Pointing to serverwide statistics isn't as strong an argument against allegations of manipulation as it may seem.
What you are really suggesting--and all without any evidence or motive--is that in 1992 Andreas Schneider (aka marvin) intentionally programmed the FIBS software to diabolically "manipulate dice" and then foisted that off on the public on July 19th of that year.
What happened is that in your rush to brush aside and casually dismiss what you thought to be a sophomoric argument presented by statistician Turner's analysis, you perhaps hadn't fully read the article which I had linked but not quoted, or perhaps you read it, but not truly comprehended it.
Imagine the following scenario: Unbiased dice generator (DG), exactly two players are logged into fibs. Fibs is programmed to behave in the following way: If on player 1's turn the DG rolls a 66, the roll is put on hold and instead the player is given the next roll produced by the DG (if that happens to be a 66 as well, the process is iterated until one of the other 35 rolls comes up). The 66(s) is(are) given to player 2 on her next turn(s). Result: Non-random dice sequences for both players, while any serverwide stats like dicetest will detect no bias, no matter how large the the number of rolls.
Though intellectually stimulating, there is no need to tax our brains with this scenario.
I don't think you understand the matrix command that Turner used to collect his 10,593,121 FIBS dice rolls and to which I also alluded in my previous post. In his analysis of FIBS dice Turner says of the matrix command which was used to collect the data:
You probably don't know about this command because it's not on the list of commands, but it produces data not only on the frequency of rolls, but on the frequency of each of the possible 1296 possible pairs of consecutive rolls.
(Bold added for emphasis)
fergy, graphic designer of the old MacFibs interface and the GamesGrid (now Grid Gammon) graphics and who still occasionally plays here puts it thusly:
Syntax: matrix http://www.fibs.com/~fergy/help/fibsCommands/pages/matrix.html
Display matrix of conditional roll frequencies. The matrix command shows you for each of the 36 possible combinations of numbers on the dice, the distribution of the next roll. For each of the 36 possible outcomes you'll see a matrix showing you how often each pair of numbers was rolled after that outcome. The numbers in the matrix represent the following rolls:
1-1 1-2 1-3 1-4 1-5 1-6
2-1 2-2 2-3 2-4 2-5 2-6
3-1 3-2 3-3 3-4 3-5 3-6
4-1 4-2 4-3 4-4 4-5 4-6
5-1 5-2 5-3 5-4 5-5 5-6
6-1 6-2 6-3 6-4 6-5 6-6 (X-X)
The last number shows how often a double was rolled after the given outcome. At the end of the list you'll also be told how often it has happened that a double was immediately followed by another double, together with Theoretical value for that number.
The values used for this test are all rolls that FIBS has made since it was last restarted. If the number of rolls is very small (i.e. smaller than 250000) the output of this command might lead you to wrong conclusions.
So I think its quite reasonable to say that pointing to server wide statistics, especially those generated by the matrix command, provide a much stronger argument against dice manipulation than pck would suggest and that Turner's analysis of 10,593,121 rolls of FIBS dice, though not conclusive, is certainly far more compelling by an order of several magnitudes than that of any utterance by a caterwauling dice whiner.
For me, the most compelling argument against FIBs being programed to manipulate dice is that there is absolutely no evidence that Marvin did this and there is absolutely no motivation for him to have done so.
Fifteen years ago in 1996, several months before Turner completed his analysis of 10,593,121 FIBS dice rolls, Kit Woolsey modestly gives the reasons why his opinion should carry some weight:
There has been much discussion on r.g.b. about the possibility of the FIBS dice being skewed, so I thought I should put in my two cents worth.
I have been playing on FIBS for about three years, and have logged plenty of experience points. It is my personal judgment from my playing experience that the dice are not biased in any way.
How qualified am I to make such a judgment? I have been playing backgammon for over 30 years, both money play and tournaments. In addition I have been active in many other gaming activities which involve chance all my life.
When one is involved in gambling, there is always the danger of cheating. It is important to be on the lookout for irregular occurrences, and to distinguish them from the normal lucky and unlucky sequences which will always happen. On more than one occasion I have sensed that things weren't right, and walked away from a situation which could have cost me a lot of money had I stuck around. Thus, from my experiences in the gaming world, I believe my judgment in this area is better than most people's judgment.
Kit goes on to address the point I made earlier that there is absolutely no reason for Marvin to have rigged the dice:
The claim that the dice are intentionally programmed to be rigged in some way is patently absurd. What would marvin's motivation be? He has no profit from FIBS. He created the program because he enjoys backgammon and wanted a free backgammon server to be available on the net. In addition, programming the dice to be rigged is much more difficult than programming random dice. Any such assumption about the dice being intentionally programmed to be rigged is completely contrary to logic.
(For Kit's full post from October 16, 1996, go here: http://www.bkgm.com/rgb/rgb.cgi?view+52
As I understand it, it was Kit and his wife Sally that initially paid the FIBS server costs.
Overall, I suspect that pck regards this discussion of FIBS dice to be more of an intellectual puzzle than he is convinced there is outright dice manipulation and conspiracy, but in the event he gets too carried away with attributing sinister motives to Marvin and dice conspiracy theories, I have Ockham's chainsaw ready.