ich hab schon viel quark gesehen, aber das dice-fest hier
replay runterladen und ansehen. so geschehen heute in der Orca Cola. und da wollten mir ernsthaft heute mittag noch leute erzählen, 2x MA8 claw frenzy ist nicht so mächtig
auch wenn mich solche spiele schwer am RNG (würfelbot) von BB2 zweifeln lassen, so gibt's in einem französischen forum doch einen nachweis dass der RNG in BB1 sehr gut funktioniert. vermutlich wird genau dieser RNG auch in BB2 verwendet, nur eben auf serverseite um manipulationen zu erschweren/auszuschließen.
There have been a nigh-infinite number of threads on various forums discussing the randomness or supposed lack thereof, of the dice system in the Cyanide/Focus version of Blood Bowl.
First off, there have been numerous programs written to analyze the distribution of the D6 and block dice rolls from the logfiles of individual games. These have universally found that the distribution of rolls is exactly what would be expected - each result has a 16.667% chance of occurring at any given time (twice that for push as it covers two sides). That doesn't satisfy the conspiracy theorists, obviously.
I took this a step further by tracing the execution of the program to find the RNG code, and the code associated with converting the RNG's output into D6 or Block rolls. I then ran all possible values of the RNG's output through those routines to obtain the probability, over time, of obtaining each of the possible values. I'll give you those results in this posting.
First of all, the RNG used in the game is called the Mersenne Twister - more specifically, a variant known as MT19937. The MT RNG is actually one of the faster and stronger RNGs - the period length (number of RNGs generated before repetition begins) is 2^19937 -1, which is officially a Really Big Number.
The standard MT19937, which is what the game uses, generates a 32-bit number from 0 to 4,294,967,295 so all of those numbers were run through each of the conversion routines (despite block dice being 6-sided dice, the two routines are completely different... different means of accomplishing the same thing... generating a number from 1 to 6). Here are the resulting percentages of result distributions:
For the D6 rolls:
Code:
Roll Instances Percentage
----
1 715827899 16.666667
2 715827899 16.666667
3 715827898 16.666667
4 715827899 16.666667
5 715827898 16.666667
6 715827802 16.666665
For the Block Dice:
Code:
Attacker Down 715827885 16.6666667
Both Down 715827882 16.6666666
Push 715827882 16.6666666
Push 715827882 16.6666666
Stumble 715827882 16.6666666
Defender Down 715827882 16.6666666
So what do those numbers mean? It means the code used to create the dice rolls from the RNG output create a dice value does generate the proper distribution. You might notice that the numbers are not perfectly distributed - that there is actually a slightly greater chance of getting an Attacker Down (skull) than another roll, or less likely to get a 6 on the D6 roll than a 4, but that needs to be looked at in context. Look at the number of decimal places you have to go before that slight offset, or how little difference there is in the number of instances. Real dice will not offer you that even a distribution... not by a long shot.
Likewise, there are no insidious routines that "look ahead" at the dice rolls - the rolls are created as they are used by the game. People can speculate all they want about that sort of thing or they can just take a look. That's what I did.
If there are any further questions about the RNG and its legitimacy then I'll happily answer them since the company answers no relevant questions on pretty much anything. I won't lay out the exact process to reproduce the RNG and dice roll creation (which is what BBOracle does, effectively) but most other things are game.Quelle
natürlich dürft ihr hier auch screenshots oder replays euerer schönsten begegnungen mit Nuffle posten