[blml] Thai braking

[Robert Geller]

A basic principle of science or indeed of measurement in general
is not to strive for unnecessary precision.   For example, if construction
laborers are filling up 20kg bags of dirt it would be senseless to worry
about measiring the weight with nonogram precision.

Similarly, the inherent variability and randomness in bridge
is sufficiently great that the present imp and VP scales
are "good enough for government work."  None of the
improved fractional scales that have been suggested
by the writers in this thread is unreasonable, but on the
other hand I question whether they are necessary.

In the first place, Swiss events have a huge luck factor,
due to the randomness of the matches (greatly variable
strength of opponents).   Indeed, this is probably precisely
why Swiss events are so popular with the paying customers.
Given this randomness and uncertainty it seems out of
proportion to worry about making miniscule adjustments
in the imp or vp scales that would make them slightly more
optimum (in a statistical sense) when the improvement is
probably one significant digit or more below the inherent 
"noise level" anyway.

-Bob



>[nige1]
>>> In a round-robin (or Swiss teams event with duplicated boards), an
>>> alternative form of scoring is possible - a cross between match-points
>>> and point-a-board...
>>>
>>> [A] Calculate the match-point scores in the usual way.
>>> [B] For each team, add the North-South and East-West match points.
>> [Harald Skj誡an]
>> There already excist a form of scoring teams which combine board a
>> match and IMPs or total score (this varies ), called Patton.
>>
>[nige2]
>We have similar hybrid scoring methods in the UK.
>
>{l] My suggestion isn't really a hybrid and has nothing to do with IMPs 
>-- it is simple match-points - scored as in an ordinary match-pointed 
>pairs competition. The score for a team is the sum of its North-South 
>and East-West match-points. IMO this scoring method is better than 
>board-a-match (or point-a-board) because it makes maximum use of 
>available data to determine a winner -- all the scores at all tables 
>contribute to your score. It also means that each hand is important.
>
>
>[2] A cruder method is available if you want to keep closer to current 
>Teams play. Cross-imp the field and, for each team add the north-south 
>and east-west cross-imps. Again this has the advantage of making maximum 
>use of relevant data.
>
>Given a specified number of teams and a specified number of different 
>boards for them all to play. My guess is that the form of competition 
>that produces the "best" ordering is a Swiss Teams scored in one of the 
>suggested ways.
>
>
>_______________________________________________
>blml mailing list
>blml at amsterdamned.org
>http://www.amsterdamned.org/mailman/listinfo/blml

-----------------------------------------------------
Robert (Bob) Geller,     Tokyo, Japan        geller at nifty.com