Ed Trice
What Is On Here Newspaper Articles >> December 4, 2001         MAIN PAGE
Stuff I Like
Good Books
Published Papers
World Championship
Newspaper Articles
July 21, 1989
August 27, 1989
December 1, 2000
November 24, 2001
December 4, 2001
September 8, 2004
January 19, 2008
December 12, 2012
Climate Change Debate
Government Essentials
Justice After Death
Grinning Dolphins
Groundhog Day
999 Girl
Software Programs
The Sniper
The Rock Fortress
Blackjack Deluxe
World Championship Checkers
Gothic Vortex
Omnia Obtorquebantur
Ed's Trainmaker
Ed's Encryption
Stuff I Sell
Gothic Chess Sets
Gothic Chess Programs
Military Grade Encryption
Stuff I'm Building
Soapbox Earth
Twitter Trains
Contact Me
  • Email : edwardtrice@mail.com
  • Phone : 610-818-5063
  • December 4, 2001

    World Championship Checkers actually had two kinds of databases: Perfect Play, and "Win-Loss-Draw" only.

    The Perfect Play database could tell you how many moves it took for the position to be won or lost (or if it was a plain draw). This database required one byte to store each result.

    The Win-Loss-Draw database needed only 2 bits to store the result: Binary 00, 01, 10, or 11 were the four different states. You could store 4 positions in a byte (and even many more) if you are clever (and we were).

    By the end of the year, we had over 132 billion positions computed that were win-loss-draw, and 19 billion that were perfect play.

    The 132 billion positions represented every way you could have 4 pieces against 4 other pieces on a 32-square checkerboard. The program could search through the game, look for instances where jumps would occur in succession, and reach an 8-piece position where it knew the result. If it was a win, the program could head for it. A loss or a draw it would avoid.

    And, once it simplified from 8 to 7 pieces, the Perfect Play database would know the best move without having to search at all! It made the program the first of its kind (to have both types of knowledge) and it was literally impossible for a human to beat.

    Shown above is the hardest position to win with only 7 pieces on the board. It is red to move and win. Of course 8-11 is the obvious move to avoid getting jumped, and then the other side soon crowns a king after 12-8 squeezes between the red pieces on 4 and 11. But from then on, the follow-up play becomes increasingly more complex. When World Checkers Champion Alex Moiseyev visited Ed Trice in 2004, he could not win this ending from the starting position.

    The Philadelphia Inquirer article came out after the local Hatfield newspaper published its story on the checkers program. Apparently reporters read each others' content in order to use it as material for their own brand. The reporter kept asking me strange questions about me when I was a kid, trying to take the angle that checkers was, after all, a kid's game. I resented that perspective, though prevalent it may be, it is still incorrect. I was able to prove that a 7-piece position shown in a book published in 1756 had been misplayed for 245 years! Not only that, it was a well-known, so-called "standard" endgame known as "Fourth Position" to the checkers-playing elite. The 7-piece perfect play database that Gil Dodgen and I created showed it could be won faster, and that there was also a better defense than those known to all of the modern players. I tried explaining such subtleties to the reporter, and she quickly brushed them aside, instead prefering to talk about how my younger brother and I would sometimes throw the pieces at one another when we were bored playing games the way they were meant to be played. Another article I hoped none of my friends read.

    Copyright ©2018 Ed Trice. All rights reserved.