A followup to Part 1 and Part 2.

Things that I am reasonably confident are true, but cannot prove or say for sure:

1. AJ Green is no longer involved with AP, UB or any related company in any way, including ownership interest.

I’m pretty sure the guy is gone. Why? Mainly because the guy probably didn’t have significant ownership interests. Unlike owners, it’s relatively simple to get rid of an employee. Sure, I’ve heard rumors that AJ either got small chunks of equity via work or investment and that’s certainly possible, but who really knows if that is true. What I am pretty sure of is that AJ did not found the company and that he did not put up initial funding money — and logically, those two sets of people are likely to be the owners. From what I’ve been told, AJ met Scott/other Montana guys a little later on and was brought into the AP/related companies mainly as an employee (or maybe consultant, not sure of the legal relationship). The key is that he was not there at Montana in the beginning and is therefore pretty unlikely to have an ownership stake. Also, ex-AP employees told me that they got the feeling that AJ did not own a piece of the pie.

Therefore, assuming my theory that AJ was not a significant owner of AP, it’s pretty easy for them to get rid of him. Also, I don’t know why they would remain loyal to someone who put them in a spot like this. In addition, there are a bunch of eyes still focused on AP and people know what AJ looks like. How would it look publicly if in May 2008 pictures surfaced of AJ walking into the AP offices in Costa Rica or Panama? It would obviously be an enormous controversy and press/forum coverage would undoubtedly be huge. AP simply cannot afford to have themselves tied to someone who they’ve basically claimed to have gotten rid of and someone who they’ve basically admitted either perpetrated or led this whole episode. It’s way too big a risk for them to have AJ communicating via email, phone, in person, etc with the known AP executives and decision-makers and I believe they are aware of that.

2. The cheating was perpetrated via a backend access tool called “Servman” that wrote hand histories to the page before hands were completed.

For all of the discussion of user #363 being some sort of superuser who was watching the table and relaying cards to POTRIPPER, I doubt that was the case. User #363 was another one of the four cheating accounts named, although I cannot publicly identify which one. On a side note, assuming that the #363 name was not changed in recent weeks, the name on the account was not a name that I recognized (note that AP did not show me the name, I found out via another method while in CR). Anyway, if #363 was another cheating account, it does not really make sense within the context of the theory that there was one superuser account that never played while the cheaters used a bunch of other names to stay anonymous. After all, the #363 account did play and did cheat and did win. Therefore, it’s pretty clear there was another method of seeing the holecards.

As has been described to me by a few ex-employees, AP uses a tool called “Servman” to manage a number of things. Servman is a web browser-based backend interface where you can do things like type in a player name and view that player’s deposits/withdrawals/rake/etc. All sites use Servman or something similar in their customer support dept. In addition, Servman had a tab called “Game Log” and, as I understand it, you could click on the tab and put in some information and bring up hand histories. According to an ex-employee of AP, the game log would typically write to Servman 20-25 seconds after the hand was completed. Therefore, during this ex-employee’s time at AP, viewing real-time admin level (ie, with all hole cards) hand histories at the customer service level was impossible. However, this employee was able to look up hands in such a manner pretty soon after hands were over. In addition, I was told that Servman had IP-restricted logins for most people. So the typical employee could not login to Servman outside of the AP offices. However, there were a number of admin level logins to Servman for the people who were running the company — and many of those logins did not have IP restrictions on them. According to someone at AP, there were probably about 20 such admin level Servman accounts and I suspect that the cheating was perpetrated via one or more of those accounts.

4. The reason Servman could see hole cards was due to a code change made by AP programmers in mid-June 2007

I don’t believe it’s really a problem for the software to let certain individuals access Servman or other backend tools outside of the office. After all, if an executive travels a lot, it may be necessary. The problem was that the software was writing hands in real time. I don’t really know if the problem existed in all Servman accounts, but I was told that in mid-June there was a software change that altered when hands in progress were written to the database. After the aforementioned change, hands were written in real-time, allowing Servman to query for partial hand histories. Why do I believe this is true?

1. Because of what ex-employees told me about how Servman used to work
2. AP was making code changes in mid-June 2007. According to this post, Scott Tom or someone using Scott Tom’s email address was looking for beta testers for the AP 8.0 software in the June timeframe. Also, another person told me that he started using AP 8.0 beta on June 15th, 2007. In addition, given the timeframe of the AP 8.0 release this past summer, it seems obvious that AP would have been beta testing it in June.
3. If AP was testing a new client, it’s very possible that they needed to make changes on their production game server. The production game server is the software that is (hopefully) running 24×7 serving games, tournaments, etc to the AP clientèle. I am far from a software expert so I really can’t tell you what goes into making a client work smoothly with a game server, but obviously a completely new client might require some changes to the game server in order to roll it out for testing. Therefore, if the code was open and being edited, it stands to reason that a database-writing change could have been made either by accident or, possibly, on purpose.

I don’t know why the change was made, but I do believe that it happened in June 2007 and I believe this change was responsible for providing real-time hand history access to whoever was in the Servman backend during the cheating incidents in August and September.

3. The aforementioned backend access software can no longer access hole cards before a hand is completed.

I don’t have any real proof of this. But, obviously, people at AP told me that it was closed and they couldn’t really “prove” it was closed because showing me around the backend doesn’t conclusively prove that it doesn’t exist somewhere else in the system. More than anything, it was a read of mine that they were telling me the truth. I didn’t believe a number of things that I was told or shown, but I did believe this one. And, logically, it seems like they could have just switched their code back over to the old system of writing hands pretty easily assuming my early theories are correct.

Related posts:

1. Absolute Poker Costa Rica Trip: Part 1
2. Absolute Poker Costa Rica Trip: Part 4
3. Absolute Poker Costa Rica Trip: Part 2
4. PokerNews Q&A Part 2: Absolute Poker
5. 60 Minutes considering Absolute Poker story