And considering people are saying this coin is no longer currently profitable to mine with EC2 instances, and afterwards the hashrate triples or something to that effect this morning, I would have to conclude someone found a way to write up a GPU miner.
Look above, crypto_zoidberg showed there is overwhelming evidence it is ec2 (and someone said 60% of their peers were amazon). The sooner the public GPU miner, the better. That would make net hashrate much more stable, and if someone has a private GPU miner, level the playing field.
At the risk of providing facts: I was perhaps 1/4th of that increase - with apologies. It was EC2, no GPU.
I suspect I know what happened: I looked at the diff, calculated that it was profitable on EC2 to mine at a particular rate, and launched quite a few hundred c3.8xlarge instances. My goal was to let the diff creep back up to where it should be based on the hashrate, and top out at the end of where it was profitable on EC2, and then kill the instances.
... what I forgot is that this wasn't a vacuum. The hash rate increased *way* faster than I thought it would, and my guess is that about 10 other people all had the same idea at about the same time when the profitability calculation looked positive. One or two of them probably launched up to 1000 instances.
The combination of all that resulted in a huge diff spike (which I think might be interacting with a mis-tuned parameter in the difficulty retargeting). When I got up this morning, I saw the diff, said words I won't utter in a public forum, and immediately killed all of my instances. Looking at the hash rate graph, I think other people still had instances starting up or running, because it kept increasing even after I killed about 600mh/s. And then it tanked when all of the cloud miners suddenly realized they were bleeding away money.
I'm not going to do that again.
