Here are the times and share totals for our last 11 payouts. I am not including the first payout because there are some cex.io shares in there.
+------------+------------+
| time | diff | hashrate
+------------+------------+
| 1402976440 | 4994809796 | 232TH
| 1402884360 | 1001667156 | 213TH
| 1402864221 | 816766904 | 151TH
| 1402841060 | 5430841828 | 256TH
| 1402750091 | 1868165592 | 222TH
| 1402714003 | 786003908 | 191TH
| 1402696412 | 596053444 | 144TH
| 1402678654 | 1687252668 | 165TH
| 1402634828 | 4918289664 | 232TH
| 1402544157 | 1333388092 | 179TH
| 1402512186 | 5877865492 | 283TH
I calculate an average hashrate of 217TH.
Using the difficulties for each day and the sum of work difficulties, I get submitted/expected= 2.493172.
This has an upper tail probability of 0.08264739, meaning that once out of every 12 reruns of those eleven days you'd see multipool have this level of luck. Not very unlucky.
2.5*diff shares per round happens multiple times per day for many different pools (well, those that solve multiple blocks per day, anyway), so it shouldn't surprise us when a particular miner submits work equivalent to 2.5*network difficulty and only solves on block.
Sorry, Entropy-uc, I'm not seeing a problem here. There might be a problem but statistically there's nothing unusual going on. If submitted/expected reach 7.0 (one in a thousand) or 8.0 (one in three thousand), I'd start to worry.
That's nice for you. If I wait for that level of proof I'm out enough money to buy a luxury car, or a small house.
Flound can't even vouch for who his users are. So how can he attest that they aren't using defective gear, or intentionally withholding?
Your numbers say that 92% of the time a honest contributor would not have performed so poorly. That's is enough reason to freeze payments until more information is available.