Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To all those discussing 'new paradigm'

Author: Joe Besogn

Date: 03:44:19 11/07/00

Go up one level in this thread


On November 06, 2000 at 15:32:02, Fernando Villegas wrote:

>Hi:
>Third party tend to be "reality", or, in another words, the unavoidable fact of
>the superior technology you can develop with one paradig over the other, or trhe
>evident better or even unique explanation a paradigm offer of certain facts the
>other cannot do. So the arbiter is the world, the fact we live in the middle of
>facts. Facts at last impose his power over opinion. Fact and his correlate,
>neccesity.
>fernando


Yes. Reality in quotes. Kuhn's point is that there is no absolute 'reality'. The
world arbitrates on a pragmatic level and with world-tools which are themselves
embedded in a model which will be flawed in its own way. Voting on complex, new
scientific themes and artifacts by foot and wallet will skew the Darwinistic
ideas search tree in strange directions; phlogiston one way, autarky another,
catholicism, McDonalds kulture, communism, human-sacrifice, bean-counting.
Where's the rational, independant arbitration of 'reality'? Rational, just
another paradigm. Foot and wallet can take the search anyplace.

But I digress. Paradigm shift. Our debate is only beginning. Czub asserts, from
soul and heart - paradigm shift. Hyatt asserts not. This is no argument, this is
contradiction. Kuhn states a generalised history for scientific development,
moving forward as evolution, punctuated by violent change. Kuhn argues the
violent change as the major progress points.

1. Pre-science (independent researchers, many different ideas, each going his
own way)

2. Normal-science (one dominant paradigm emerges, development is
'puzzle-solving' within the dominant framework, common agreed set of measurement
tools, common agreed puzzles to solve, exemplars solved)

3. Crisis (exemplars become anomolies, new problems appear to which paradigm has
no solution)

4. Revolutionary phase (individuals set out to solve anomolies, use what went
wrong in the exemplar to make new exemplars)


From Kuhn's framework, we can see that the Hyatt assertion 'no new paradigm'
implies that Hyatt considers computer chess is in stage 2. Normal Science,
moving forward, solving problems evolutionarily. Hyatt considers there is no
crisis. There is no anomolous exemplar, no problem that can't be fixed within
the current framework.

Czub asserts computer chess is in or around phases 3 and 4.

Moving away from contradiction to argument, the question new paradigm or not,
hinges on

a) are the examplars of stage 2, normal science now anomolies? begging the
question, what are the underlying assumptions of the stage 2, normal science?

b) have individuals set out to solve anomolies?

Underlying assumptions. Kuhn tells us that these will often be 'taken for
granted'.

1. Quiescence search - idea of searching until the position is quiet - assumes
(a) it is possible to search to quietness, (b) quietness can be evaluated, (c)
if quietness can't be reached, then evaluate anyway.

2. Measurement systems - rating by SSDF, blitz rating at ICC - historically
accepted as ratings rose 1600, 1700, .... 2200. 'Progress' to IM and GM levels
at 2400, 2500, 2600 came under increasing questioning, (a) programs don't play
at GM level, (b) one program can play some games (tactical?) at 3000 ELO, some
games (positional?) at 2000 ELO.

3. King attack issues - that these complex calculations slowed the search so
much that better results (higher win/lose) percentages were obtained by keeping
the evaluation fast and simple - idea being that the tactical skill brought
about by increased depth would more than compensate for any positional
understanding from knowledge.

4. Risk issues - better to stick to measuring on the 1-3-3-5-9 material basis,
with positional factorised in centipawns. Known wins (rook and two pawns on
sixth, or two connected pawns on seventh as examples) to be measured in pawns,
but unknown, risky evaluations of king attack to be either not measured, or else
done in centipawns, to avoid danger of sacrficing material and then losing in
the endgame.


Individuals set out to solve anomolies? New ways of thought. News ways to view
problem anomolies.

1. Quiescence search - new paradigm says 'drive into the fog'.

2. Measurement system - new paradigm says rating lists not important.

Quote "These days chess programs are so strong that I often wonder if strength
is really the main important point any longer. Maybe playing style becomes
more important than playing strength the coming years. Ed Schroder"

3. King attack issues - new paradigm says this must be evaluated in detail else
the program plays checkers not chess.

4. Risk issues - new paradigm says unclear, chaos positions must be evaluated -
drive into the fog.


As Kuhn states, these is no independent arbiter between different paradigms. All
I think we can show is that there are two (maybe more) competing paradigms, and
point to the differences. The peer-acceptance decision will be pragmatic based -
which ideas 'work'.










This page took 0.01 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.