There's a bug in this game where if you click on any of your units during your opponent's setup before a kickoff and it is left selected when it becomes your turn to setup for the kickoff then that player will disappear off the field and you'll only have 10 players instead of 11. This bug was reported around 3 years ago. How is it that such a major bug has not been fixed by now?
Clicking on units to see what skills they have is a natural thing to do, especially if you have many teams and don't remember exactly what kind of skills your players have. I've accidentally run into this bug many dozens of times and run into it probably at least a hundred times before that before I found out what caused it. If I have run into it so many times then I think probably a lot of new players have run into it and thought "wow, this game is buggy" and left and never come back. Is there no concern for the quality of the product?
That's the vanishing player thing sorted.
Since the original article about the RNG has been requested to be removed by the author I decided to do some cursory digging and write my own. My main source of information for this is pmcc, Cyanide's lead programmer for BBLE. Specifically, I asked him what type of RNG was used, whether the result for d6 and block dice was calculated in the same manner, whether the algorithm turning the RNG output into a die roll does so evenly, and whether there were any other inputs altering the die roll (such as the AI). Clearly for the die-hard conspiracy-theorists asking the actual programmer to tell us how it works is akin to asking Nixon if he is a crook, but perhaps for those people even the original article would be unconvincing.
What RNG is used?
The RNG used is a Mersenne Twister MT19937. Wikipedia (if you trust it) tells us that this particular RNG has 3 desireable properties:
1. Long period. This is the number of iterations before the RNG repeats itself. In this case that is (2^19937) - 1, so really quite a lot (in the order of 10^6001).
2. It is k-distributed to 32-bit accuracy for every 1 ? k ? 623. That means that all sequences up to 623 numbers long are equally probable. This is important in BloodBowl, where typically a few hundred dice will be thrown. The "32-bit" part means that it produces a number between 0 and (2^32)-1, or 4,294,967,295. This is basically the output of the RNG.
3. It passes various statistical tests for randomness.
What does the game do with the RNG results?
What we need to play BB is generally d6 results or variants thereof (e.g. d3). Sometimes a d8 is needed as well, but I don't know how this is produced.
If we want to take our 32-bit number and turn it in to a number between 1 and 6 inclusive then there are two ways to do so. We can either simply divide the number which is output by 1/6th of itself (715,827,883) and round up (producing 1 to 6), or we can perform a modulo +1 operation on it. A modulo, for those who don't know, is the remainder after division. We take our number and divide it by 6, which will give us an integer and a remainder. The remainder will be a number between 0 and 5, so adding 1 to it gives us our 1 to 6 range.
BloodBowl uses both. One operation is used to convert the number to d6 results, and the other is used to convert into block dice. Interestingly, 2d6 results are calculated using the same algorithm as the block dice results, such that if DD were to correspond to 6 then two DD results would be 12 on 2d6, but it might be any numbers on two separate d6 rolls - there is no correlation between the two.
Are the dice fair?
The last two questions I asked pmcc were to do with the fairness of the dice.
Q. Does the algorithm turning the RNG output into a die roll do so evenly?
A. Yes.
That seems reasonable to me, given the simplicity of turning the output number into a number between 1 and 6 inclusive. I believe the original researcher found some very slight discrepancy in the number of 6s compared to other numbers, but the difference was in the order of 0.000014% difference.
Q. Are there any other inputs altering the roll?
A. No.
Again, given the fact that three different dice checking programs were created and all showed that the dice were fair, even against the AI, this seems fair.
Given the above, I for one am happy with the dice. If you aren't then maybe you could ask VoodooMike to explain it to you, since it was him who wrote the article with the original research, and him who wanted me to not post it; hence, you are stuck with my own ramblings instead
Digital Nuffle in all His glory!
