On October 19 2010 18:09 Fudyann wrote:
Pluzich: I'm trying to understand but I'm not sure what services exactly you think would be centralized. Nobody gets "control assigned". There is no "authorization stuff". In that sense, the network is completely decentralized.
It's still semi-centralized because we need to give some nodes the chance to build a reputation strong enough for escrowing.
I'll break it down.
poker1.com has 100 players who are suspected to cooperate in DoNs. Some players from poker2.com say this is the case and demand money back.
poker1.com does not see enough evidence, poker2.com does.
Would you expect some random group of nodes to resolve such things?
Note that things can get arbitrarily complicated and you'll need lawyers etc to be able to make informed decisions.
Well the easiest and most expensive solution would be to hire a neutral 3rd party with non-disclosure agreements that handle investigations with all the data sent to them.
Hence the service, for which you'd have to pay (a lot, btw).
The next problem you'd have to deal with is that when you hire a 3rd party to investigate you have to ship lots of data to them. Because of the amount of data will be huge you won't be able to anonymize it. So you will be potentially shipping millions of HHs and other data to a 3rd party.
For instance-in the scam scandal with DoNs, you'd either have to have actual, not showdown, HHs, (with all cards face up), something only PS could potentially have.
Otherwise, to prove scam beyond reasonable doubt they would have to analyse maybe 100x larger volume of data, but only "as seen by observers on the table" (you see cards if there is a showdown, otherwise not).
Either sending open HHs or showdown HHs but on much larger volume to a third party is so dangerous. You have to trust them THAT much information. Also, your users have to trust you/the third party, otherwise they won't sign up.
Hence the trust problem.
Brian Townsend would be very happy to work for that 3-rd party btw
The whole point of the cryptographic scheme is that you don't need to trust the server is trustworthy. The server doesn't know any of the holecards. Only you know your holecards. Does that mean we can't investigate cheating?
1
jchysk United States. Oct 20 2010 11:15. Posts 435
It'd be nearly impossible to investigate collusion in the typical ways a poker site would be able to. They can screenshot a desktop and see what applications are running in the background, they know hole cards, same table frequency, and they know personal information like location and IP address.
On October 20 2010 08:19 Fudyann wrote:
The whole point of the cryptographic scheme is that you don't need to trust the server is trustworthy. The server doesn't know any of the holecards. Only you know your holecards. Does that mean we can't investigate cheating?
In principle it may be possible to "open" the HH, you need a special protocol for that which is invoked, for instance, when there is a court case or something like the investigation we discussed above. But you'd probably need the agreement of all the servers because of cryptography (i.e. everyone has to use their private key).
We know same table frequency. The others are unreliable (ip address and location) or too invasive (screenshot desktop). Potentially we can figure out a protocol for revealing hole cards during an investigation.
Would typical players mind if their hole cards were made public in the course of such an investigation? Many live players get pissed if they cards are accidentally revealed by the dealer at the end of the hand, and in poker it's obviously valuable to hide how you play.
Bump. I am still interested, but I'm stuck for the moment on the problem of collusion tracking.
Does anyone have a clue how we're going to prevent collusion in a p2p network? My best idea is this:
You rate the people you play with if you suspect them of colluding. The rating is sent into the network and clearly visible whenever you play with someone. You avoid people that are often suspected of colluding.
The basic problem remains that there is no central authority that has access to full hand histories and the like.
1
KeyleK_uk United Kingdom. Nov 07 2010 01:52. Posts 1687
That's no good, too many people have both HU and 6max as their main games and want to be able to play both on one site.
Can somebody step in to tell me how collusion is usually investigated? What is needed?
1
genjix2 United Kingdom. Nov 09 2010 20:35. Posts 46
Collusion is found using hypothesis testing. It says given a sample mean and standard deviation from our population of players, what is the chance these measured variables (standard deviation for folding hands, playing speed, ...) is down to chance?
If these probabilities are unacceptably low then you could have a bot. However if many of these variables are all low then you most likely do have a bot. These actions don't reflect how a human player would play.
That's what PokerStars means by statistical analysis. The more data to data mine and find patterns amongst, the better.
On October 16 2010 18:47 Fudyann wrote:
It is possible to play a hand of poker "peer to peer" or without a trusted party without anyone being able to cheat.
No, it is not. /thread
On February 19 2009 22:21 Confedrate wrote: i dont get it
On October 16 2010 18:47 Fudyann wrote:
It is possible to play a hand of poker "peer to peer" or without a trusted party without anyone being able to cheat.
No, it is not. /thread
Cheat as in fake your cards. Preventing collusion is also an easy problem to solve: simply make all matches headsup. The hard problem is solving it for ring games. Any ideas?
On October 16 2010 18:47 Fudyann wrote:
It is possible to play a hand of poker "peer to peer" or without a trusted party without anyone being able to cheat.
No, it is not. /thread
Cheat as in fake your cards. Preventing collusion is also an easy problem to solve: simply make all matches headsup. The hard problem is solving it for ring games. Any ideas?
No - 2 players in different parts of the world can always be on the phone and there is no way to even try to inspect them if you can't see their holecards or their previous hand histories or table selections.
Collusion is not a very big problem in poker. It's not really a problem at all frankly. My main worries as someone without computer security knowledge is security of hole cards and security of my money. Convince me of that and I'll play on your site in a shot and so will the fish - they don't care about collusion either.
Researchers used brain scans to show that when straight men looked at pictures of women in bikinis, areas of the brain that normally light up in anticipation of using tools, like spanners and screwdrivers, were activated.
We can log table selection and hand histories, just not hole cards. The whole point of the system is that it's impossible to see somebody's holecards without that person revealing them, and equally impossible to misrepresent your holecards.
On November 12 2010 01:47 Fudyann wrote:
We can log table selection and hand histories, just not hole cards. The whole point of the system is that it's impossible to see somebody's holecards without that person revealing them, and equally impossible to misrepresent your holecards.
Which brings me back to no it isn't. especially not p2p when it's infinity easier.
On February 19 2009 22:21 Confedrate wrote: i dont get it
1
jchysk United States. Nov 12 2010 09:16. Posts 435
The problem with recording holecards would be the possibility of abuse for investigations beyond collusion. You'd probably have to have a pretty sick analysis system for flagging statistical anomalies and only bring those up for review.
On November 11 2010 20:24 Oly wrote:
Collusion is not a very big problem in poker. It's not really a problem at all frankly. My main worries as someone without computer security knowledge is security of hole cards and security of my money. Convince me of that and I'll play on your site in a shot and so will the fish - they don't care about collusion either.
Don't agree with collusion. I think it would be an even larger problem on a site where people think they can get away with it. As far as security goes, I would feel more secure on an open source poker room than a proprietary closed client.
On November 12 2010 01:47 Fudyann wrote:
We can log table selection and hand histories, just not hole cards. The whole point of the system is that it's impossible to see somebody's holecards without that person revealing them, and equally impossible to misrepresent your holecards.
Which brings me back to no it isn't. especially not p2p when it's infinity easier.
Are you claiming that it's possible to see somebody's holecards without that person revealing them? If so, I think you probably don't understand the cryptography behind this.
i feel we cant really plan this thing to perfection unless someone makes one first for people to try it, even if it's just play money at first then just fix or check things as you move along and see actual probs. talking about this constantly will only kill interest as time goes by and it would be a shame to see this thing just fade away.i mean no disrespect in any way, i hugely support what you are trying to do. we gotta get a beta thing started if that's possible.
1
genjix2 United Kingdom. Nov 13 2010 09:04. Posts 46
im working on it. give me time jeez.
Can I have 4 beta testers please?
- You must be able to use IRC.
- Be able to arrange amongst yourself a playing time within a days notice.
- Participate in a mailing list to offer feedback.
The more the better since arranging testing sessions would be easier as everyone need not attend.
Plus points:
- Any technical knowledge like admin of Linux.
- Mac OSX users.
- Power users who use Windows a lot.
- Graphics artist.
I need various work done. Every small task slows me down and adds up from doing the balls & meat core tech.
