we had a very bad run from nov 27 to jan 1
rest of the time we have been very good.
Yeah but it was actually a bit shorter than that - from the last block of November, as you said, but only up to the 745% block on the 16 December.
19 blocks and 1 orphan.
The CDF[Erl] for those 20 blocks
20 Blocks with a mean difficulty of 202.028% - yeah way high.
Gets:
http://www.wolframalpha.com/input/?i=CDF%5BErlangDistribution%5B20%2C+20%5D%2C++2.02028%5D= 0.999859 i.e. 1 in 7092 of being that or worse - sooooo very unlikely even for a very short run of only 20 blocks.
I've included the orphan in the block count since it was a valid block found and even sent out to the network first, but due to a complete double failure in the relay network at the same time as a GFW blockage sending it into china, it was orphaned by a block in china.
In that case, being orphaned was a systematic problem, not related to determining the CDF[Erl] when evaluating the probability of the data.
The whole month of December is 42 Blocks with a mean difficulty of 139.638% (again including the above orphan as a block)
since the remainder of the month averaged a mean difficulty of 106% for the last 21 blocks which is well within expectation.
