Look, this is approximately how the statistics were worked.
Let's say MN has been a bit unstable tonight. There has been brief loss of service at
20:17
20:21
20:38
20:55
20:56
21:05
21:19
21:23
21:33
21:36
21:49
There were some updates done during the day that one of the technical crew said might mean that MN is unstable.
However they decide this must be due to a hacker, without considering the updates.
They chance upon this thread and realise that FireFox has posted at
20:17
20:38
20:55
21:19
21:23
21:33
21:49
They conclude without looking at any other threads that FireFox definitely must have had something to do with it because they posted at those times when Mn was unstable.
They decide that the other times when they'd initially thought it was unstable when FireFox hasn't posted don't count. There isn't really any difference between them, but now they can see FireFox hasn't posted at those points they ignore them.
So the reality being FireFox was on at 7/11 or 63% of the time, but they're considering them being there 100% of the time.
They also haven't taken into account that when FireFox starts posting on this thread they tend to post a lot, increasing the likelihood that their posts are more likely to correspond with times. If someone only posts normally once or twice on a thread, they're less likely to hit the right times.
So when FireFox appeals against being banned, they put to the appeal committee (that I don't think they have) the following information:
FireFox has posted at the following times which are all the times it was unstable.
20:17
20:38
20:55
21:19
21:23
21:33
21:49
They don't mention that FF posted more than anyone else on that thread between those times.
And they also fail to mention the updates that they were warned might cause issues.
So the appeal upholds the ban because they assume that
- MN has looked at all threads and this is the only person who has posted at the times when Mn was unstable
- FF was on every time it was unstable
- There is no other obvious cause.
All of the above assumptions are incorrect, but this is all the information the appeals committee has so they are not considering the full picture.
Yes, this is a basic example, but surely you can see how by assuming it has one answer they are making it a self-fulfilling conclusion.