I’ll answer that for Mike.
A hash is a hash. For the statistic above, we are using the total number of hashes delivered, as recorded in our database. This is not an estimate of cumulative revenue, but merely a measurement of total work done.
Measuring the ZEC that would have been mined cumulatively is a different issue. I just ran some numbers on a python script that uses the difficulty for each block and the varying slow-start reward, and it looks like our customers should have earned about the same amount by now if we had constantly hashed at 1.0x versus if we did what actually happened (0x for 2 days, 1x for 2 days, then 1.75-2.0x since). It would vary from customer to customer, though, depending on whether they got more or less than they were owed on days 2-3.
We are not hashing at 1.95x instead of compensating customers for the downtime. We are doing it in addition. I do need to finish up the compensation plan, though; mea culpa for not getting it out by now.
Here are some interim numbers to tide you over. These are the ideal revenue for someone with 1 H/s active since block 0 hashing on a perfect pool (comprising 100% of the network hashrate) with 0% fees and with 0% orphan rate:
0.0013592462 ZEC by block 1637 (t = +48 hours) – when we came mostly online
0.0017693052 ZEC by block 2803 (t = +96 hours) – when we began to exceed targets
0.0037666281 ZEC by block 8872 (t = +350 hours) – now
Let’s say customer B got this (with a perfect pool, etc):
0% for the first 48 hours
75% for 48-96 hours
175% for 96-350 hours
B would have earned roughly:
0.0000000000 ZEC by block 1637 (t = +48 hours)
0.0003075442 ZEC by block 2803 (t = +96 hours)
0.0038028593 ZEC by block 8872 (t = +350 hours)
Our actual performance was better than this Customer B model for most of our customers. Customer B represents roughly the 20th percentile for what our customers received.
Our compensation plan will be based solely on the 0.0013592462 ZEC/(H/s) and 0.0017693052 ZEC/(H/s) numbers, and will ignore everything that came after t = +96 hours.
Edit: These numbers were all 1/4 of what they should have been due to a bug in my python estimator code. Fixed.
Edit 2: I made a mistake with the block 8872 lines. I had used the amount for block 5228 by mistake. It has been corrected to 0.0037 and 0.0038 ZEC.