Ultimate COO Paul Leggett has given an update to the blog at Ultimate Bet about the Phil Hellmuth v DOUBLEBALLER hand that caused all sorts of controversy in the online community since Saturday morning, a portion of that comment is excerpted below:
As we explained in our press release, the incorrect payout was caused by a software malfunction. The software malfunction occurred when the winning player was disconnected from the hand at the precise millisecond the software was determining whom to award the prize. This in conjunction with the “player’s state†data being cleared from memory caused this extremely improbable incident to occur.
Originally we were unable to reproduce the error in our test environment, until we programatically forced a disconnect to happen at the precise millisecond the hand was being awarded.
We have never had an issue like this reported previously and we have reviewed all of our database and application logs for December and November and have been unable to find any historical occurrences. We are continuing to go back further into the historical data to ensure no hands from the past need to be corrected. Analyzing this data takes time but we will continue to go back further and further into our history to ensure no one else was affected.
Leggett\’s entire statement can be found at the Ultimate Bet blog.
Tokwiro Enterprises, who own Ultimate Bet and Absolute Poker, issued their own statement on the matter, after the jump:
WINNING HAND CORRECTION
MONTREAL (DECEMBER 23, 2008) – Tokwiro Enterprises (\”Tokwiro\”) issued the following statement today, concerning the correction of a winning hand played on the CEREUS Poker Network on Saturday, December 20, 2008.
Paul Leggett, Tokwiro’s Chief Operations Officer stated:
On Saturday, December 20, 2008, two players DOUBLEBALLER and PHILHELLMUTH were playing heads up at a 200/400 table on UltimateBet.com. DOUBLEBALLER won hand #1162170993. The software malfunctioned and awarded the losing player PHILHELLMUTH the pot, instead of the actual winner DOUBLEBALLER. After learning of the incident, we credited the player the correct prize to his account. We then began an investigation into the malfunction. We have now completed our preliminary investigation into the software malfunction regarding an incorrect payout in hand #1162170993 on the CEREUS Poker Network.
We have concluded that the malfunction occurred because the winning player disconnected from the table at the precise millisecond when the software was awarding the pot, in conjunction with the \”player’s state\” data being cleared from the memory cache.
Phil Hellmuth, who was involved in this incident, said, \”The only time in my life where I’ve had one pair beat three of a kind is when I made a big bluff in a poker tournament and now because of a software malfunction in online poker. I am one of the most watched players on the Internet, and all I can say is that it is pretty obvious that there was no malicious intent and even more obvious that UB handled this problem well. Sometimes these things happen on the Internet. The important thing here is that I continue to have a ton of trust in UB’s software and new management team.\”
There is a prevention system in place designed to protect hand results against disconnections. However, after conducting an extensive source code analysis, we found logic in our code that we have enhanced to further ensure this malfunction cannot reoccur. We were able to reproduce the error on our testing systems by forcing a hand to disconnect its winning player at the precise millisecond the hand in question was awarding the pot and simultaneously flushing the \”player’s state\” data. Had the disconnection happened literally a millisecond sooner or later the error would not have transpired.
While enhancing the software to prevent this malfunction from occurring again, we introduced a new defect that caused 36 hands to award incorrect payouts. All players have been reimbursed for these hands as well.
We have implemented a new method into the code and updated the game servers. Both errors are no longer reproducible. Furthermore, we are conducting extensive tests to simulate every possible disconnection scenario to ensure this type of malfunction cannot occur in the future.
We are currently reviewing historical hands in our database to see if there ever were similar incidents in the past. Thus far, our investigation has been unable to unearth any additional historical occurrences of this malfunction. Our investigation will continue until we are satisfied that there are no additional historical occurrences.
We recognize the severity of these issues and we will continue to give them the attention they deserve until we are absolutely certain that they cannot happen again.