New GiB record?

[nullve]

From my BBF Challenge match against jeffford76:

 

[hv=http://www.bridgebase.com/tools/handviewer.html?bbo=y&lin=pn|nullve,Robot,Robot,Robot|st||md|3SJ8752H42DT43C742,SQT63HQ63DA75CKJ3,SK94H87D862CQT865,SAHAKJT95DKQJ9CA9|sv|E|ah|Board%209|mb|P|mb|2C|an|Strong%20two%20club%20--%2019+%20HCP;%2023+%20total%20points;%20forcing%20to%202N|mb|P|mb|2N|an|Positive%20notrump%20--%202-5%20!C;%202-5%20!D;%202-4%20!H;%202-4%20!S;%208+%20HCP|mb|P|mb|3H|an|5+%20!H;%2019+%20HCP;%2023+%20total%20points|mb|P|mb|4N|an|Blackwood%20(H)%20--%202-5%20!C;%202-5%20!D;%203-4%20!H;%202-4%20!S;%208+%20HCP;%209+%20total%20points|mb|P|mb|5D|an|One%20or%20four%20key%20cards%20--%205+%20!H;%2019+%20HCP;%2023+%20total%20points|mb|P|mb|5N|an|Query%20kings.%20Have%20all%20keycards.%20Does%20not%20promise%20extras%20--%202-5%20!C;%202-5%20!D;%203-4%20!H;%202-4%20!S;%208+%20HCP;%20!HQ;%209+%20total%20points|mb|P|mb|6D|an|King%20of%20D%20--%205+%20!H;%2019+%20HCP;%20!DK;%2023+%20total%20points|mb|P|mb|6H|an|Signoff%20--%202-5%20!C;%202-5%20!D;%203-4%20!H;%202-4%20!S;%208+%20HCP;%20!HQ;%209+%20total%20points|mb|P|mb|P|mb|P|pc|S5|pc|S3|pc|S4|pc|SA|pc|H5|pc|H2|pc|H3|pc|H8|pc|CT|pc|C9|pc|C2|pc|CJ|pc|H6|pc|H7|pc|HJ|pc|H4|pc|D9|pc|D3|pc|D7|pc|D2|pc|H9|pc|S2|pc|HQ|pc|C6|pc|ST|pc|S9|pc|DJ|pc|SJ|pc|D4|pc|D5|pc|D8|pc|DK|pc|DQ|pc|DT|pc|DA|pc|D6|pc|S6|pc|SK|pc|CA|pc|S7|pc|C5|pc|HK|pc|C4|pc|C3|pc|HA|pc|C7|pc|SQ|pc|CQ|pc|HT|pc|S8|pc|CK|pc|C8|]400|300[/hv]

 

At the other table:

 

[hv=http://www.bridgebase.com/tools/handviewer.html?bbo=y&lin=pn|jeffford76,Robot,Robot,Robot|st||md|3SJ8752H42DT43C742,SQT63HQ63DA75CKJ3,SK94H87D862CQT865,SAHAKJT95DKQJ9CA9|sv|E|ah|Board%209|mb|P|mb|2C|an|Strong%20two%20club%20--%2019+%20HCP;%2023+%20total%20points;%20forcing%20to%202N|mb|P|mb|2N|an|Positive%20notrump%20--%202-5%20!C;%202-5%20!D;%202-4%20!H;%202-4%20!S;%208+%20HCP|mb|P|mb|3H|an|5+%20!H;%2019+%20HCP;%2023+%20total%20points|mb|P|mb|4N|an|Blackwood%20(H)%20--%202-5%20!C;%202-5%20!D;%203-4%20!H;%202-4%20!S;%208+%20HCP;%209+%20total%20points|mb|P|mb|5D|an|One%20or%20four%20key%20cards%20--%205+%20!H;%2019+%20HCP;%2023+%20total%20points|mb|P|mb|5N|an|Query%20kings.%20Have%20all%20keycards.%20Does%20not%20promise%20extras%20--%202-5%20!C;%202-5%20!D;%203-4%20!H;%202-4%20!S;%208+%20HCP;%20!HQ;%209+%20total%20points|mb|P|mb|6D|an|King%20of%20D%20--%205+%20!H;%2019+%20HCP;%20!DK;%2023+%20total%20points|mb|P|mb|6H|an|Signoff%20--%202-5%20!C;%202-5%20!D;%203-4%20!H;%202-4%20!S;%208+%20HCP;%20!HQ;%209+%20total%20points|mb|P|mb|P|mb|P|pc|C2|pc|C3|pc|C8|pc|C9|pc|H5|pc|H2|pc|H3|pc|H8|pc|CT|pc|CA|pc|C4|pc|CJ|pc|HJ|pc|H4|pc|H6|pc|H7|pc|D9|pc|D3|pc|D7|pc|D2|pc|HT|pc|S2|pc|HQ|pc|S4|pc|S3|pc|S9|pc|SA|pc|S5|pc|DK|pc|D4|pc|D5|pc|D8|pc|DQ|pc|DT|pc|DA|pc|D6|pc|ST|pc|SK|pc|DJ|pc|S7|pc|C5|pc|HK|pc|C7|pc|CK|pc|HA|pc|S8|pc|S6|pc|CQ|pc|H9|pc|SJ|pc|SQ|pc|C6|]400|300[/hv]

[The_Badger]

hi nullve,

 

OH MY GOD! Nothing else to say. Quite unbelievable.

[Stefan_O]

Yes, unfortunately, the cardplay engine is buggy, too.

Quite similar examples, were reported here some weeks ago.

[broze]

Wow. The worst I've seen!

[lycier]

http://i64.tinypic.com/zl4ug2.jpg

[sfi]

Tough way to lose 3 imps!

[lycier]

I guess Gib playing server is too old to do the work so well.

[steve2005]

I guess Gib playing server is too old to do the work so well.

This isn't an age problem. A human beginner or a young program almost certainly wouldn't make that mistake and if it did it would be quickly fixed.

This is big problem. I saw Gib with Qxx if trump offside play the Q on first round of trump.

There is a major bug in card play.

 

 

[Stefan_O]

The thread with two earlier cases I mentioned above -- just as ugly:

 

http://www.bridgebase.com/forums/topic/74881-gib-bug/

[Stefan_O]

And then we have these masterpiece defenses...

 

http://www.bridgebase.com/forums/topic/74844-bug-worst-lead-ever-in-world-history/

 

Not sure if the first-lead is a different algorithm than subsequent cards, but buggy it is.

 

...unless I missed a BBO announcement, that any of these have now been fixed, of course.

[jdonn]

BBO is aware of this hand, and investigating. Thanks

[Vampyr]

Can the bot be taught to count top tricks?

[Stephen Tu]

It knows how to count top tricks. My home copy of GIB just claims at trick 1.

 

Why it went haywire here, they'll have to investigate.

[iandayre]

Once thing GIB seemingly does perfectly is accept or reject claims. I would not mind if BBO's GIB would claim when it has all remaining tricks certain.

[smerriman]

BBO is aware of this hand, and investigating. Thanks

 

We've moved on from hundreds of threads never being replied to, to your replies saying they're being looked into. Yet I haven't seen a single response with the result of said investigation.

 

Was anything found? Or did they give up? Bugs like the ones in this thread seem seem to be the highest priority given how common they are.

[MrAce]

This isn't an age problem. A human beginner or a young program almost certainly wouldn't make that mistake and if it did it would be quickly fixed.

This is big problem. I saw Gib with Qxx if trump offside play the Q on first round of trump.

There is a major bug in card play.

 

 

+1000

 

I think GIB needs a major overhaul in its card play program.

 

 

 

 

[diana_eva]

These absurd plays or bids occur when bots run out of memory. I don't know the technicalities but it's something about servers and memory, not about how robots are programmed. Happens only to challenge and daylong robots, you won;t get these when renting basic robots or in robot tournaments.

[MrAce]

These absurd plays or bids occur when bots run out of memory. I don't know the technicalities but it's something about servers and memory, not about how robots are programmed. Happens only to challenge and daylong robots, you won;t get these when renting basic robots or in robot tournaments.

 

This statement is untrue about the rented bots.

I gave examples for the bot that I rented for 1$, an advanced robot which refused to take the defeating trick vs their slam in 2 different hands in 1 week.

Yes they make much less mistakes than regular bots but they still have issues.

[diana_eva]

Sorry I didn't mean to say rented bots have no bugs.

[MrAce]

Diana, here is an example of tournament GIB

 

 

http://tinyurl.com/zxghuoz

 

Check the 9th trick

[diana_eva]

Diana, here is an example of tournament GIB

 

 

http://tinyurl.com/zxghuoz

 

Check the 9th trick

 

That's not nearly close to bot losing a trick from AKJT9x vs Qxx. That;s the kind of thing I was referring to when I said rented/tourney bots won;t do that.

[smerriman]

Happens only to challenge and daylong robots, you won;t get these when renting basic robots or in robot tournaments.

 

Does that include the free weekly instant tournament? Just saw this at another table:

 

[hv=http://www.bridgebase.com/tools/handviewer.html?sn=civra1&s=SJ832HAQ8DKT98CA4&wn=Robot&w=SAT976HJ972D4CT72&nn=Robot&n=SKQ4HKT6DQJ763CJ3&en=Robot&e=S5H543DA52CKQ9865&d=w&v=b&b=4&a=P1D(Minor%20suit%20opening%20--%203+%20%21D%3B%2011-21%20HCP%3B%2012-22%20total%20points)P3N(Balanced%20--%202-5%20%21C%3B%202-5%20%21D%3B%202-3%20%21H%3B%202-3)PPP&p=H2H6H3HADTD4D3DACKC4C2C3C6CAC7CJDKS6D6D5D8H7DQD2DJC9D9S9D7H5S2S7HTH4HQHJH8H9HKC8S4S5SJSTS3SASQC5CTSKCQS8]400|300[/hv]

 

See trick 11. Similar to MrAce's, except there are only 3 cards left, two suits, and it plays the card guaranteed to be worse than any other.

[diana_eva]

Sorry I even attempted an explanation. I shd leave this stuff to programmers.

 

Anyway, no, it doesn't apply to instants of any kind. Only to challenges and daylongs. I was trying to reassure people that absurd plays like the OP, or losing a trick from KQ9x vs AJxx like here, or playing low-low-low with KQxx on AJTxx like here won't occur with rented robot or in robot tournaments. I didn't expected as counter example defense goof ups that I could be making at the table :P Not that those aren't bugs, but they aren't 100% impossible like the OP IMO.

 

We're testing advanced robots in Challenges which should hopefully fix these bugs, and in general the robots shd be smarter. If you register for the BBF Challenge event #4 barmar will enable advanced robots for participants so we can get some feedback on whether the advanced robots in challenges are better than the bots used now.

Edited October 21, 2016 by diana_eva

[smerriman]

We're testing advanced robots in Challenges which should hopefully fix these bugs, and in general the robots shd be smarter.

 

Right - that's definitely a good thing, but is unrelated to the actual issue here - why is basic Gib making these defensive mistakes, when according to the definition of how it defends, it should be 100% impossible (more impossible than the declarer play, where the algorithm involved is a lot more complex). And will BBO ever look into it?

 

In terms of defense, the only difference between advanced Gib and basic Gib is an increased number of simulations - but even with a small number of simulations, the above mistakes should never happen, implying there's a larger fault here.

 

(Even if the programmers looked into a single example, and said the reason Gib lead the worst possible card is because it only ran 10 simulations, by a 1 in a million fluke, all of the 10 happened to not include the usual distributions, so it didn't think the card it played mattered. That would satisfy me.)

[nige1]

If you register for the BBF Challenge event #4 barmar will enable advanced robots for participants so we can get some feedback on whether the advanced robots in challenges are better than the bots used now.

Good! that should remove MrAce's objection :)