[Flame]

Anyone read this book by mike lawrence ?
Is it good ?
Does it really go against the law or just show when and how to make adjustments ?

[ArcLight]

The book is not yet out. I ordered it in advance, but Mike Lawrence said it should ship in January (that the publisher had some problems).

I have not read it, but I don't think its just a rehash of the LAW. I get the impression that the author is not as strong a believer in the LAW as others.

I'm very curious about its contents myself. Considering how well the LAW is known, writing an "anti-LAW" book seems controversial. There will surely be many people looking to test out the ideas in this book. Right or wrong its a high profile topic, and will certainly get notice.

[whereagles]

I tend to agree the law of total tricks is better used as a guideline than gospel.

[Cowology]

whereagles, on Jan 1 2005, 03:52 AM, said:

I tend to agree the law of total tricks is better used as a guideline than gospel.

I tend to feel that way about everything relating to bridge.

[mr1303]

Just been reading an extract from "Bridge, Zia and me" by Michael Rosenberg

Larry Cohen has a dream, where he ends up in a room surrounded by 3 witches. He holds:

AKxxx
xxxx
x
xxx

bidding goes as follows (RHO opens)

(2D) P (2NT) P
(3D) P (P) X
(P) 3S (4D) P
(P)

and Larry has a think. He decides to bid 4H (to be flexible) and his partner corrects to 4S, which ends the auction

Dummy comes down with:

QTxx
AKQ
xx
QJTx

Left witch cashes the AK of clubs, and her partner plays the 9 then the 4, and a 3rd round is ruffed, then the A of diamonds cashed. 1 down.

The full deal was:

---

Left witch says "19 total trumps, 17 total tricks".

"But...." began Larry, before being promptly told to shut up

Later Larry holds the same hand again, and the bidding proceeded to follow exactly the same path, except this time left witch doubled.

"You don't learn from your mistakes do you?" said left witch

The deal is played a third time. This time Larry passed 4D

The full deal was:

---

Both 4S and 5D are cold

"19 total trumps, 21 total tricks" said the witch.

Larry woke up.

[cwiggins]

The book can now be ordered through through at least these three sites:

* Baron Barclay www.baronbarclay.com (item 5077 $17.95 plus shipping and handling)

* Mike Lawerence's site: Mike Lawrence order page

* From the site Lawrence and Wirgren have set up about the book: www.newbridgelaw.com (click "Order our book")

As of January 12, Amazon did not have it for sale.

[mike777]

Got the book, reading now.

[MickyB]

Also carlritner.com, the ACBL Library's book sale. $15.95 plus P+P. It is well worth a look for those outside the US as well, particularly with the current exchange rates.

[Chamaco]

cwiggins, on Jan 13 2005, 04:27 AM, said:

* From the site Lawrence and Wirgren have set up about the book:  www.newbridgelaw.com

From what I could read on this site, their claim that the LOTT works only in 35-37% of the deals is based on an adjustment of Matthew Ginsberg's statistical analysis.

According to Ginsberg:
- total tricks = total trumps in 40% of the deals (adjusted to 35-37 by the authors)
- total tricks-trumps difference is exactly +/- 1 trick in 46% of the deals
- total tricks-trumps difference is +/- 2 or more in 13 % of the deals.

So, authors' claim that LOTT works in 35-37 % of the cases is a a little "trick" , maybe to advertise the book.

Actually, anyone using the Law, accepts a +/- 1 total trick deviation, which is also included in the undetermination of the knowledge of how many trumps the opponents actually hold.
On average, a +/-1 *total* trick, is 0.5 trick for each side, so that's where common sense, adjustments, hand evaluation and "feeling" come into play.

If we accept a +/- 1 total trick deviation, Ginsberg's stats reveal that LOTT approximates by *at worse* +/-1 TOTAL trick about 86% of the hands (or 83% using Lawrence & Wirgren adjustment) , which is quite a high percentage, IMO.

[ArcLight]

I'm halfway through the book.

The author methodically demolishes the LAW, and especially what Larry Cohen says in his 2 books. (he refutes each of the LAW rules such as moving a card from East to West not affecting the total number of tricks, or adding a trump inceasing the total number of tricks).

The LAW as Vernes described it is generally accurate over a large number of hands, but it has a very high variance, meaning that frequently it will be off by 1 or 2 tricks in either direction. Its more accurate with balanced hands, and when the total number of trumps is low, and less accurate the greater the number of total trumps.

44% accuracy on average, with 56% accuracy with 14 total trumps (while being off 2+ tricks in either direction 5% of the time), dropping to 30% with 20 total tricks (while being off by 2+ tricks in either direction a quarter of the time!!!!).

The author demonstrates how moving cards around does indeed make a difference in the number of total tricks, in contradiction to what the LAW says (and Larry Cohen wrote). Adding a trump can make no difference at all. And which partner is the declarer also can make a difference. Lots of factors that throw the LAW off.

The authors approached this as a scientist would. They didn't just say "based on looking over some hands" . They used a sample size of 1 million (1,000,000) hands to get a frequency distribution of total trumps.
Then they analysed 2,000 hands to get an exact count of total tricks.
(While a scientist would prefer more than 2,000 hands, it is probably large enough that the numbers are accurate to within 1% or so. And I wouldn't be surprised if more hands are eventually analyzed to get a slighly more accurate table)


What does make a difference? ....

DISTRIBUTION!!! And working HCP.

(no surprise there)


(its a fast read, I'll have it done in a day or so)

[Chamaco]

I ordered the book too (not here yet).

Quote

The authors approached this as a scientist would. 

I really hope their approach is MORE scientific than their way to QUOTE other people's data.

Any scientist knows that a rule holds true +/- some error.

So, to say that "the law is right only 37% of hands" is a tricky way (because it means *exactly right*) : using the same Ginsberg's data (those quoted in the book advertisment, and in the web page), one could as well say "in 86% of the hands the law approximates total trick by a max deviation of 1 total trick" ...

[ArcLight]

>I really hope their approach is MORE scientific than their way to QUOTE other people's data.

I'm not aware of any of their quites being inaccurate.

>Any scientist knows that a rule holds true +/- some error.

Their proint is that while the LAW is accurate on average, its not nearly as good an evaluation tool as Larry Cohen has said. (they don't beash him directly, they let the facts speak for themselves). Larry Cohen wrote in one of his books that the LAW was more accurate than experts judgement, and that is something that I think Mike Lawrence would object to.


>So, to say that "the law is right only 37% of hands" is a tricky way (because it means *exactly right*)

The book jacket (back of the book) says the LAW accurate on average 44% of the time. But the LAW is less accurate than that as there are more total tricks.

In addition, it's not always the case that the right trump suit is selected.
The author makes a good point - lets say your side has 9 spades and 10 Clubs, making 10 tricks with either suit as trumps. You want the game bonus, so you prefer Spades to Clubs, but thats not the Lawfully correct suit.


> : using the same Ginsberg's data (those quoted in the book advertisment, and in the web page), one could as well say "in 86% of the hands the law approximates total trick by a max deviation of 1 total trick" ...

It depends on the # of total tricks. In general the LAW is 95% accurate with 14 total tricks, meaning off by 1 or fewer tricks. Its far worse with 18 total tricks.

The author points out that underbiding, where you win the acution is not a problem. But if you over bid by 1 and are set an additional trick (taking a bad save) it can have a big impact on the score.


LAW accuracy (from I Fought the LAW of Total Tricks)

# of Frequency Exactly
total tricks Percentage Right
14 10.5 56
15 10.5 42
16 27 44
17 23 36
18 16 36
19 8.5 34
20 3.5 30
21 1
22+ 0.3


(I wouldnt be surprised if these numbers would change a percent or so with a larger sample set than 2,000 hands. The author also says so. Maybe in time this will happen)

[Chamaco]

Quote

I'm not aware of any of their quites being inaccurate.

At this web page, one of the authors (dunno if they wrote it together) says:
http://web.telia.com...27101/lott.html

"One reason why the Law has been so popular is because it is so simple to use. But simplicity is one thing, accuracy another. In the statistics section you will learn (if you didn't know it already) that at the table the Law predicts accurately no more than 35-37% of the time. "


Then at this other web page they talk of statistics and things look quite different:
http://web.telia.com...27101/stat.html

"In preparing for our book, we did a lot of statistics by looking at thousands of hands in many different ways. But before us, Matthew Ginsberg had done a large study published in The Bridge World (May 1996). He had let his double dummy engine (the same that he has based his very strong computer program GIB on) analyze almost 450,000 deals in order to see how accurate the Law of Total Tricks really was.
The result of his study showed these things:
the average difference between total tricks and total trumps is 0.05
the average error per deal is 0.75
total tricks equals total trumps on 40% of the deals
on 46.9% of the deals there is a difference of one (up or down)
on 13.1% of the deals there is a difference of at least two (up or down)

........................................

So, for practical purposes, to say that "The Law of Total Tricks" is right on 40% of the deals is a little too high. 35-37% will be closer to the mark. Or, put differently, slightly more often than every third deal.
"


So, the trick is using in the first sentence
"In the statistics section you will learn (if you didn't know it already) that at the table the Law predicts accurately no more than 35-37% of the time."

So, the key is the term they use, "ACCURATELY", which is not accurate :-)
They should use the term *EXACTLY*.
If they want to introduce "accuracy", they should give a +/- range.
Otherwise it is just a way to play with number to make it appear the Law works less often than it does...

Indeed, from the same number they quote, it appears that on those deals the LOTT predicts the number of total tricks within a +/- 1 total trick about 83-86% of the hands, not a bad percentage indeed...

[inquiry]

Chamaco, on Jan 13 2005, 10:58 AM, said:

I ordered the book too (not here yet).

Quote

The authors approached this as a scientist would. 

I really hope their approach is MORE scientific than their way to QUOTE other people's data.

Any scientist knows that a rule holds true +/- some error.

So, to say that "the law is right only 37% of hands" is a tricky way: using the same Ginsberg's data (those quoted in the book advertisment, and in the web page), one could as well say "in 86% of the hands the law approximates total trick by a max deviation of 1 total trick"...

I will order the book today as well. But as for the law, even if it was "more" precise than it is, that precision is hardly useful, because it is impossible for you to know with any certaintly the number of trumps both sides hold. The "law" as applied in the real world also includes a number or well known adjustments ("purity", doulbe-double fit, voids).

However, I look foward to reading the "new law" espoused in this book.

Ben

[arrows]

My humble opinion:

Although the claim of "LOTT is more accurate than expert judgement" is
argubly an exaggeration, I think predicting the EXACTLY total trick about
40% of the time make it quite accruate a tool in the game of bridge.
and generally correct with a deviation of 1 trick over 80% of the time
make it practically very useful.

I have been seeing many players making poor decisions which I believe
they wouldn't have made, were they aware of it was a violation of the LOTT.

I am not sure whether studying the LOTT isolately, without put it into a
context of auction, is a scientific approach or not, but I am pretty sure
it is not the bridge approach.

Since the LOTT often off by about 1 trick, we need frequently make some
adjustment on the "raw" prediction. Fortunately, there are always some
bidding context for us to evaluate our hands from other perspectives,
such as, shape? losers? fit in side suit? wasted high card? etc. we can then
form an evaluation on these factors and choose to bid aggressively (making
a positive adjustment to the "raw prediction") or conservatively (making
a negative adjustment to the "raw prediction").

Therefore, it is usually not hard for an average player, who undertands
all the basic concepts, to make better decision by following the LOTT and
the adjustment routines. And IMO, with ordinary hands, if you think you made
a "lawful" decision and it seems not worked well, chances are you have
overlooked some adjustment factors.

It is well known that LOTT is far less accurate when dealing with wild
distributions, off by 2 tricks or even more. And there's no reliable guidence on
how to make adjustment of 2 tricks or more.
But with wild distributions, it's more or less a guessing game any way,
even for world class players. Most bridge veterans could recall the hand
that Hamman made an unfortunate lead and the french team made slams
at both tables.

Demanding LOTT working with wild distribution just makes no sense. why?
because eventually we use these tools in the CONTEXT of some bridge AUCTION.
With wild distribution, chances are the auction will start at
, or immediately reach to the game level or above, at which,
most of the time, we are not even given a chance to investigate what is our
trump, why should we be worried about the accuracy of the LOTT?
In these cases, if I do know what the trumps are, I will think of following
Grant Baze's advice, "with 6-5, bid one more".

In conclusion, I think it's safe to say LOTT is a good start point to consider how
high one should bid in competition, and make the necessary adjustments along
the way, and this part, is the art of experts. So for me, I'll say I wouldn't want to
make a decision that crudly violates the LOTT unless I am 100% sure what
's going to happen next and prepared for it, like having a well-calculated escape
plan when partner trying to get the throat of me after the game.

[RonH]

Jusy received the book. It is not a light read . Takes a much more scientific approach in presenting arguements, pro and con.

Ron

[ArcLight]

>Jusy received the book. It is not a light read . Takes a much more scientific approach in presenting arguements, pro and con.

Ron, compared to Mike Lawrences other books I find this a much faster read. It is not a light read compared to Points Schmoints. But compared to Mike Lawrences Overcalls or Opening Leads, its a breeze.


A chunk of the book consists of hands where the authors show how the law fails when the hand is slightly altered. (i.e. moving a king from East to West)

[ArcLight]

I just finished the book. Well worth reading.

A fast summary is:

1) The LAW of Total Tricks is accurate over a large number of hands, but it has a wide variance, meaning you will frequently be off by 1 or 2 tricks in either direction.

2) The LAW is less accurate the greater the number of combined trumps.

3) The method the authors suggest has 2 components:
Distribution (they call this Short Suit Total) and
Working Points


Working Points are HCP that are useful. (Ex. Qx in the opps bid and raised suit is probably useless. A king behind a 2NT opener is suspect.) You get a bonus for long suits, and for implied ownership, meaning if you have 10 total trumps including the AK, you probably will drop the Q and J and will get creedit for owning them too.

SST = the sum of the length of your partnerships 2 shortest suits. If your third suit is a doubleton, reduce your SST by 1. IF its a stiff, reduce SST by 2, and a void reduces the SST by 3.

This method seems to work well. And the authors give many hands where the LAW fails and these work. To be fair they also show some hands where their idea doesnt work. Its in the case where you take a save with low HCP and few trumps. Thats because the opponents will lead trumps and you wont be able to ruff.

I think the authors tried to be fair and present the strengths and weaknesses of their system and the LAW. It seems their system is a lot better. And its easy to use.

use 13 - SST + adjustment for WP.

with 19-21 WP, no adjustment.
For every 3 additional or fewer WP, increase/reduce the number of tricks by 1.

Ex.
With 2 doubletons, you have a SST of 4 (2+2)
13 - 4 = 9 expected tricks.
If you have 23 WP you get a +1, and can expect to take 10 tricks.


(its a much faster read than Mike Lawrences other books, perhaps a bit deeper than Larry Cohens LAW book, but still easier than many advanced books)

[Chamaco]

Quote

1) The LAW of Total Tricks is accurate over a large number of hands, but it has a wide variance, meaning you will frequently be off by 1 or 2 tricks in either direction.

Off by 1 total trick is far within the tolerance used by most law followers: that's where judgement comes in.

Quote

3) The method the authors suggest  has 2 components:
Distribution (they call this Short Suit Total)  and
Working Points
Working Points are HCP that are useful.  (Ex. Qx in the opps bid and raised suit is probably useless.  A king behind a 2NT opener is suspect.)  You get a bonus for long suits, and for implied ownership, meaning if you have 10 total trumps including the AK, you probably will drop the Q and J and will get creedit for owning them too.
SST = the sum of the length of your partnerships 2 shortest suits.  If your third suit is a doubleton, reduce your SST by 1.  IF its a stiff, reduce SST by 2, and a void reduces the SST by 3.

This seems to converge more and more towards ZAR evaluation concept ?
ZAR accounts for length and shortness as well as for fitting honors.

[ArcLight]

First, its certainly worth buying the book. Its $16 from Carl Ritner at www.carlritner.com. Ive bought many books from him, he is reputable.

Secondly, the authors do a better job than I of stating their points and making their points.

Having said that ...

>Off by 1 total trick is far within the tolerance used by most law followers: that's where judgement comes in.

It can be off by a lot more than 1, and frequently, as the total number of tricks increases. The variance (meaning deviation from the arithmetic average) is high. Also, it can make a great deal of difference if a King is switched, which side plays the hand, etc. The LAW says this shouldn't matyter, but it does in some cases. The LAW is not a good method to use, there are better. Read the book and the authors will explain this in greater depth, with many more examples.


>This seems to converge more and more towards ZAR evaluation concept ?
ZAR accounts for length and shortness as well as for fitting honors.

Its not ZAR points at all. The authors discuss that having lots of trump doesnt necessarily add anything. Distribution and working points. ZAR dosnt address working points.

The book gives many examples of typical hands where moving a card around between the hands has a big impact on the total tricks (despite the LAW saying this wont happen), while the authors system does generally account for this.


If you are in doubt, try reading the book. See what you think then. At least keep an open mind. Time will tell which is a better system.