For the last few weeks I've been developing a web-based bidding program that would let you input two hands and generate a proper Standard American bidding sequence. This could be of great help to new players who seem forever confused (and perhaps rightly so) by modern bidding methods.

Ironically, the more I realize how intricate bidding judgement it is, the more I am convinced it can be quantified. All this talk about how a computer will never play good bridge - I don't believe it. The field of artificial intelligence has more important issues to address right now than playing cards; otherwise millions of dollars in funding would be going into bridge research and a solution may have already been created by now. You may point out that chess-playing computers are already first-rate, but just because bridge is a tougher task does not mean it is impossible.

Anyhow, I wanted to share some of my experiences to date. The first order of business was hand evaluation, and that meant fixing the old 4-3-2-1 point count system. Everyone says that queens and jacks are overvalued, right? So I accordingly devalued them in my methods. (I also weighted tens slightly.) After some preliminary testing, however, I realized that it would be better to separate suit contract evaluation from no-trump evaluation. I think we would all agree that soft values have greater potential in no-trumps than suit play.

So I isolated no-trump evaluation and have been focusing on that ever since. I am using a 15-17 1NT, but that doesn't mean good players don't occasionally open 1NT on great 14-counts and poor 18-counts. Consequently, I started working on evaluating suit texture and tenaces. Clearly it's better to have an A-K in one suit than in two.

What though, about semi-balanced hands? Everyone finds themselves wanting to open 1NT on 2452 shape occasionally. How good do the doubletons have to be? Personally I think they need to combine for at least 3 points, although this has not yet been finalized. Something to worry about later, along with NT hands with a 5-card major.

Another thing to consider is suit length; not just adding a point for a 5-card minor, but for even tripletons and 4-card suits. A worthless spade tripleton is still worth a little more than its doubleton counterpart, because the chance of the opponents running 5 spades on you decreases. Similarly, QJ tight is worth less than QJx, which is a guaranteed stopper.

And even the points themselves may carry different weights depending on whether you are opening or responding to 1NT. Let's assume that the average 1NT opening is done on 16 points (although it's probably closer to 15.8). That means the other 3 hands share 24 points, or 8 points each. Therefore, partner has the average equivalent of 4 queens.

If you are responding to a strong 1NT, though, partner has the average equivalent of 4 aces. Now suppose you hold KJT9 in a suit. Opposite a queen you have got 3 tricks. Opposite an ace you have some play for 4. It seems to me that you would therefore much rather have KJT9 in responder's hand than opener's. Should this tenace be weighted more for responder than opener?

In conclusion, here's a hand from the 1997 World Bridge Championships in Hammamet, Tunisia. Alain Levy writes of one of the boards when France played Poland:

"Do you try for game holding: J1097 A65 108 Q1095, when partner opens 1NT (15-17)? No, said Romanski, and passed. Mouiel drove to 3NT. Balicki was upset when he saw dummy, saying 'You have forced to game with only 7 HCP.' 'I think my hand is worth 9.5 HCP,' answered Mouiel. And he was right.

Board 49. Love All. Dealer North.

K3 K97 A63 J7642

J1097 A65 108 Q1095 A4 Q103 QJ974 AK8

Q8652 J842 K52 3

"Nine easy tricks, while in the other room Chemla/Perron played 2 after a Landy 2 overcall; 7 IMPs to France."