Leo LaSota, on 2012-March-28, 16:18, said:
Apparently you misunderstood my post when you say that "you bid like you have to do to win robot games". There is no need at all to make unusual bids to win robot games. What is needed to win robot games regularly is the ability to declare the hands extremely well. As others have posted, the norm is to declare about twice as often as defend. So it makes sense that the players that routinely declare the hands well will have good results. Although I have not run a set of data on my hands played and the results I have when making all "normal" bids versus hands that I make an "unusual" bid, I am pretty sure that my average is higher on the "normal" hands. Again, I like the declarer play practice and so I open some non-standard nt hands in order to have more practice. I could certainly see your point and some others as valid if it were found that "unusual" bids created much better results than "normal" bids. Again, I believe the reverse is actually true.
now that would be an interesting case for Picket's bridgebrowser software
Granted on some of the hands where Leo sticks his head out by off shape NT openings GIB doesnt work it out
and he makes the contract, but might also happen against real human opponents also....but look at the ones
where he gets 100% on a board, he also many times is headed for 0% but GIB does something else.
with Stephens software you could actually run the statistics and see what the frequency gain is for
opening 13-14 hcp 1NT with singelton or void. So maybe there is a flaw in the robot defending mechanism that
allows that to happen on a more regular basis.
In the GIB bot forums we have all posted how GIB tends to lead passively instead of aggressively against 3nt contracts, and
even when leading right, other GIB wins first trick and shifts to another suit....this way of defending does seem
to work for GIB quite alot but never does for me when I try it