[diana_eva]

Today, at the Denver NABC, BBO's Fred Gitelman presented a new anti-cheating device he's been working on. Full presentation here.


Recent high-level cheating scandals have resulted in considerable discussion about (among other things) the use of technology (among other things) to make it harder to cheat at bridge and easier to detect those who do cheat.

Bridge Base Online (BBO) has developed a device which attempts to do just that. For the purposes of this presentation, the device will be referred to as the Bridge Base Online Hybrid (BBOH).

BBOH can be thought of as a hybrid between playing bridge with traditional screens and playing bridge electronically. The BBOH consists of 2 basic components: a screen component and a tablet/software component.

Many leading players believe, reasonably enough, that a fully electronic solution with players sitting in separate rooms would be best for security purposes, but that something important would be lost in the form of "table presence".

BBOH aims to leverage some security benefits of a fully electronic environment while not detracting from table presence. In fact, BBOH improves over traditional screens in this area (where table presence is essentially limited to only your screenmate).

At this point in time, we have built the first prototype of the BBOH screen component and written some prototype software that allows BBOH to be tested. We eagerly await feedback and expect to improve both the screen component and software component as a result of such feedback.

Comments and feedback welcome, either here, or via email to Fred.

[manudude03]

It would have been nice to see a tablet screen when dummy has been shown. Some concerns do come to mind though:

1. When using screens, if a player requires information from the opponents, they use a notepad they share with their screen mate. What would happen with this new layout? If you use 4 notepads (one for each "link" at the table), it seems there would be a very real chance that partner could see the notepad, possibly transferring UI.

2. What happens in the case of a "misclick" either in the auction or during the play? Is it just treated as pulling out the wrong card?

edit: Also, those with less dexterity might find it a bit of a problem to type in lengthy explanations for alerts (assuming they aren't held to the 30 character limit or whatever BBO has)

[hrothgar]

Hi Fred,

Thanks for posting such a well written description. In particular, documenting your cost assumptions is much appreciated. Here are a couple additional points that you might want to consider.

First: I suspect that you might encounter problems where people accidentally “see” a hand on a tablet at another table. It is possible to equip the screens with a privacy filter so the screens can not be viewed at an angle. The following link provides a representative example.

http://www.amazon.co...0/dp/B00028ONIA

Second: I think that there is value in having a mechanism by which you can demonstrate that the hands for a session are being dynamically generated in a “secure” manner. You want to make sure that people have faith both in the hand generator program and the mechanism used to seed the system. I’d recommend a system like the following

1. Use an open source program to generate your hands (Hans van Staveren’s Big Deal would be the obvious choice

2. Have a set of teams collaboratively generate the seed the program at the start of the event. Here’s one possible (elaborate) implementation. At the start of the round, four teams are chosen at random. A hand of bridge is dealt out, with a member of each team receiving the cards. Each player shuffles his own hand, marks down the order of the cards, and then hands the cards to the TD. The TD then shuffles and deals a second deck. The seed for the round is generated from (the set of four hands specified by the players salted with the second deck). At the end of the round, each principle (the four teams and the TD) publish their input in generating the seed to verify that the set of hands match the specified seed.. This may seem silly and over engineered. Perhaps it is, however, this system will provide enough randomness to be cryptographically secure. People trying to social engineer the system would need to have players on four different teams collaborating, plus have the Tournament Director in on the act.

[Fluffy]

Tablets have the problem of opponents peeking at where you put your finger when you press your card. The software can counter this in several ways:

-Allow to rearrange cards
-Allow to use a personal USB mouse instead of touching

Drag and drop is the way to avoid missclicks. Easier with mouse.

You didn't explain how the self alerts would happen. If you write them on paper, there is UI between opponents that someone asked, and probably to partner given that he could see opponent asking and hear you writing.
If you do it electronically, typing on a tablet is slow, and also noisy. So still UI will be present.

Having said so, I love the idea and I really hope it gets developed at high level bridge.

PS: I second Richard's advice about deal generating. Generate them after everyone is sit and another way of cheating is removed.

[mgoetze]

Return of the foot signals?

[lycier]

If you employ Apple ipad,its Visual Angle is 178 degrees , bigger than general display,you should need narrow-angle ipad,it would better keep 120 degrees so as to enhance the confidentiality of the match.

[straube]

I really like this. I like that both opponents get explanations from both partners (which may be stated differently or wrongly in a bidding misunderstanding). I like that it preserves a lot of the social aspect of the game (which I really missed that the last time I used a laptop sitting by myself for a Grand National event). Even if cheating weren't an issue, it would be nice just to reduce infractions and not have to worry so much about unauthorized information. Thanks for working on this Fred.

[wank]

my problem is that, in my case, 'cards' on a screen don't process in the same way as real cards. i can't count a hand on bbo. i can't even remember what happened on the previous trick. it's all just a blur of colour. btw i'm 35 so i'm not some geriatric who can't stand modern technology.

[gordontd]

wank, on 2015-December-04, 23:36, said:

my problem is that, in my case, 'cards' on a screen don't process in the same way as real cards. i can't count a hand on bbo. i can't even remember what happened on the previous trick. it's all just a blur of colour. btw i'm 35 so i'm not some geriatric who can't stand modern technology.

Yes, I think for most of those who disliked the idea of moving to an entirely electronic environment, the physical cards were an important part of what they liked about face-to-face bridge.

It seems to me that Bridge+More or a similar offering from a competitor might have the possibility of providing the best of both worlds:

• Physical cards
  
• No pre-dealing
  
• Tablets only used for the auction
  
• Software can reconstruct the play
  
• Can be used for BBO without needing an operator
  
• Only need two tablets per table when played with screens
  
• Could be used in club setting without screens and only one tablet per table

btw I have no brief to advertise Bridge+More, but I have seen it in action and liked the process it uses. However it's not yet ready for market and I'm hopeful that there may be a competitor in the wings, which would probably help make them affordable.

[Mbodell]

Fwiw I had the same problem with online play when I only did it occasionally. But now I regularly play 50+ serious hands a week and after about six months of this online is fine and works as well as in person. In fact, I played in the lm pairs with one of my online regular partner's and having to alert his bids, instead of self alert, felt odd the first day.

[Fluffy]

I liked bridge+more it also, but had lost track of the project. However I found the tablet solution for bidding not very good, as it is slower than just biddingboxes, and I see no improvement.

Anyway brdge+more has nothing to do with BBOH, BBOH is anti cheating enhancement for screens, while bridge+more is for clubs without screens. There could be a mixup solution with bridge+more dealing the hands, people hlding cards in hands, and play them in front of them on BBOH enviroment, to be captured with the dutch software for automatic vugraph reproduction to show cards played to partner. But all that would be too complicated and expensive.

[barmar]

hrothgar, on 2015-December-04, 18:01, said:

Second: I think that there is value in having a mechanism by which you can demonstrate that the hands for a session are being dynamically generated in a “secure” manner.

Is there anything about BBOH that suggests that hands would be generated by BBO software, rather than whatever software ACBL (or other organizations) uses for their f2f tournaments?

[gordontd]

Fluffy, on 2015-December-05, 03:13, said:

Anyway brdge+more has nothing to do with BBOH, BBOH is anti cheating enhancement for screens, while bridge+more is for clubs without screens.

Certainly that wasn't the motivation for Bridge+More or similar devices, but they could certainly be used so as to have that effect.

[hrothgar]

barmar, on 2015-December-05, 04:24, said:

Is there anything about BBOH that suggests that hands would be generated by BBO software, rather than whatever software ACBL (or other organizations) uses for their f2f tournaments?

I am making a basic presumption of intelligence on your part.

The existing system is badly flawed in that it requires trust in individuals rather than trust in a secure and auditable process.

This creates problems when people make a mistake (for example, reusing the same seed for more than one event). This sorts of cock ups happen more often than one would like.
It also creates problems if some person is corrupt and (hypothetically) decided to sell hand records for a major event.

Last, but not least, this creates problems when people are paranoid and start making accusations that people are stealing hand records and there is no way to disprove this. (Take a look at some of the stuff that Gabby was posting over at Bridge Winners about the 2013 USBF team trials)

The best way to avoid all of this hassle is to use Open Source code so that third parties can inspect and validate the hand generation system, generate the seed for each tournament in a public/collaborative fashion, and publish the seed at the end of each event.

Or, alternatively, do whatever you damn well please and leave yourself exposed to all sorts of headaches after the fact as a horde of nutcases crawl out of the woodwork...

In short, if you are going to implement this system, do it right rather than handcuffing yourself to an obsolete process.

[RMB1]

gordontd, on 2015-December-05, 02:48, said:

It seems to me that Bridge+More or a similar offering from a competitor might have the possibility of providing the best of both worlds:

• Can be used for BBO without needing an operator

Fixed?

[gordontd]

RMB1, on 2015-December-05, 07:39, said:

Fixed?

Yes, thanks

[Manastorm]

Current cheating scandal has shown how useful hand records and video data is. Could the hand records be used to disclose actual bidding and carding agreements?
Pairs should provide a dataset how their methods work by test deals and actual tournament deals. Players could automatically see statistical information of
bids: expected values, minimum value, maximum value, possible outliers, suit lengths etc. If bridge is played with laptops, why not take advantage of
the new techonology?

[blackshoe]

Some years ago, there was a bidding program called, iirc, "Bridge Partner". It used AI technology to allow the software to "learn" whatever bidding system you wanted to teach it. You taught it by bidding hands. Unfortunately, it would take thousands of hands for it to learn a system, and even then you might well introduce anomalies. Far as I know, the project is dead. What you're proposing, Manastorm, seems similar.

[Fluffy]

Someone on BW said that bridge+more can also record the play of the hand by reading the cards when you put them back on the device at the same time it is dealing a new one. I don't know if its true, but would also be a step forward. Although it would need to read also how many tricks were left when there was a claim.

[gordontd]

It does. You put the hands back in unshuffled and it reconstructs the play. I imagine it could be a card or two either way in determining the exact moment of the claim.