PokerChips: Hosting Fair RNG Sessions

[santaclause102]

For anyone interested in a Poker DAC https://bitsharestalk.org/index.php?topic=14372.0

[bytemaster]

With our voting DAC we will have certified IDs that are verified voters.  You could use that to identify unique users.

[CoinHoarder]

Can I get ya'lls opinion on this?

https://www.dropbox.com/s/rhvqsa702k8p5vg/v0.01%20-%20The%20Future%20of%20Decentralized%20Online%20Poker%20As%20I%20See%20It.docx

On version 0.02 now.. I added a lot more data points to catch collusion and bots, and made some other minor changes such as adding "infractions" and initiating a three strike rule for innocent players that are mistakenly flagged.

My concerns:
1. The vote system might be able to be gamed, any suggestions to make an improvement on it?
2. What is the best way go about stopping multi accounting?
3. Eliminating duplicate data points and adding ones I haven't thought of.

I decided that giving people three chances wasn't a good idea and removed the infractions and "three strike rule". If cheaters are caught, they should be banned. Not sure what I was thinking there, but it seems ridiculous to have to catch someone cheating three times to ban them.

Anyways.. I could really used some feedback on what I wrote. I've been to like 4 forums, emailed a couple people I know are really interested in decentralized poker, and I haven't got any feedback on this is it is rather frustrating. I am trying to help, but I need a little feedback to improve upon what I have done this far. I can keep working on the data points... consolidating them, deleting duplicate searches, and adding ones I haven't thought of/researched yet. However, I need help on the following:

My main concern at this point is stopping multi accounting (which I feel might be impossible in a decentralized poker network after doing some research), and people being able to game the cheating detection system. Multi accounting is tough, because the way centralized sites stop it is by asking players to send in photo IDs, utility bills, etc.. then they analyze them for fakes and make sure no one has more than one account. In a decentralized poker network I feel there may be no good solution for this.

[CoinHoarder]

Can I get ya'lls opinion on this?

https://www.dropbox.com/s/rhvqsa702k8p5vg/v0.01%20-%20The%20Future%20of%20Decentralized%20Online%20Poker%20As%20I%20See%20It.docx

On version 0.02 now.. I added a lot more data points to catch collusion and bots, and made some other minor changes such as adding "infractions" and initiating a three strike rule for innocent players that are mistakenly flagged.

My concerns:
1. The vote system might be able to be gamed, any suggestions to make an improvement on it?
2. What is the best way go about stopping multi accounting?
3. Eliminating duplicate data points and adding ones I haven't thought of.

[gamey]

Awesome! I will definitely read the paper within the next couple of days.  I am very interested because it seems like the solution of this would have exceedingly broad applications beyond poker. 

While somewhat smart I have adhd which makes reading academic papers a lot of effort so I am always hesitant to start.

Sent from my SM-N900P using Tapatalk

[toast]

I could read the paper but I am almost certain to be let down.  decentralized does not work with poker.  The network has to know cards.  There is no way to guarantee no collusion between host and a player. 

Sent from my SM-N900P using Tapatalk

Not coinhoarder's paper, that's just anti-collusion techniques.

Sergio's MPF paper. There is trustless poker. Clever crypto.
http://www.dc.uba.ar/inv/tesis/licenciatura/2010/lerner

There is always collusion, but that is equally problematic in centralized poker solution. If Full Tilt / PokerStars could run it profitably then it is possible to decentralize.

Does the paper say how a network can know holecards and guarantee with a very high degree of certainty that only the network knows?  You were the one to point out to me that a network can't know private keys. (Which led to a lot of insight as to what is possible/the necessity of "tokens") 

If a network can guarantee that no one knows holecards then you have solved the same problem AFAIK.  I'm sure that there is some way to fragment the data where one node won't know, but that isn't quite acceptable.  The expectation lost due to collusion with the house far outweighs any collusion between players.  This is likely true in most any multiplayer game of imperfect information.  IMO that is why so much of the effort to make poker honest is misplaced.

The difference between centralized vs decentralized poker is that the number of actors you have to trust is finite in centralized poker.  In a network, it is unknown.

There is almost nothing I know more about than the meta game in/of poker.  I had an account on Planet Poker which was the first for money poker site.  Not trying to pull rank or whatever, just saying this is like the one thing I've wasted far too much of my life being involved with.  So I'd love to see a decentralized poker solution that solves house collusion.  Catching player collusion has little value to me and is relatively straight forward.  This isn't intuitive to most people though. It appears more value is placed on the solvable problem of detecting (with some certainty value) inter-player collusion or some simplistic fixation on the "fairness" of the RNG.

YES, it is solved. I know it seems impossible. I don't know how else to say it, besides "read the paper".

[CoinHoarder]

Does the paper say how a network can know holecards and guarantee with a very high degree of certainty that only the network knows?  You were the one to point out to me that a network can't know private keys. (Which led to a lot of insight as to what is possible/the necessity of "tokens") 

If a network can guarantee that no one knows holecards then you have solved the same problem AFAIK.  I'm sure that there is some way to fragment the data where one node won't know, but that isn't quite acceptable.  The expectation lost due to collusion with the house far outweighs any collusion between players.  This is likely true in most any multiplayer game of imperfect information.  IMO that is why so much of the effort to make poker honest is misplaced.

The difference between centralized vs decentralized poker is that the number of actors you have to trust is finite in centralized poker.  In a network, it is unknown.

There is almost nothing I know more about than the meta game in/of poker.  I had an account on Planet Poker which was the first for money poker site.  Not trying to pull rank or whatever, just saying this is like the one thing I've wasted far too much of my life being involved with.  So I'd love to see a decentralized poker solution that solves house collusion.  Catching player collusion has little value to me and is relatively straight forward.  This isn't intuitive to most people though. It appears more value is placed on the solvable problem of detecting (with some certainty value) inter-player collusion or some simplistic fixation on the "fairness" of the RNG.

Mental Poker solves this, which is what Toast was implying. If you read the paper you will see that Sergio has all of that figured out, collusion in between the host and players will be mathematically impossible through cryptographic functions. The problem of Mental Poker has been researched since 1979, there is a ton of work that has been done on it in academia. It is practically solved at this point after 30+ years of research.

[gamey]

I could read the paper but I am almost certain to be let down.  decentralized does not work with poker.  The network has to know cards.  There is no way to guarantee no collusion between host and a player. 

Sent from my SM-N900P using Tapatalk

Not coinhoarder's paper, that's just anti-collusion techniques.

Sergio's MPF paper. There is trustless poker. Clever crypto.
http://www.dc.uba.ar/inv/tesis/licenciatura/2010/lerner

There is always collusion, but that is equally problematic in centralized poker solution. If Full Tilt / PokerStars could run it profitably then it is possible to decentralize.

Does the paper say how a network can know holecards and guarantee with a very high degree of certainty that only the network knows?  You were the one to point out to me that a network can't know private keys. (Which led to a lot of insight as to what is possible/the necessity of "tokens") 

If a network can guarantee that no one knows holecards then you have solved the same problem AFAIK.  I'm sure that there is some way to fragment the data where one node won't know, but that isn't quite acceptable.  The expectation lost due to collusion with the house far outweighs any collusion between players.  This is likely true in most any multiplayer game of imperfect information.  IMO that is why so much of the effort to make poker honest is misplaced.

The difference between centralized vs decentralized poker is that the number of actors you have to trust is finite in centralized poker.  In a network, it is unknown.

There is almost nothing I know more about than the meta game in/of poker.  I had an account on Planet Poker which was the first for money poker site.  Not trying to pull rank or whatever, just saying this is like the one thing I've wasted far too much of my life being involved with.  So I'd love to see a decentralized poker solution that solves house collusion.  Catching player collusion has little value to me and is relatively straight forward.  This isn't intuitive to most people though. It appears more value is placed on the solvable problem of detecting (with some certainty value) inter-player collusion or some simplistic fixation on the "fairness" of the RNG. 

[toast]

I could read the paper but I am almost certain to be let down.  decentralized does not work with poker.  The network has to know cards.  There is no way to guarantee no collusion between host and a player. 

Sent from my SM-N900P using Tapatalk

Not coinhoarder's paper, that's just anti-collusion techniques.

Sergio's MPF paper. There is trustless poker. Clever crypto.
http://www.dc.uba.ar/inv/tesis/licenciatura/2010/lerner

There is always collusion, but that is equally problematic in centralized poker solution. If Full Tilt / PokerStars could run it profitably then it is possible to decentralize.

[gamey]

I could read the paper but I am almost certain to be let down.  decentralized does not work with poker.  The network has to know cards.  There is no way to guarantee no collusion between host and a player. 

Sent from my SM-N900P using Tapatalk

[CoinHoarder]

Good collection of anti-collusion suggestions.

Serio Lerner who wrote the 100% trustless Mental Poker Framework paper is providing consultation for BitShares =D

Awesome, yes he is an expert on mental poker.. I am sure he has it all figured out!

I am trying to focus more on ways to fight collusion, as that is another challenging problem that perhaps developers are not fully aware of. I've played a lot of poker in my time, so I'm fairly familiar with the logistics of online poker networks.

I really like Bitshares and Bitshares AGS.. wish I had some money to invest! Hopefully I can get some in before AGS closes up. I've been recommending Bitshares AGS to some of my friends/family, as you guys have a lot of good projects in the works.

Good luck to you guys, I'll be around. This is only my first iteration of that paper, I would like feedback and it will be revised as feedback is received.

[toast]

Good collection of anti-collusion suggestions.

Serio Lerner who wrote the 100% trustless Mental Poker Framework paper is providing consultation for BitShares =D

[CoinHoarder]

Pokerstars has 'Zoom' tables at multiple stakes.
There are usually a couple of hundred people in the zoom pool at any one time.
After each hand you are zoomed to another random table & given a new hand, this is to speed up the action but it also means you can't collude because the likelihood of you sitting at the same table as your colluders enough is very low.

Also I'm sure 80% of the poker volume is under 2/4 ($400 buy-in tables) and at those levels colluding is pretty pointless. Winners at those stakes are either bum-hunting (seeking out & getting position on weak/recreational players, no collusion needed, not worth it.) or they're mass multi-tabling playing 8-24 tables at the same time, making so many decisions so quickly it's not possible to collude. (They break even and live off rake-back*)

Rake* you showed the rake table earlier in the thread which looks lucrative for shareholders but bear in mind players can get anywhere from 20-60% of that back on most sites via rake-back. So you'd probably have to undercut that model by 75% to be attractive.

A bigger problem at lower stakes than colluding is bots, many people have automated bots that just play basic strategy on multiple tables and eek out a profit.

People talk about collusion etc, but the real problem is insuring that the DAC is not leaking information to another player.  How do you guarantee that ?  The goal seems to go against the nature of "Distributed" in DAC.

Research the term "Mental Poker"

[CoinHoarder]

Can I get ya'lls opinion on this?

https://www.dropbox.com/s/rhvqsa702k8p5vg/v0.01%20-%20The%20Future%20of%20Decentralized%20Online%20Poker%20As%20I%20See%20It.docx

[gamey]

Pokerstars has 'Zoom' tables at multiple stakes.
There are usually a couple of hundred people in the zoom pool at any one time.
After each hand you are zoomed to another random table & given a new hand, this is to speed up the action but it also means you can't collude because the likelihood of you sitting at the same table as your colluders enough is very low.

Also I'm sure 80% of the poker volume is under 2/4 ($400 buy-in tables) and at those levels colluding is pretty pointless. Winners at those stakes are either bum-hunting (seeking out & getting position on weak/recreational players, no collusion needed, not worth it.) or they're mass multi-tabling playing 8-24 tables at the same time, making so many decisions so quickly it's not possible to collude. (They break even and live off rake-back*)

Rake* you showed the rake table earlier in the thread which looks lucrative for shareholders but bear in mind players can get anywhere from 20-60% of that back on most sites via rake-back. So you'd probably have to undercut that model by 75% to be attractive.

A bigger problem at lower stakes than colluding is bots, many people have automated bots that just play basic strategy on multiple tables and eek out a profit.

A good assessment.  One thing about random tables/identities is it reduces transparency.  If a host was to give information to a player, this would effectively hide any way of a player protecting themselves.  This is where I think poker is not a good idea.  People talk about collusion etc, but the real problem is insuring that the DAC is not leaking information to another player.  How do you guarantee that ?  The goal seems to go against the nature of "Distributed" in DAC.