I paste here an interesting comment of a dev of World of Warcraft explaining the guessed problem of roll hacks. But the reason why I do so aren’t about the game:

There have been a bunch of questions, a lot of concern, and a ton of speculation about how WoW’s “rolling for loot” works. I want to give you guys some info to help everyone better understand this system.

– The random numbers are generated on the server, not the client
– There is no “re-roll” mechanism. You tell the server you want to roll and it generates a single random roll for you. Network retries will not make it re-roll, nor can the client ask for one. If you asked it to roll again, it knows you have already rolled and will not do it.
– The loot roll does not use the same packets as /random
– The packet for a loot roll conatins this info: which corpse, which item, and boolean (yes/no) if you want to roll or pass; that’s it.
– The server sends the roll results (the value rolled or that you passed) to all the party members. You could hack it to make your client see the wrong roll, but the server would still know what you really rolled and award the loot accordingly.
– WoW currently uses the IBAA pseudo-random number generation algorithm. We have a library of generators and can switch at any time if the need were to arise.

Another popular theory I want to touch on is “hot streaks”. This is the same logic that made La$ Vega$ into the city it is today. People like to see patterns where none exist. They also like to look at some statistics but not others. Let’s take an example. What are the chances of rolling 98 or higher? Simple: 3%. What are the chances of rolling 98 or higher 5 times in a row? About 1 in 4.1 million. If you stop there you would think that if you saw somebody roll 98+ 5 times in a row, they *must* be cheating! But let’s not stop there. How many random rolls are there each night? Many millions. So what are the chances that a noticable number of people did indeed roll 98+ 5 times in a row? Virtually guaranteed.

At the first reports of suspected roll hacking, we did a review of the loot rolling code and found no vulnerabilities. We still decided to go ahead and log all loot rolls so that we can address suspected roll hacks with real information. Since that time we have checked these logs against claims of roll hacking and have as yet found nothing to substantiate any of the claims. In some cases we have seen direct evidence that the claims were false in the first place.

In summary, there are no known roll hacks. We will continue to monitor and investigate these issues.

I particularly like the line: “People like to see patterns where none exist.” It fits perfectly in the model described by Raph Koster.

This doesn’t apply to the players. This applies to market analyst, in particular those who love polls and statistics. The statistics itself is a completely arbitrary “science” that, since is considered science, is able to pretend to be objective. A contradiction, of course.