[david_c]

What a wonderful program FD is. I've been playing around with it a lot for the last week or so, and I now have about 50KB of system in FD format. It'll make a huge difference to our bridge playing once it is set up so that opponents can see the explanations.

Right, now that I've said how wonderful it is ... I'm afraid that the more I use it, the more I wish that it had been set up differently. What follows is a plea for all the mandatory parts of the definitions to be dispensed with, leaving just the "description".

There are a few specific problems that I have with each of the mandatory parts of the definition. But one general problem is that, because they appear every time, there is no distinction made between information which is important and information which is unimportant, or even very important. For example, if you were playing the Fantunes system, having "Forcing" appear in the top-left hand corner of your 1♥ opening isn't really good enough. What it needs to say is "FORCING <--- look". It follows that any important information ought to be repeated in the "description" anyway. (Also, when viewing the list of bids, you only see the "description".) Meanwhile, all the information which is unimportant or irrelevant just makes life harder for whoever it is trying to understand the explanation.

Now let's look at each of the parts of the definition in turn.


Disposition

Sometimes the disposition nicely sums up what a bid means. For example, 1♠:2♦ playing 2/1 - this is great, I can write "forcing to game" and this does it nicely. Except, I'm going to repeat this information in the description anyway. Otherwise the reader might not realise that this is what really defines the bid.

Sadly, a more common situation is that there is no disposition which is useful. You find yourself thinking, "Well, it just shows hearts, doesn't it? What shall I put? I suppose it's forcing, because my partner has promised a rebid. OK, I'll put that. But that's not very helpful, it's too obvious. The bid is just ... a description of the hand."

This isn't solved by adding more dispositions. In fact, having lots of dispositions can be a bad thing, because it only works if everyone will understand what they mean. "Puppet" may be a perfect explanation, but it's no good if the reader isn't a bidding theorist.


Minimum / maximum length

For a start, this is too inflexible. For example, an opening bid in third seat might be "5+ cards, occasionally a good 4". Similarly, a natural bid might typically promise 4 cards, but occasionally be the best bid with a 3-card suit. In a large number of cases, the correct explantion is "natural", and the suit holding expected follows logically from that. Giving a specific number is therefore often misleading. The reader is unsure whether the length is an integral part of the agreement, or just a number decided on by the writer of the FD file because they are forced to do so.

It's a nightmare trying to work out the suit lengths allowed in complex auctions, particularly for artificial bids. For example, in writing the file for Keri over NT, a lot of opener's artificial bids promise a suit length of "2-5". But sometimes a 5-card suit has been ruled out, so it changes to "2-4". And it's important to get these things right, because if the opponents play you for "2-4" when actually this is an error and you have 5, that is clear misinformation. But the suit lengths aren't actually part of the convention at all - it would be so much better if they were left out.

And yet there are some artificial bids where the length promised is relevant. For example, we might have a 1♣ opening promising at least a doubleton. As usual, the problem is that we are forced to give lengths every time, so it's not clear to the reader when they are important.


Possible outcomes

I think enough has been said about this one before, but as a reminder:

• It makes no sense to people who don't use FD ...
  
• ... and isn't exactly clear to those that do.
  
• Denominations are usually removed from the "possible outcomes" because they are logically impossible, rather than because the call says so.
  
• It doesn't even deal properly with suit agreement, because it's possible to agree one suit and then go back to another.
  
• It takes a huge amount of work for the writer of the file.
  
• It provides a large amount of totally useless and very distracting information.

I notice that I'm not the only person who gave up on trying to list the possible outcomes in the FD standard systems forum.


Artificial

This one is OK, if it is used as a flag like an alert. I don't have a problem with it. Except that it's quite easy to forget about, for some reason.


Best wishes to anyone who actually read all that

[barmar]

I agree with most of what you say. However, I don't think it's quite as bad as you make it out to be. For instance, in the case of suit lengths, I think everyone should understand that this is the usual, expected lengths, not a guarantee, like the suit length check-boxes on the ACBL convention card. If you play 5-card majors, you normally have 5 cards in the suit when you open a major, and your partner will respond accordingly.

All experienced bridge players understand that it's sometimes necessary (or at least reasonable) to deviate from your systemic agreements. There's no bid that accurately describes what you have, so you find the "best lie". It's not gross enough to be a psychic bid, but it doesn't agree exactly with your conventions. So no matter what mechanism you use to provide system information to the opponents, these inaccuraces will occur, and it's just part of the game.

I suppose the problem that's unique to FD is that all these predefined meanings make things look more cut-and-dried. I think we just have to realize that just because we're storing all this information in a detailed database, we're still not playing as automatons that work off the database. It's a more detailed description of agreements, but you're still allowed to use judgement to deviate.

[kfgauss]

david_c, on Sep 24 2005, 02:43 PM, said:

It's a nightmare trying to work out the suit lengths allowed in complex auctions, particularly for artificial bids. For example, in writing the file for Keri over NT, a lot of opener's artificial bids promise a suit length of "2-5". But sometimes a 5-card suit has been ruled out, so it changes to "2-4". And it's important to get these things right, because if the opponents play you for "2-4" when actually this is an error and you have 5, that is clear misinformation. But the suit lengths aren't actually part of the convention at all - it would be so much better if they were left out.

And yet there are some artificial bids where the length promised is relevant. For example, we might have a 1♣ opening promising at least a doubleton. As usual, the problem is that we are forced to give lengths every time, so it's not clear to the reader when they are important.

I would just leave the suit length blank when no specific info has been given regarding it. Perhaps others disagree, but this seems a reasonable method. For example, when you transfer, nobody needs to see that your 2D bid caps the number of diamonds you hold at 8 (if only 5+ hearts is guaranteed) or maybe 6 (if hearts > diamonds is guaranteed). Obvious negative inferences (ie if I have 5 hearts I can't have 9 diamonds) aren't something you're required to tell your opponents.

I also would ignore the possible outcomes. Perhaps if there's some specific situation where it's particularly useful to know you could fill it out then, but then it will be noticed as it's different from the default setting.

Andy

[barmar]

I'd go further and wish that the suit-length section allowed you to specify which suit when you click the Artificial box. For a transfer, the information about 5+ in the next suit should be in the predefined section, not just the description. It should also be possible to enter two-suiter bids this way, rather than relying on the descriptions. The hardest type would be bids with an unspecified suit, or an OR (e.g. a bid showing either both majors or both minors, as in CRASH).

[david_c]

kfgauss, on Sep 26 2005, 10:18 PM, said:

I would just leave the suit length blank when no specific info has been given regarding it.

To some extent I agree - it would be good to have the option of leaving it blank. But, why go through all the trouble of adding this sort of flexibility to the mandatory parts of the definition, when you can achieve maximum flexibility by putting everything in the "description"?

Lack of replies in this topic makes me suspect I'm not going to win this one. But is there anyone out there who likes the current arrangement and can explain to me why it's better to have a fixed format? (Bonus points for sticking up for the "possible outcomes" ... )

[barmar]

david_c, on Sep 27 2005, 08:03 AM, said:

But is there anyone out there who likes the current arrangement and can explain to me why it's better to have a fixed format? (Bonus points for sticking up for the "possible outcomes" ... )

I'm not sure I can defend the specifics of the current design. But the general approach of having standard fields seems worthwhile, because it encourages standardization in the attributes that are common to many conventions and treatments. It's the same philosophy that prefers standardized convention cards rather than just pages of system notes.

But it has its drawbacks, much like convention cards, when meanings don't fit neatly into the standard fields. For instance, the ACBL convention card is OK if you play a mostly natural system, but I'll bet it's practically worthless for Hrothgar.

The problem with the current architecture, as I alluded to above, may be that it doesn't have enough standard fields. Instead of just a single suit-length field, it could have fields for "usual length" and "possible length". E.g. in the auction 1C-1S-2S, opener normally has 4 spades, and responder usually continues as if he did, but occasionally will have to support with 3 because his hand was unsuitable for any other bid.

Not to cast any aspersions on Fred's work, but if I were designing this program I'd be looking at how bridge *playing* programs store their convention cards. When I'm playing against GIB bots, I can click on one of their bids and it describes it in a concise format, which I assume has to be encoded somewhere in the program, it's not just free-format text, since the program has to use that same database to select the bid in the first place. Something like that encoding would probably be a good idea to adapt for FD.