Today, the total computation speed of the network moved from a low of 3000 GH/s to a high of 5000 GH/s, while the percentage of non-pool mining increased dramatically, involving deepbit and slush both going down.
Clearly something doesn't add up. Why does the computation speed go up when large pools are out of action?
There are also claims on the forum that two large FPGA clusters were brought online today. And other claims that the pools going down were due to DDOS attacks.
Draw your own conclusions.