[Jinksy]

For those of us *cough* new to the forums, could someone explain what the algorithm is?

I know about the page at http://www.jeff-gold...cgi-bin/knr.cgi but I haven't found anywhere online that tells you what it's actually doing - and personally I'm not that excited about a black box that could contain just a random number generator for all I know.

[mgoetze]

http://www.jeff-goldsmith.org/knr.txt

[etha]

mgoetze, on 2015-September-16, 15:31, said:

http://www.jeff-goldsmith.org/knr.txt

Thomas Andrews has done some work on various evaluation methods here is a link to one of his articles.
http://bridge.thomas...rstandings.html

[Zelandakh]

Unfortunately we often see KnR values posted for justifying actions on 1NT openings, for which I think it is wholly unsuitable. Almost all evaluation methods using a 3-2-1 (or equivalent) count tend to show an improvement over Milton in suit contracts. You can adjust for this by adding 0.5 for aces and subtracting 0.5 for queens and the resulting base (4.5-3-1.5) is functionally identical aside from whatever you decide to add for jacks. The other modifications are also things you can incorporate into Milton, although I am personally not convinced all are absolutely correct in practice.

Part of improving is fine-tuning evaluation but the impression I get is that very few top players do this consciously using any sort of algorithm but rather just "know" what the features of a hand are worth within the context of the auction. I actually think it would be a very useful exercise for bridge in general to engineer a knowledge-based system of evaluation based on input from the very top players. This was a big part of the success of chess computers, who are generally better than humans at evaluating most types of complex position these days. That was something that was a huge weakness before the work done with the best GMs to create algorithms that put into numbers the calculations that they were doing by feel. Computers ought to be the best at evaluating hands in the context of an auction, not the worst!

[Trinidad]

Regarding "Huh" (thread title):

The Ig Nobel Prize for Literature was awarded for research on the word "Huh". It seems that this word is universally known in every language. (Link to the publication)



Rik