Different algorithms and service locations were differently affected. For some (Sha256 and Scrypt on NiceHash) were almost unaffected, because these most likely had big fat orders with enough of BTC loaded.

Quark on NiceHash for example, was severly affected - I remember the price was 10 times lower than usual when we started web back up, but then in few minutes price went up as buyers started placing new orders.

A lot of buyers do not put a lot of BTC, they rather use refill method every so often. If you check Quark order list now, you can see that not a single order has more than 0.2 BTC as remaining balance. The total hashpower of Quark before outtage was around 50 GH/s and the total when service came back around 30. You can do some simple math to figure out that these slim orders were quickly depletted in hours of downtime.

Also worth to note is that with parameter p= we give every miner an option to stop mining if price falls below this treshold. Not a single other pool has this option. If you do not want to mine for change, you should always set this parameter. Did you use it?

We did not skim or earn any BTC from this unfortunate event. Buyers are paying miners. The only beneficiers of this event are buyers that placed low paying orders. Maybe you should blame them, because they didn't put high enough prices.