Jun

13

The first chess computer, 1960There is a vast literature supporting the use of mechanistic decision in repetitive situations, [instead of] over relying on human expertise. Forgetting about accuracy for a moment, which is key, humans are quite inconsistent in the way they use information. Show an expert the same fact set on repeated occasions and the conclusions only correlate at about 0.50. In other words, the facts only account for about 25% of the variation in the expert’s final conclusion. This suggest that the way information is being weighted from instance to instance is inconsistent or the expert is considering information outside of the fact set. When it comes to accuracy the decision algos do better, overall.

Rich Bubb adds:

Here is an interesting example. Soon all of you will be replaced by machines [LOL]:

[An automated investing] system was developed by Robert P. Schumaker of Iona College in New Rochelle and and Hsinchun Chen of the University of Arizona, and was first described in a paper published early this year.

It's called the Arizona Financial Text system, or AZFinText, and it works by ingesting large quantities of financial news stories (in initial tests, from Yahoo Finance) along with minute-by-minute stock price data, and then using the former to figure out how to predict the latter. Then it buys, or shorts, every stock it believes will move more than 1% of its current price in the next 20 minutes" and it never holds a stock for longer."

Source: MIT Technology Review Blog

Nigel Davies writes:

That's a pretty sad comment on fund management standards. Humans have put up one hell of a fight against computers on the chess board using raw (unaided) brain power and still beat the best machines if they're armed with an ordinary PC and have a longer time limit. And that's to say nothing of the drubbing that Go players have given computers, even giving so far as to give them odds.


Comments

Name

Email

Website

Speak your mind

1 Comment so far

  1. Mitchell Jones on June 17, 2010 1:12 pm

    I suspect that a large part of the advantage a computer has over a human player in chess stems from the fixed opening position, rather than merely from the computer’s unarguably superior calculative abilities. Because the opening setup is always the same, there is an “opening book,” by which I mean the accumulated experience of the masters in the opening, over the century or so that the moves of the game have been constant. That accumulation of opening knowledge has made it possible to memorize vast numbers of well-tested opening variations and, as a consequence, spring myriad traps on an opponent who lacks a similar knowledge base. It is not an accident that IBM’s “Big Blue,” which defeated Kasperov, was “booked up out the kazoo,” as one writer put it at the time.

    So here is my thought: why not jettison the fixed opening position in chess? Start with the pieces on the sides of the board, and begin by taking turns—white goes first, of course—with each player placing a piece anywhere he wants it on his half of the board, or placing a pawn anywhere he wants it other than on the first rank. Then, when the last piece or pawn has been set down, actual movements of pieces or pawns would commence, and the game would assume its more traditional aspect.

    Under such a regimen, memorizing an “opening book” would become pointless, and chess would come closer to being a pure test of the depth of a player’s understanding and judgment. And, I suspect, under such a regimen human chess masters would once again be able to give odds to the best of computers, and still beat them.

    Such a change will never be accepted, of course. The ascendancy of such a chess variant would shake all the rote memorizers out of the master rankings in short order, leaving only those who actually possessed deep understanding of the game. Result: the rote memorizers within the chess establishment would fight such a rule change to the death, and so it will never happen.

Archives

Resources & Links

Search