Join Bridge Winners
All comments by Csaba Czimer
1 2 3 4 5 6 7 8 9 ... 24 25 26 27
You are ignoring the author of this comment. Click to temporarily show the comment.
I just checked Fredin's profile.
It looks like this:

Peter Fredin is suspended from posting until July 21, 2020 . Reason: accusing someone of cheating in public, multiple times


There is a hungarian proverb, it's about “Tell the truth and your head will be broken”. I guess, something similar exists everywhere.
18 hours ago
Csaba Czimer edited this comment 18 hours ago
You are ignoring the author of this comment. Click to temporarily show the comment.
I don't think there is enough space to explore partner's exact diamond holding in commonly played systems / conventions after a weak 2 opening.
Opener's hand is borderline though. Many players would open 1 nowadays. And they are right, this hand is worth 13.75 points by http://www.jeff-goldsmith.org/cgi-bin/knr.cgi?hand=AKT732%20752%20KT%2042">K&R valuation
July 11
Csaba Czimer edited this comment July 11
You are ignoring the author of this comment. Click to temporarily show the comment.
Amazing. Why wasn't is used in BBO vugraph broadcasts? At least there, where video was recorded too. The video is from 2018, and operators sometimes far from perfect (it's a hard job, I know).
June 23
You are ignoring the author of this comment. Click to temporarily show the comment.
I just took a look at your site, nukk.ai.

The staff is amazing professionally. If you don't create a world class robot player, noone will.
June 23
Csaba Czimer edited this comment June 23
You are ignoring the author of this comment. Click to temporarily show the comment.
If you play oldschool (penalty-oriented) double then you double with natural maximum, thus it should be minors, very probably with only 3, because you may risk 3 when you have 4+.
June 22
You are ignoring the author of this comment. Click to temporarily show the comment.
Paul, because programming bidding is also a f*cking complex thing with billions of possible bidding situation.
Probably GIB ran a (too short, thus bad) simulation and found that passing is better in EV.
June 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Writing a good robot is a ten (or hundred?) times bigger task than all the rest together with endless possibilities of investing time (or wasting, depending on the result).

About GIB on BBO: I have a copy of the desktop version of GIB and it plays much much better than the one we encounter on BBO. I think it is quite good in play and defence and mediocre in bidding.

Why is the difference between desktop and BBO? Disclaimer: it's only my guess, not an official opinion. As many of you probably know GIB's heart is a fast double dummy solver. Deciding single dummy play is based on simulation: it generates some (the more the better) deals that correspond with the bidding and the so far exposed cards, solves all of them double dummy, then chooses the line of play which was successful in most cases (or had the highest average score). Obviously the bigger the sample, the more accurate the result. When it runs on our computer, we can set the time limit of the simulation, more time results in better play.
What is the situation on BBO? GIB runs on the server side for many users simultaneously, thus if it were enabled the same computing power (processor time and memory usage) it would crash any non-super-computer. Still, the robots play incredibly fast. The only reason I can think of that they use very small sample size for the simulation.
The consequence: if you don't want to spend hell a lot of money on server capacity and still want better robots (playing in many instance at the same time), then they must run on client side. It won't go using JavaScript, it's too slow, very probably it must be written in good old C, C++ or something similar that can be compiled.
June 21
Csaba Czimer edited this comment June 21
You are ignoring the author of this comment. Click to temporarily show the comment.
For our relay precision it is definitely too hard.
South opens 1, north asks and learns the 6-4 at 2NT.

First problem: choose a 6-? suit instead of a known 6-3. Not impossible, but not obvious either.

Second problem: if we choose spades, we will know that partner has 1 heart, upper range, and 1 keycard, that is 2 keycards (maybe 2 aces) are missing. We can stop in 5 (or 5) however.

If we choose hearts, we learn that partner has 1 heart, upper range, 1 keycard, no king of spades, and we are at 5.
Still don't know about the minor kings, let alone the QJ (we need the jack too).
June 19
Csaba Czimer edited this comment June 19
You are ignoring the author of this comment. Click to temporarily show the comment.
This was one of my reasons too.
June 19
You are ignoring the author of this comment. Click to temporarily show the comment.
This is a popular method here (in Budapest):
- 3: general game try (range ask)
- 3oM: asks shortness (slam going hand)
June 19
You are ignoring the author of this comment. Click to temporarily show the comment.
None of them. I played it yesterday and saw the opportunity of making against West's K, but rejected it, and went down playing the simple way. Then asked myself whether I should have played the other way, then decided to ask better players here.
June 19
Csaba Czimer edited this comment June 19
You are ignoring the author of this comment. Click to temporarily show the comment.
I see no problem with the presentation.
June 17
You are ignoring the author of this comment. Click to temporarily show the comment.
Splinter in the other major, same strength as 1M-4m. 1M-4oM is nat.
June 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Nice look, annoying behaviour.

Programming:
I'd load a whole deal in one piece (into client app) with all of its possible branches, because it's painfully slow the way it is now. (my computer is OK, my internet connection is fast).

Content:
Don't try to teach bidding unless you are completely sure what you are doing (and double checked it with other experts). In the endplay example deal 9 (or 10?) out of 10 expert plays 3 as a preemptive bid.
May 11
You are ignoring the author of this comment. Click to temporarily show the comment.
I was thinking about it too, but we play equal level conversion doubles, thus partner can double with 4 spades and long diamonds, and my clubs were, well, not too strong.
April 29
You are ignoring the author of this comment. Click to temporarily show the comment.
If we play strong NT, then a simple Stayman will just be OK. They'll have a hard time to bid anything.
March 28
You are ignoring the author of this comment. Click to temporarily show the comment.
If east is afraid of 6 (he should), then 3 is better, south can rarely bid more than 4. Pass may be even better.
March 28
Csaba Czimer edited this comment March 28
You are ignoring the author of this comment. Click to temporarily show the comment.
we play it with strong club:
NV/V 1-2: 1N = (9)10-12, 1 includes 13-15 BAL, 1 16+
NV/V 3rd: 1N = 10-14, but you can open 1 if you feel like
other: 1 includes 11-14 BAL, 1N = 15-17, 1 is 18+ when BAL
March 26
You are ignoring the author of this comment. Click to temporarily show the comment.
You are right, but let's face it, nowadays most 4M opener don't have 8 tricks in hand, not even when vulnerable. Anyway, I have no problem opening 1, I'd probably choose it too, but 4 is a possibility.
March 19
Csaba Czimer edited this comment March 19
You are ignoring the author of this comment. Click to temporarily show the comment.
North might open 4.
Given 1
- 2 is OK (however i'd worry a bit about fit if opener bids 2)
- 3 is perfect IMHO.
Very probably South should pass 3.
March 19
Csaba Czimer edited this comment March 19
1 2 3 4 5 6 7 8 9 ... 24 25 26 27
.

Bottom Home Top