I'm not so sure it's just variance. I wrote a script for monero that xnbya runs with some modifications here ( http://minexmr.com/pools.html ), and in my experience it's not sometimes, it is all the time that the sum of the known pools hashrates adds up to be 20-33% larger than the hashrate calculated from the diff. I think the diff is based on last 24 hours, and the pools base their hashrate on the diff of blocks over the last 10 or 30 minutes. Maybe I'm wrong, but if it was just variance in the hashrate, then at some point the calculated network hashrate would be higher than the sum of the pools (since this isn't even including unknown/private pools and solo miners), but in my experience this is never true - the sum of the known pools is always greater than the calculated network hashrate. I don't know why this would be, just an observation from monero that also seems to hold true for boolberry and bytecoin.