Board Thread:Game Discussion/@comment-184.58.159.118-20130905114258/@comment-11490363-20130905162234

SchappX54G wrote: How can you still hold this line after the BS email you got from TFL support?

You also repeatedly claim that there's "more" evidence for 50/50 than not. How do you quantify your "more" statement? It's just your assumption that you choose to mentally validate, no different from those who say it's NOT 50/50. But after that BS email they sent back to you as well as previous correspondence that would suggest there is bot or alt weighting in the paid bridge, I think the "evidence" combined with the raw data doesn't smile at you. What email? The email *I* got said he 'did not have access to that information', so I have no conclusive evidence either way from them. And I am not in the habit of jumping to conclusions just because I hate Mobage or think they are out to screw us. If people want to assume they are being unfair then so be it, but I just look at the numbers and the numbers don't support that theory to the best of my knowledge.

Someone else claimed that Mobage told them it was weighted to alt or bot depending on the episode but no one offered the email as proof, and the evidence we have seen directly refutes that.

My evidence: If it is weighted to one side, there would be less streaks of whichever one is less likely, and we are seeing streaks of both types and very long ones, which is more consistent with a 50/50 guess than not. Their appears to be equal numbers of streaks indicating that the relative percentage is equal, hence 50/50. If there were 3 times as many X bot streaks as X alts, then that would imply a 75/25 split, but that is not the case.

If it *is* weighted it is a very minor weight, indistinguishable from 50/50 due to the small sample size. The more samples we get, the more refined the potential range becomes, and it seems to be narrowing on 50/50, not anything else.

That said, the 20 or 30 people that post their results is not conclusive either way, we just have a confidence interval at best. We would need more samples and a distributed sample set, voluntary reporting is ALWAYS biased.