Saturday 5 July 2008

Tarting up the top tools

Having given 3NT values-for-four a good long workout over the last year or so, we have finally given it the thumbs down.

Let me explain first why it was there in the first place. In a sequence like 1 - 2NT, opener with extras needs to say 'opposite your raise to at least three, I have enough for four'. But if he has nothing to splinter at the three-level, without values-for-four he would need to bid 4.

What's the problem, you may ask? Well we have spent quite some time engineering Chilli so that keycard asks can be made below game, so that with an unfavourable response we may still be able to stop in game or 4NT, rather than be forced to the five-level. In the given sequence (and many others like it), responder is unlimited and may well be interested in going to slam. If opener does indeed rebid 4, responder can no longer ask at the usual safe level.

So 3NT was given a conventional meaning of values-for-four in some fit auctions. The 'some' is the root cause of our eventual disillusionment. It was clear from the start that 3NT should be to play in some other fit auctions, most obviously minor fit auctions, but also situations where partner could have no interest in going to slam, or where the fit was 5-3 and the hands were balanced. Each exception was logical and well defined and tied into the concept of hard and soft fits.

But the end result was too much thinking about rules at the table – just what Chilli is not about. So we are abandoning it, and 3NT returns to its former glory of always being to play in all types of auction.

Our sense of relief at this outburst of simplicity has been enhanced by the re-discovery of the extra nuances this allows. Consider this sequence, for example: 1NT - 2; 2 - 2NT; 3 - 3; 3NT. Opener has shown three hearts and an empty diamond suit, and has suggested 3NT as an alternative to 4, so is probably 3334 or perhaps 3343. This leaves responder ideally placed to choose. (We do better here than with standard transfer technology, which would go 1NT - 2; 2 - 3NT, and now it is opener that has to decide, with less information with which to do so.)

We've retained the idea of soft and hard fits, but they now differ only in the meaning of the three-level side-suit bids – empty suit (soft) or splinter (hard). That also means that we can re-classify some fits as hard when the only reason for their being soft was that 3NT needed to be to play.

To minimise the downside of all this, we have clarified the rules about the four-level bids for when partner (or the opposition) eats up our space. When there is only one side-suit left at the four-level, it is the keycard ask – we don't need a values-for-five. When there is no side-suit left at the four-level, 4NT is the keycard ask.

Then we aim to get the most out of the keycard ask responses. Usually we use the Chilli three-step responses, which generally perform better than the traditional four steps. But when the ask happens to be exactly one step above the trump suit, four steps are more effective, and we use those. (Five minutes with pencil and paper should convince you that this combination is optimal.)

So in our first example sequence, 1 - 2NT; 4, 4NT is now the keycard ask and the responses are zero, one, two no queen and two with queen. Does that sound vaguely familiar?

Alan Williams
July 2008

The Chilli bidding system is described and defined at chillibidding.org.