Join Bridge Winners
All comments by John Uchida
1 2 3 4 5 6 7 8 9 ... 15 16 17 18
You are ignoring the author of this comment. Click to temporarily show the comment.
From BBO notes, https://www.bridgebase.com/doc/gib_system_notes.php#Soloway_Jump_Shift

“Strong jump shifts are only from the 1 level to a higher suit on the 2 level. Jumps to a lower suit on the 3 level are natural and invitational.”
Feb. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
You could create a 2nd bidding poll for the auction that starts 1 - 1.
Feb. 15
You are ignoring the author of this comment. Click to temporarily show the comment.
There used to be a PGA tournament called The International. In the early years of the tournament, the field was cut after every round, and the winner was based only on the final round score.
Feb. 8
You are ignoring the author of this comment. Click to temporarily show the comment.
No club finesse on the opening lead?
Jan. 20
You are ignoring the author of this comment. Click to temporarily show the comment.
I assume you were using BBO's basic bots and not the advanced bots. The basic bots randomly and inexplicably make crazy, 0% bids and plays that nobody can explain. The BBO explanation is that basic bots have limited processing time and don't use the most advanced algorithms to reduce strain on BBO's servers.

On their forums, BBO moderators have said to not expect fixes for most of the basic bot deficiencies (an exception might happen if the same error was happening in their advanced bot, depending on the error).
Jan. 3
John Uchida edited this comment Jan. 3
You are ignoring the author of this comment. Click to temporarily show the comment.
Any intermediate defender would know declarer does not have a club remaining because with AJx would have won the jack at trick one to make 3 club tricks. GIB doesn't remember/analyze that declarer would have made the a different club play at trick one with 3 clubs. Sometimes GIB just has a hiccup and makes an absolutely 0% play that can't possibly be right for no apparent reason.
Jan. 3
John Uchida edited this comment Jan. 3
You are ignoring the author of this comment. Click to temporarily show the comment.
I don't see any money savings since the announcement said that apps and software were going to remain the same (separate). Funbridge has their own bridge bots.

Since nobody seems to know how to substantially modify BBO's GIB bots except the inventor who is out of the picture, the solution seems to be to switch to the Funbridge bridge playing engine.
Jan. 3
You are ignoring the author of this comment. Click to temporarily show the comment.
GIB loves to “rescue” partner when holding a void. Fuzzyquack got lucky when North did not rescue again with 6 over 5
Dec. 28, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
East-West GIBs do know that J is not guaranteed to be a singleton. Look at the descriptions for 2NT, 3NT and 6NT which show “1-2 S”. The bots know that declarer can have 1 or 2 spades. Presumably the East bot can figure out that declarer does not have the 24-26 HCP that 6NT shows :-)

If there actually was any kind of advanced AI, East would know that a singleton spade was not possible because of declarer's line of play and West's failure to discard a spade.
Dec. 19, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
The ACBL has plenty of money. Management and the BOD could afford to flush 2 million dollars for a zero return in software in the past few years. We can probably afford 2 or 3 more years of NABC's with Honolulu size losses. Let's not worry until the checks start bouncing.
Dec. 7, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
GIB isn't trying to make a deceptive play in diamonds. It's just making a random play because double dummy, those diamonds are all equal. If South read the 9 as a true card (e.g. J9, J109, etc), South should be planning to cash another top diamond and then reevaluate when partner plays the 10 or J on the 2nd round.

I don't see that the GIB's initial diamond play affected the result.
Nov. 11, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
Kudos to ACBL management for only losing $375K on the CRM software. Any loss less than $1M has to be considered a success.
Oct. 11, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
Under the GCC chart under Competitive,

7. DEFENSE TO:
a) conventional calls (except see #10 RESPONSES and REBIDS
above and #7 under DISALLOWED below),
b) natural no trump opening bids and overcalls, except that direct calls, other than the following, must have at least one known suit.
- Double
- Two Clubs
- Two Diamonds showing a one suited hand in hearts or spades
c) opening bids of two clubs or higher.
Sept. 30, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
Gwynn, Alex - Interestingly enough, Wikipedia has an entry about Plagiarism,

https://en.wikipedia.org/wiki/Wikipedia:Plagiarism

If you take the time to read the article, you might want to apply the duck test,

https://en.wikipedia.org/wiki/Duck_test
July 25, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
Play the ace the first trick, and if the queen, jack or 8 are played by LHO, then go to dummy and run the 9. Picks up the offside singleton queen, jack, or 8 for 1 loser and loses 2 tricks to void or singleton 2 or 3.
June 10, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
A reasonable person might ask whether that first step was already done for the last CEO search. If the answer is yes, the followup question should be “What went wrong and how can we fix the search?”.

Or maybe the BOD needs to get some outside “counselling” to be more successful.
May 3, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
On the BBO forums, BBO representatives have said that a couple of years ago, GIB bots scored about 55% in robot MP tournaments where 1 human plays with 3 bots. They keep doing periodic bug fixes so I think that percentage could be a few percent higher these days.

No way to tell if the people playing in the average BBO robot game are stronger or weaker than an average live club game.

For the most part, people who in BBO robot games are at least somewhat familiar with the bot bidding system and play quirks.
March 27, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
>> “(1) The bots analyze play positions double-dummy. If a finesse is onside, they know that you know that, so that card becomes useless.”

>> “(2) If bots drop the highest cards from equals, stupid humans, who aren't counting, may lose track, and the bots gain a trick.”

>> “Humans might well be aware that the missing queen of clubs is still out, but they might not notice if the only club out is the high five-spot.”

Point 1 is absolutely true. BBO has said that's what is going on and observational data confirms this. That's why you see leads of singleton king of trump because it's a no lose play for the double dummy analysis by the bots.

I doubt that point 2 was a consideration in the programming (although it certainly would improve the effectiveness of the defense if it was done correctly). Unfortunately any gains are lost in a tsunami of bad results from not knowing when cards are equals in a single dummy analysis and saving declarer from misguessing at a later trick.

>> “Their algorithm shows that you will win the queen, and they may never get that trick. So, they grab it. On the rare occasions when they duck, they have to run their simulations, and that takes time! So, they never duck in tempo.”

BBO may be running hundreds or even thousands of robot games at the same time. Sometimes delays can come from other tables.

Someone posted a hand on BBO forums where a bot on defense ducked from AKQx in a side suit for absolutely no reason and let an impossible game make. I wouldn't trust making a decision based on bot tempo.
March 15, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
Board 12 - I know your post was supposed to be about the GIB bots so I won't comment on what you actually had, but never make an offshape takeout double playing with GIB. Even if you show a strong single suited hand or big NT hand later, GIB still assumes you have normal takeout double support for all the unbid suits. Worst case is you get in a bidding war with GIB bidding and rebidding a weak 4 card suit (opposite one of your short suits).
Feb. 25, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
The article says that for the 100 deals shown in detail, only the 1st 10 and the last 10 were random, and the other 80 were chosen non-randomly because the option chosen made a difference.

That makes the 100 deal set decidedly non-random.
Dec. 29, 2017
1 2 3 4 5 6 7 8 9 ... 15 16 17 18
.

Bottom Home Top