[fred]

Computer software advances over the years have, by and large, made it easier for TDs to do their jobs well. For example, when I started playing bridge, scoring matchpoint games was done by hand and this was a stressful, time-consuming, and error-prone process for many TDs. The development of scoring software has resulted in TDs not having to get involved in this anymore. Life has become better for both TDs and players as a result.

However, one relatively recent software development, that of fast double dummy solvers (like GIB) and the integration of such software into programs like BBO gives me cause for concern. I am concerned because correspondence with both TDs and players in BBO tournaments suggests to me that some TDs do not really understand what GIB is all about and depend far too heavily on GIB to help them to make scoring adjustments.

The purpose of this post is to try to explain my thoughts on this subject. It should be noted that I am neither a TD nor am I am expert in the Laws. For those of you who don't know me, what I am is a very experienced bridge player and one of the people who runs BBO.

It should also be noted that BBO generally does not get involved in how TDs of free tournaments do their jobs. If you are a TD of a free tournament and if you care about running tournaments that are as fair as possible, then you should probably read and pay attention to what follows. But we recognize that free TDs are volunteers and we greatly appreciate the time and effort they contribute to making BBO a better place for our members. Thus, if free TDs want to ignore what follows and continue to rely heavily on GIB to make their lives easier, that is their business.

This post is really about the difference between could, should, and would.

GIB, being a double dummy solver, answers the question "how many tricks could be made if everyone was able to see all the cards and if everyone played perfectly?".

Obviously these are big ifs - players cannot see all the cards and they do not play perfectly.

For an example, click on the following link and imagine that the round ends just after West leads the King of hearts against South's 6S:

Link to example deal

If you click the GIB button near the bottom of the window that opens you will be told that South could make an overtrick in 6S and that the defense cannot do anything to stop this. Probably you don't need GIB's help to see that - South can finesse against West's Queen of spades and later finesse twice against West's Jack-Ten of clubs.

But it would be a seriously bad ruling to adjust the score to 6S making 7 just because GIB says that 13 tricks are possible.

In general, your job when adjusting a score is not to concern yourself with what is possible or what could happen. Instead you should focus on what is likely or what would happen.

Where does should fit into this?

Suppose you are not only a TD, but that you are also a bridge expert. Suppose that you are able to figure out what you think is the best line of play in 6S. You can click the "Next" button repeatedly at the bottom of the window that opens when you click the link above to see how I think declarer should play (I could be wrong about this - I did not put a great deal of thought into the analysis of this hand!).

Essentially I am suggested that the way declarer should play 6S involves stripping the hearts and diamonds, playing one round of clubs, and then intentionally taking an anti-percentage play in the trump suit knowing that, if the percentage play ("9 never") would have worked then the trick declarer loses will come back thanks to an endplay.

On the actual layout, this line leads to 12 tricks, but it would not necessarily be correct for you to adjust the score to 6S making on the basis of how declarer should play.

Should is hardly more important than could when adjusting a score. What really matters is would.

Unfortunately it is the case that the question "how many tricks would declarer make?" is arguably the most difficult to answer. To make things worse, it is hard for me to imagine anyone writing software that could answer this question for you

But it is your job, as a TD, to do the best you can to answer this question. This will often involve trying to assess the skill-level of the players. For example, in the example deal, if you know the declarer is a world famous player, it would be reasonable to assume that he/she would play this deal as he/she should and adjust the score to 6S making. On the other hand, if you knew that the declarer was a novice, it would be reasonable to assume that it would not even occur to him/her to find the best line of play. The novice would put his/her faith in "8 ever, 9 never" thereby losing a trump trick. Later he/she would lose a club trick and go down in 6S.

So sometimes it will be the case that the skill-level of the player(s) should be a factor that you need to consider when adjusting a score.

Of course, especially in online bridge, it is often impossible for a TD to know *anything* about the skill-level of a random player which in turn can make it difficult or impossible to intelligently guess what such a player would have done in a given situation. All I can really suggest is:

- It is safe to assume that nobody on BBO takes as many tricks as GIB can take when looking at all 4 hands! What GIB has to say really shouldn't enter into the equation.

- If you have time then it can never hurt and often helps to discuss the situation with other TDs. Other TDs may be able to give you insight into both the skill-level of players you are unfamiliar with and the analysis of the deal in question.

- Try not to let you own ego as a bridge player or your natural bias of being able to see all four hands get in the way of your judgment. If a given play seems "obvious" to you, that does not necessarily mean it will be obvious to a random player (who can only see his/her own hand and the dummy and who may not play bridge anywhere near as well as you).

- I am not sure what the Laws have to say about this, but as far as I am concerned it makes sense for a TD to be able to ask questions like "how would you have played that hand?". Ideally it is best to ask such questions as soon as possible after the deal is over. Still, you should take the answers you get with a grain of salt - the player you ask will often have been able to see all four hands by then and may have had enough time to come up with an answer that is both credible and would result in a good score for his side. Even if you choose not to put a lot of stock in such answers, they may be useful for helping to determine the expertise of the player in question.

- The myhands web site can be used to tell you how well the player in question tends to score when playing on BBO thereby providing you with a rough guess as to how well the player in question plays.

Being a TD is far from easy and even the best TDs make plenty of poor rulings. For a variety of reasons, it is even harder for a TD to get things right in an online environment. IMO all that anyone can reasonably ask of our TDs is that:

- they try their best
- they are polite to the players
- they try to evaluate their own rulings objectively and try to learn from their mistakes

The politeness part is often a challenge because competition brings out the worst in many people - even people who are normally delightful sometimes turn into monsters in the heat of battle. Also, it is the often the case that the TD is placed in a no-win situation and that someone is going to be upset (and quite possibly behave abusively) regardless of what ruling the TD makes.

Hopefully it will make you feel a little better about such things if you keep this in mind: I know I speak on behalf of many 1000s of BBO members from all over the world when I say how much your service to our community is appreciated. Thank-you

This post may make it harder for you to do your jobs (since I am basically suggesting that you should not be relying on what GIB has to say), but hopefully it will also result in generally higher-quality rulings.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[blackshoe]

Very well said, Fred.

I'd add just a couple of comments, in no particular order:

Relying on GIB (or other such tools) to guide one to a ruling is akin to relying on the traveller. The latter has long been deprecated in f2f bridge, should be deprecated in online bridge as well, and the use of new tools like GIB for this purpose should also be deprecated. In the case of travellers in particular, you have no idea whether the bidding and play at other tables was identical, or even similar, to what went on or might have gone on in your judgement at the table concerned.

Note that in some jurisdictions, you can weight your score adjustment according to the (rough) probabilities you assign to various outcomes. For example, if you think the player concerned might make 6♣ 60% of the time, and go down 1 40% of the time, you can assign a weighted score of 60% of (for example, NV) 920, and 40% of -50, or +530 to the declaring side (and -530 to the defense). If you judge that defenders might double 10% of the time, you would weight (remember, I said "roughly") 55% of 920, 5% of 1090, 35% of -50, and and 5% of -100. You can't do this in ACBL sanctioned games, but if you're not running one of those, you can.

As for "obvious" I always remember an incident that occurred in my undergraduate quantum physics class. Nobel Prize winner Hans Bethe was the guest lecturer, and he wrote an equation on the chalkboard, turned to the class and said "and from this, obviously..." turned around and wrote another equation on the board. About fourteen hands went up. "It may be obvious to you, professor, but it's not obvious to me. Could you explain, please?" It took him an hour and a half.

Online, things happen more quickly than they do offline. Offline, the TD usually has time to consider (and to consult) on judgement rulings. Online, that may not be the case. This can lead to errors, but as my first Navy skipper told me "it's not always so important that you make the best decision, as that you make a decision". IOW, don't sink the ship because you did nothing while deciding whether to turn ten degrees to port, or fifteen. Or, to put it in bridge terms, don't spend all day on whether declarer would have played the hand this way or that way. Take your best shot and move on.

There will be times when what's written in your law book (or for that matter, in the laws for online bridge) does not match what the software requires or allows you to do. For example, in law, the round ends at a particular table when the players move to the next table (Law 8B1, 8B in the online laws), but the software requires the round to end when time runs out. Live with it. If it means giving a technically illegal adjusted score, well, that's better than throwing up your hands and cancelling the whole game.

[bluejak]

Very sensible posts, but if I may take a very small issue with one detail. While could is wrong, and should is often wrong, would may be wrong too!

Rather than look at a hand, just imagine one. It is played 400 times in 4♠ [and maybe a few times in other contracts]. Two players make 12 tricks, GIB makes 12 tricks, three players take 9 tricks. 395 times it makes ten or eleven tricks. Let us say looking at the hand you can see this. You can see the double dummy line for 12: you expect the players making 9 forgot to draw trumps. But 10 or 11: which?

Now in fact at the line that players are somewhat more likely to take makes 10, and a bolder line makes 11. If you happen to look at the results [I know you should not really! ] you will find they support this: 205 made 10 tricks, 190 made 11. Now if you were deciding what "would" happen you might think the right answer was 10 because that is the most frequent. But that is not the way to make a North American ruling, ie one were weighted scores are not allowed.

What you do is to consider all the reasonably likely results: in this case that means 10 or 11 tricks. Then you give the benefit of doubt to the non-offenders. So if the non-offenders are in 4♠ you adjust to them making 11 tricks: if they are defending you assume 10 tricks. I suppose what I am saying is that not only are could and should wrong, but I am afraid would is as well!

Best is to consider everything with a reasonable probability, then lean towards the non-offenders. But Fred's basic theme is correct: GIB or Deep finesse might help to avoid silly results, but it really does not help to decide likely scores to which to adjust.

[fred]

Thanks very much for your feedback and clarifications, David and Ed.

It is good to know that my sense of how a TD should think about such matters is mostly solid from a legal point of view

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[barmar]

Just to remind everyone: Double dummy solvers never guess a two-way finesse wrong, and they always follow the Rabbi's Rule when it helps. On defense they lead or underlead aces whenever it's best, and don't crash honors.

[Dealing_Don]

It is nice, though, to have GIB there to be used as a second opinion. In the heat of battle when faced with adjusting multiple boards, it's nice to have a quick second opinion in simple cases just to make sure you haven't missed anything. If there is a discrepancy between what a TD determines the result would be and what GIB says it could be, it is usually pretty easy to discover the line used by GIB and see if it is more reasonable than the one assumed by the TD.

Of course checking the line of play up to the point of cessation sometimes gives you a clue about which line the declarer and or defender might take. Often the line the defense takes is as important as the line declarer chooses.

As a director, I do not feel it improper to conclude that a board cannot be fairly adjudicated with a result and assigning averages to both pairs. When I do this I usually explain my reasoning to the players involved. e.g. "I cannot determine the outcome on board 6 because it involves guessing on a two way finesse" or "There are too many variables in the line of play to determine a fair result." If it is an IMP game and we are talking about overtricks, I usually stretch try to come to a result rather than "average it out".

It would be nice, though, to be able to assign NP (not played option) to the pairs involved so that it would not affect their percentage score for the game. This option would also be good for completely unplayed boards rather than assigning Avg= and pulling a pair's 60% game down to a 59.2% game (or if two averages from 60 to 58.3) . If we had this option I think there would be less "demand" for Avg+ boards due to "slow opponents". Until we have an objective measure of slowness, the requests to adjust unplayed boards to Avg+ will continue to flood TDs, especially in quicker games with low minutes per board.

[Mbodell]

Dealing_Don, on Aug 28 2009, 06:15 AM, said:

This option would also be good for completely unplayed boards rather than assigning Avg= and pulling a pair's 60% game down to a 59.2% game (or if two averages from 60 to 58.3) . If we had this option I think there would be less "demand" for Avg+ boards due to "slow opponents". Until we have an objective measure of slowness, the requests to adjust unplayed boards to Avg+ will continue to flood TDs, especially in quicker games with low minutes per board.

One other change that would make things much better IMHO would be for the end of round to work the way it does in F2F bridge and some other online sites. There if there is a pair that has started a board and is still playing it then they don't stop playing the board in question. They just lose time from the next round. And that way if there is a slow pair eventually they will have some round where they only play 2 boards instead of 3 because they are too deep in a round to start a new board. If you only routinely skip boards that are unplayed you will have far fewer boards to need adjustment (only MI/UI boards + maybe disconnects that couldn't get a sub).

I recognize this would take a change in software and may be difficult depending on how the pairing, scoring, and other aspects are programmed. However, I think this would be a much more preferable solution and would minimize the number of boards a TD would consider for adjustment and give them more time to both use GIB as a safety check, and also perform the lawfully correct procedures that bluejak and others describe.

[barmar]

That's how tourneys on OKbridge work. If you run long you're allowed to finish the board you're playing, but you can't start a new board within N minutes of the end of the round.

This doesn't work well in a Mitchell movement, because the opponents who are waiting for you also lose time, and that's not really fair to them. On OKbridge the tournaments are all Swiss pairs, and the late pairs are paired with other late pairs.

The other thing OKbridge does is that no one moves until at least 70% of tables are finished. So if there's a tough hand that most pairs have trouble with, everyone gets some extra time.

[blackshoe]

barmar, on Aug 31 2009, 11:12 PM, said:

The other thing OKbridge does is that no one moves until at least 70% of tables are finished. So if there's a tough hand that most pairs have trouble with, everyone gets some extra time.

The lack of that can lead to problems in f2f, too - I've seen games where half the field is nearly a full round ahead of the other half, because the TD doesn't control the movement, the players move when they're ready to move. And then they complain because not everyone is as fast as they are.

[Oof Arted]

As this seems to have moved on to software movements 2 questions arise

1 Why oh why does the Swiss Pairs NOT match pairs of roughly = scores each round that have NOT played each other before even though Clocked:: Pairs get upset when they play the same pair more than once

2 Why if you wish to run a Clocked non Swiss pairs does it feel necessary to split into 2 or more sections with 16 tables or more when if you run an UNclocked movement it does not

[tacheon]

Thanks Fred,
Your post was informative and illuminating. I am a very new TD just in the "Free" section. I would not dare to get involved in the "Money" area.
My mentors just accept GiB's analysis as indeed I do too. I am a poor player myself and although I have completed BM2000 ( a fabulous program) there are quite a few hands I just cannot make,even after studying the explanations very carefully.
As a result of my inadequacy in this area I just have to accept the GiB result when adjusting scores . But then there is no money involved and although there is the occasional opp that may object I , like my mentors, just say that the computer is never wrong and leave it at that.

[drahcir]

Fred,

Can I compliment you on this thoughtfull explenation. Your attitude in writing is something I can only hope to master someday (I am 54 years of age ).
GIB is very usefull in the post mortem and it helps to understand the complexity of our "hobby". Like GIB, many persons watching or commenting the game with all cards in sight have problems to keep their objectivity, unlike GIB they even make mistakes withouit realising it.

Regards,

Richard van Houten (drahcir)

[helene_t]

bluejak, on Aug 27 2009, 08:30 PM, said:

What you do is to consider all the reasonably likely results: in this case that means 10 or 11 tricks. Then you give the benefit of doubt to the non-offenders.

Typically there will be no obvious "non-offending" side. The table got delayed which may be the fault of one or both of the pairs. You don't know whom.

A split-score (giving both parties the benefit of the doubt) seems unreasonable since at least one of the pairs, possible both, were at fault. You don't want to encourage stalling by awarding it with more than 100% MPs to the two pairs combined.

A weighted score is obviously the best but the software does not support that. I would be tempted to estimate the median of likely scores but that presumably has no legal basis. What to do then? Are we back at the most likely score?

[bluejak]

The thread seemed to be about rulings. If a table is slow and a board is taken away mid-board that is of course illegal, but in online bridge some practical solutions are illegal.

Probably best is to work out a code specifically for this case, and follow it. But if my opponents were slow, a board was taken away part way through, and it was assumed I was not going to get a good score I should probably sue!

Far better if at all possible is to control slow play by some legal means.

[nige1]

A devil's advocate would point out that double-dummy analysers open up the long-term possibility of objective rulings. For example, if Gib has sight of only two hands, he will usually come up with a plausible if not ideal line of play that could be useful for assessing likely results. Directors are human and their judgement is imperfect too. For example, they sometimes exercise 20-20 hind-sight, when judging an action to be gambling or a serious error.

Incidentally, when assessing or weighting scores, why shouldn't the director take into account results at other tables? Quite often the events at other tables will be similar enough for fair comparison. When they are not, real results are still likely to be a better indicator than a director's superficial analysis. Even if the director chooses to ignore most of the real results, he should take notice of some. For example, he may be forced to accept that a result he rates as impossible, has occurred before (even several times).

[bluejak]

An adjustment is based on the position before the infraction occurred. How does the Director know the position was the same at the table where the other result occurred?

[Geoduck2]

I enjoyed reading this topic. As a very new TD with limited bridge play experience I am struggling with this part of the task. I am wondering if there is any place that I can go to read about how decisions and equitable adjustments are made. One example that I comes to mind immedaitely is what is the correct adjustment when, on the last board of a tourney, after the bidding in complete, the lead made, and the dummy down, one of the remaining players (delcarer or defender) leaves or gets disconnected and does not return? I figure that it's too late to sub and too early to use the GIB so what is left? A == ?

Also, is there such a thing as a psyche on an opening bid? And if so could someone please define it for me? I am confused. Thanks,

[blackshoe]

Geoduck2, on 2011-April-08, 19:31, said:

I enjoyed reading this topic. As a very new TD with limited bridge play experience I am struggling with this part of the task. I am wondering if there is any place that I can go to read about how decisions and equitable adjustments are made. One example that I comes to mind immedaitely is what is the correct adjustment when, on the last board of a tourney, after the bidding in complete, the lead made, and the dummy down, one of the remaining players (delcarer or defender) leaves or gets disconnected and does not return? I figure that it's too late to sub and too early to use the GIB so what is left? A == ?

Also, is there such a thing as a psyche on an opening bid? And if so could someone please define it for me? I am confused. Thanks,

Read the International Bridge Laws Forums, which are down the page on this very site.

I've not directed online so there may be some cultural thing, but... when a contestant (a pair in a pairs contest) is in no way at fault for an irregularity (as when one of their opponents leaves or disconnects) and you award an artificial adjusted score, the contestant in no way at fault gets Avg+. A contestant directly at fault gets Avg-. A contestant only partly at fault (this does not mean that only one player of the pair is at fault!) gets Avg. See Law 12C2. If a player was disconnected, I'd say the contestant (the pair) is only partly at fault (but I would not balk at a ruling of "directly at fault"). If a player willfully leaves, the contestant is directly at fault. If you can't tell which, I'd lean toward "directly at fault".

The definition of a psych is in the laws:

Quote

A deliberate and gross misstatement of honor strength and/or of suit length.

It is certainly possible to psych an opening bid.

[barmar]

Geoduck2, on 2011-April-08, 19:31, said:

Also, is there such a thing as a psyche on an opening bid? And if so could someone please define it for me? I am confused. Thanks,

If you play 5-card majors and open 1♠ with only 3 spades, it's a psyche. If you play sound opening bids, and open at the 1 level with less than 10 HCP, it's probably a psyche. Why do you think opening bids are any different from anything else when it comes to psyching? A psyche is when you deviate significantly from your agreements; if you have agreements about opening bids, you can deviate from them.

[Geoduck2]

The definition of a psych is in the laws: It is certainly possible to psych an opening bid.
[/quote]


Isn't there something additional about intent to mislead the ops?