
> OK try this.
>
> Run a coin flipping tournament. Assign ratings to the coins based on
> their
> win/loss results.
I'm sorry, but chess results are not random, but dependent on chess play,
some say skill, and most people understand Dr. Elo to have made a relative
scale which places those performed skills in its index.
You may have been thinking that if people flip an assortment of GMs up in
the air, the really heavy GMs could only be flipped by the strongest
flipper?
> Some coin at the end of the tourney will have a high rating. Some coin
> will
> have a low rating.
>
> Would you say the coins have "strength"?? Or would you finally understand
> that ratings are only a direct relection of results....and we *assume*
> that the
> causal factor behind the results is "strength" (which is measured only
> indirectly).
When you say "we *assume*" are you referring to yourselves? It wouldn't
occur to me to assign strength to the coins, any more than to the GMs in my
own example.
Is it the flipper who has the strength, not the flipped? I'm thinking
that could be it!
Anyway, please share more of your thoughts on this subject, since I am a bit
confused when people /do/ talk about 'strong' players, and so on, since they
co-incidentally [?] seem to be referring to players with high ratings - and
the relative difference in ratings is the way we talk of a player being
stronger than another.
A final example, if I flip a 2600 player up in the air, in your opinion, is
this the same as if I flip two 1300 players? <No! that's not it. >
One 2600 player is the same strength as two 1300 players? <no, wait, I still
haven't said it right... >
Confused in Vermont. PI
> ECJ
>
|
|
| |
Date: 16 Nov 2007 12:55:41
From: help bot
Subject: Re: STRENGTH, and the GM flip
|
On Nov 16, 7:00 am, "Chess One" <[email protected] > wrote:
> Now and again I chat with the geniuses who administer chess in this country.
Can you be specific: which country? Do you /believe/
you are in Ireland, for instance? Or England? (There is
a strange flag by your name at ChessWorld -- not the
stars and stripes, nor even the flag of the Confederacy.)
> One problem in conducting conversations is agreeing what words mean, and
> yesterday we stumbled over this one: STRENGTH
>
> So when you say 'strength' what do you mean?
>
> >> Some people say that ratings don't equal strength, but I never understand
> >> what that means - at least as a general statement ratings indicate
> >> playing strength, but no computer program has a real, honest-to-goodness rating.
Ratings and strength are two different things (if I
counted correctly).
A player who retires from play for twenty years
may still have a rating of, say, 2825 or so, but
no longer be anything but a mere vestige of his
former self.
Some players can be observed to put far less
effort into games against their vast inferiors, and
expend far greater effort when playing those few
they recognize as their superiors.
But the ratings system blurs all results together,
mixing and stirring until all that remains is a blah
number. The number is accurate and fair, but it
contains less information, perhaps, than an
informed observer well-acquainted with the
subject in question.
Let me give a real-world example: here in the
state of Indiana there is a player named Emory
Tate, and he has been "taken out" of contention
more than once by far lesser players, when
competing for say, the state title. Yet when he
faces off against "other GMs" (yes, he actually
believes he is one!) his results are much better
that his rating would predict -- or at least his
upset potential is. This is most likely because
of his far greater effort in those games, and his
relative lack of effort when playing patzers.
At the other end of the scale there are many
players who have low ratings, but who are in the
process of studying and improving rapidly; such
players' ratings could well be off by a class or
more, since they knew so little to begin with.
If an average player were to sit down thinking he
was expected to win, and take the improving
player's rating too seriously, the result could be
disastrous.
Where ratings come in handy is in removing
the problem of human bias from the equation;
there are countless cases where someone will,
for whatever reason, proclaim that player-X is
far better than his actual rating, but where this
is entirely the result of personal bias, not facts.
When such unsubstantiated claims surface,
it is possible to look at results /objectively/ by
taking raw numbers and doing math -- where
the human-bias element is removed from the
equation altogether, *if* the equations and
relevant data were determined in an objective
manner, with no room for human meddling.
-- help bot
|
| | |
Date: 16 Nov 2007 18:43:00
From: help bot
Subject: Re: STRENGTH, and the GM flip
|
On Nov 16, 5:18 pm, "Chess One" <[email protected] > wrote:
> >> Now and again I chat with the geniuses who administer chess in this
> >> country.
>
> > Can you be specific: which country? Do you /believe/
> > you are in Ireland, for instance? Or England? (There is
> > a strange flag by your name at ChessWorld -- not the
> > stars and stripes, nor even the flag of the Confederacy.)
>
> There is also a strange rating attached to it, improbable both for someone
> my age and some who cheats, though these exhaust your own criterion
I see. So you cannot answer due to your "clock" of
secrecy pact.
> O. But we do not discuss such epehemral subjects as rating floors
That must have slipped my mind.
It so happens that in perusing a current listing of
the players in my area, I saw that several were
plastered on the floor -- held up by a purely artificial
construct (not unlike Parr-logic) of no substance.
My current goal is not to fall flat to whatever mine
happens to be -- at least not rapidly.
> > Let me give a real-world example: here in the
> > state of Indiana there is a player named Emory
> > Tate, and he has been "taken out" of contention
> > more than once by far lesser players, when
> > competing for say, the state title. Yet when he
> > faces off against "other GMs" (yes, he actually
> > believes he is one!) his results are much better
> > that his rating would predict -- or at least his
> > upset potential is. This is most likely because
> > of his far greater effort in those games, and his
> > relative lack of effort when playing patzers.
>
> an eccentric example, to example what point?
I expect it is a point far beyond the very limited
powers of a nearly-an-IM to grasp; all you need to
know is that 99% of readers here will "get it", so
your unique level of intelligence is an anomaly.
> > At the other end of the scale there are many
> > players who have low ratings, but who are in the
> > process of studying and improving rapidly; such
> > players' ratings could well be off by a class or
> > more, since they knew so little to begin with.
> > If an average player were to sit down thinking he
> > was expected to win, and take the improving
> > player's rating too seriously, the result could be
> > disastrous.
>
> humm.
Okay... ummm hmmm haa ha hmmm. Now you
try it.
> is the rating held by any player currently active in their rating pool,
> relective of their relative performance against other players in the same
> pool, similarly active?
Reflective? Yes. Always reliable? No. The key
to remember is that ratings reflect results -- no
matter how they may have been obtained. This
includes thrown games, fluke losses and obviously
unrepeatable events like Frank shall losing 0-7.
No, wait -- he did repeat that, over and over!
Currently, the best example I can give (and it's not
a very good one) is that of the GM (or IM) who
routinely takes a draw in the final round of a Swiss
tournament, so as to be able to leave early while
still collecting his first-place prize. Obviously, as
these are not real, contested games their being
rating skews the numbers, rendering them far from
perfect.
> > When such unsubstantiated claims surface,
> > it is possible to look at results /objectively/ by
> > taking raw numbers and doing math -- where
> > the human-bias element is removed from the
> > equation altogether, *if* the equations and
> > relevant data were determined in an objective
> > manner, with no room for human meddling.
>
> you cannot remove 'human bias' from things to do with human beings.
I can if I kill them all and then run the numbers
through an alien-designed ratings system to see
who was best! Besides, one day ants will rise up
and conquer inferior societies, like those of humans
and squabbling geese.
> did you say anything here I failed to reflect?
Did you learn to read yet? Only after completing
your course will you be able to finally grasp what
more intelligent people write here (which basically
includes everyone, unfortunately).
> the proposition that ratings are equivalent to strength is the issue, and is
> one a [relative] synonym for the other, is the question
The answer is "not necessarily".
But this is no meal-ticket for posturing that in
spite of rating X, you are really nearly-an-IM.
You see, there needs to exist a /real/ reason
of cause for the anomaly to exist, not merely
the emotional desire!
In my case, the reason my rating is not
reflective of my /current/ strength is that I have
not played in a long time, AND what I believe
is a very real ratings deflation which has taken
place in the interim. While former rivals sit on
cold floors, helpless to get up, I am still floating
in thin air... .
-- help bot
|
| | |
Date: 16 Nov 2007 22:18:46
From: Chess One
Subject: Re: STRENGTH, and the GM flip
|
"help bot" <[email protected] > wrote in message
news:9a1c5057-521a-43aa-8ffc-ee1646ac2ab9@i37g2000hsd.googlegroups.com...
> On Nov 16, 7:00 am, "Chess One" <[email protected]> wrote:
>
>> Now and again I chat with the geniuses who administer chess in this
>> country.
>
> Can you be specific: which country? Do you /believe/
> you are in Ireland, for instance? Or England? (There is
> a strange flag by your name at ChessWorld -- not the
> stars and stripes, nor even the flag of the Confederacy.)
There is also a strange rating attached to it, improbable both for someone
my age and some who cheats, though these exhaust your own criterion
>
>> One problem in conducting conversations is agreeing what words mean, and
>> yesterday we stumbled over this one: STRENGTH
>>
>> So when you say 'strength' what do you mean?
>>
>> >> Some people say that ratings don't equal strength, but I never
>> >> understand
>> >> what that means - at least as a general statement ratings indicate
>> >> playing strength, but no computer program has a real,
>> >> honest-to-goodness rating.
>
>
> Ratings and strength are two different things (if I
> counted correctly).
>
> A player who retires from play for twenty years
> may still have a rating of, say, 2825 or so, but
> no longer be anything but a mere vestige of his
> former self.
O. But we do not discuss such epehemral subjects as rating floors, and
statistical widows - this is entirely central to everyone's chess
> Some players can be observed to put far less
> effort into games against their vast inferiors, and
> expend far greater effort when playing those few
> they recognize as their superiors.
>
> But the ratings system blurs all results together,
> mixing and stirring until all that remains is a blah
> number. The number is accurate and fair, but it
> contains less information, perhaps, than an
> informed observer well-acquainted with the
> subject in question.
By all means explain yourself...
> Let me give a real-world example: here in the
> state of Indiana there is a player named Emory
> Tate, and he has been "taken out" of contention
> more than once by far lesser players, when
> competing for say, the state title. Yet when he
> faces off against "other GMs" (yes, he actually
> believes he is one!) his results are much better
> that his rating would predict -- or at least his
> upset potential is. This is most likely because
> of his far greater effort in those games, and his
> relative lack of effort when playing patzers.
an eccentric example, to example what point?
> At the other end of the scale there are many
> players who have low ratings, but who are in the
> process of studying and improving rapidly; such
> players' ratings could well be off by a class or
> more, since they knew so little to begin with.
> If an average player were to sit down thinking he
> was expected to win, and take the improving
> player's rating too seriously, the result could be
> disastrous.
humm.
> Where ratings come in handy is in removing
> the problem of human bias from the equation;
> there are countless cases where someone will,
> for whatever reason, proclaim that player-X is
> far better than his actual rating, but where this
> is entirely the result of personal bias, not facts.
is the rating held by any player currently active in their rating pool,
relective of their relative performance against other players in the same
pool, similarly active?
> When such unsubstantiated claims surface,
> it is possible to look at results /objectively/ by
> taking raw numbers and doing math -- where
> the human-bias element is removed from the
> equation altogether, *if* the equations and
> relevant data were determined in an objective
> manner, with no room for human meddling.
you cannot remove 'human bias' from things to do with human beings.
did you say anything here I failed to reflect?
the proposition that ratings are equivalent to strength is the issue, and is
one a [relative] synonym for the other, is the question
phil innes
>
> -- help bot
>
>
|
|
