“Thus the computer doth make cowards of us all.” So might William Shakespeare write if he were to view the current chess scene. We sit down with our silicon friend to select strategies and openings, and we close doors.

Our friend sits beside us, looking over our shoulder, and he makes comments on our move selection. What’s the problem with that, I hear you ask? By his very nature, our silicon friend cannot be in doubt. He will offer his commentary, tell us we’re doing well, doing badly, or hanging in there. He’s always got an opinion. And sometimes he’s right, sometimes he’s wrong, but he’s never in doubt.

I’m reminded of a line from Douglas Adams: “We demand rigorously defined areas of doubt and uncertainty!” It sounds silly, but there’s some point to it. It’s in these areas of doubt and uncertainty that the best chess is often played.

Back in the day, when a human annotator reached a complicated complex position that he couldn’t see the end of, he marked it with the “unclear” symbol. That let players and future analysts know there was something undiscovered there, and it could prove fruitful ground for the player to explore. It was a position either side might hope to make something of.

Our silicon friends today don’t suffer from that sort of humility. Their egos don’t permit them to say “I’m just not able to see this clearly enough to know.” Instead, they cheerfully spout off with an evaluation whenever we ask, whether they can clearly see the answer or not.

This isn’t exclusive to chess. In the financial analysis world, this syndrome was termed “visi-knowing” (after the first really popular spreadsheet application, VisiCalc). It meant you had a detailed financial sheet in front of you, and all the numbers squared up, with a mountain of detail. But, in part because of the detail, it wasn’t clear to the human reading the report just what was actual, and what was fiction.

Correspondence players will tell you, computers will be happily churning along, sure they’re doing alright, until after they’ve arrived in a position they cannot win; they just don’t see their fate coming, even though the human player has been sure for several moves now.

But when we mere mortal OTB players prepare, we take the silicon monster’s word, and close down the line of research, a line we might have followed up on had we not had a friend whose omniscience we relied upon.

Case in point: The Blackmar-Diemer Gambit. No computer likes this opening. Yet it, better than any other opening system I know, exemplifies the phrase “good practical chances.” The positions in it are filled with tactical opportunities for both sides, with complicated play which ensures that at least one side, and quite possibly both, will go horribly wrong.

Yet I once had a computer walk the tree for a 25,000 BDG game database, making the move with the best result for White or Black, depending upon which side was to move. Black won in all variations, meaning no matter what White’s move choice had been, Black had a variation which had resulted in a won game for him. There were plenty of improvements available, though, so for grins I started looking for them (starting in the Teichmann Defence, which a friend assured me held the secret to refuting the BDG).

While I was slogging through there with my silicon friend, I realized something: The likelihood of me, or any of my opponents, finding some of these continuations were practically nil. Some of the “only” moves I was finding were moves only a computer could love. And then came the epiphany.

How many players were crunching away at positions in the BDG, or any other opening, for that matter, and stopped analysing a line because their computer said it slightly favored the other side? How many variations which would have pitted one player’s skill and calculating ability against another have been ignored, because the computer said “it doesn’t work,” whether it actually did or not?

We see draw rates of 55% at the GM level. Maybe this is one reason why. Today those dubious lines go unplayed, even though they might win. I wonder how many future Mikhail Tals are being smothered, because their imagination gets stepped upon by the computer’s verdict of “=”.

And I think back to Kramnik-Kasparov, and I wonder: if the match were being prepared today, would Kramnik have had the courage of his convictions and fly in the face of the silicon beast’s “slightly worse” pronouncements to build his Berlin Walls anyway?

What have we already lost, and what will we lose in the future, because of the computer’s oblivious analytical certainty?

This entry was posted on Sunday, September 3rd, 2006 at 12:57 pm and is filed under Chess Software, General. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.