Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: the guy who said cm6000 is stronger than 8000 is right!

Author: Dann Corbit

Date: 18:18:55 05/22/01

Go up one level in this thread


On May 22, 2001 at 20:46:36, James T. Walker wrote:
[snip]
>Hello Dann,
>Of course I agree with your numbers since they are based on solid math
>principles.  The problem is this same "excuse" has been used since I first came
>here about 3.5 years ago and started testing Crafty 14.something.  It was
>approximately 150 points behind the latest commercial programs then and as near
>as I can tell it's about the same distance behind now.  Of course you can use
>the uncertainty factor to explain that they may be closer than the numbers show
>but that has been the same excuse for 3 years.  I think if anything the past 3
>years has shown that the data is more reliable than you are willing to accept.
>Yes you change commercial programs every year or two and Crafty changes every
>month or less so you will always have this "excuse".

Well, you might be right.  I suspect that a lot of crafty's losses are book
losses.  In any case, I still think the difference is not very great between the
top professional programs and the top amateur ones.  Given a commercial quality
opening book and a properly tuned platform, I think there is evidence that they
can compete on equal footing (see the last CCC combat zone where Crafty won).
On the other hand, this is not strictly scientific and I am doing a great deal
of extrapolation.

>The problem with your
>uncertainty factor is that they have never merged together as you indicate they
>possibly could.  I have no doubt that Bob is leading the way in several areas
>concerning chess programming and a lot of programmers are following his lead in
>these areas but to me it is obvious by now that Crafty's strength overall is
>still lagging behind the top commercial programs.  I have my own database of
>over 12000 games now and several versions of Crafty are included and the data
>shows that Crafty is improving at about the same rate as the top commercial
>programs in computer vs computer games.  I don't have a clue about computer vs
>human contest since I believe this is very different and does not necessarily
>follow the comp vs comp statistics.  In any case we can agree to not agree on
>this and there is no need to argue about it.  Believe me if/when Crafty tops the
>SSDF list I will be one of the loudest in cheering.  I will also cheer when
>Ferrett becomes available to the public.

I think we missed the last chance on 10/13/2000 so we will have to wait until
9/13/2019 at the earliest.

E:\nr\fixed>type badluk.c
#include "nr.h"
#define ZON -5.0
#define IYBEG 2000
#define IYEND 2100

int main(void)
{                               /* Program BADLUK */
  int ic, icon, idwk, im, iyyy, n;
  double timzon = ZON / 24.0, frac;
  long jd, jday;

  printf ("\nFull moons on Friday the 13th from %5d to %5d\n", IYBEG, IYEND);
  for (iyyy = IYBEG; iyyy <= IYEND; iyyy++)
    {
      for (im = 1; im <= 12; im++)
        {
          jday = julday (im, 13, iyyy);
          idwk = (int) ((jday + 1) % 7);
          if (idwk == 5)
            {
              n = (int) (12.37 * (iyyy - 1900 + (im - 0.5) / 12.0));
              icon = 0;
              for (;;)
                {
                  flmoon (n, 2, &jd, &frac);
                  frac = 24.0 * (frac + timzon);
                  if (frac < 0.0)
                    {
                      --jd;
                      frac += 24.0;
                    }
                  if (frac > 12.0)
                    {
                      ++jd;
                      frac -= 12.0;
                    }
                  else
                    frac += 12.0;
                  if (jd == jday)
                    {
                      printf ("\n%2d/13/%4d\n", im, iyyy);
                      printf ("%s %5.1f %s\n", "Full moon", frac,
                              " hrs after midnight (EST)");
                      break;
                    }
                  else
                    {
                      ic = (jday >= jd ? 1 : -1);
                      if (ic == (-icon))
                        break;
                      icon = ic;
                      n += ic;
                    }
                }
            }
        }
    }
  return 0;
}

E:\nr\fixed>badluk

Full moons on Friday the 13th from  2000 to  2100

10/13/2000
Full moon   4.0  hrs after midnight (EST)

 9/13/2019
Full moon  23.6  hrs after midnight (EST)

 8/13/2049
Full moon   3.8  hrs after midnight (EST)

 2/13/2082
Full moon   1.9  hrs after midnight (EST)

10/13/2084
Full moon  15.2  hrs after midnight (EST)

 6/13/2098
Full moon  15.6  hrs after midnight (EST)


> ( I will cheer for about any
>advancement in computer chess :-) )  In the meantime I enjoy matching the
>programs against each other on equal hardware and watching the fireworks.




This page took 0 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.