DON'T PANIC!

This bot has actually been active for several days.  I first noticed the pattern in the filtered maps before the weekend.  It was originally solving maps rather slowly, and not running full time.

It is, in some ways, a significantly more impressive bot than anything that we saw in "round one."  It appears to be filtering for low density maps, with low "diagonal" weightings to the perlin.  It then solves these maps through (I'm assuming) the same sort of annealing, but does so very quickly compared to round 1 solvers. It seems to be running with some decent processing power behind it as well, given the rates.

Does anyone want to step forward and take credit for it?

I'm not sure why the bot operator decided to "push" much harder all of a sudden. The only thing I would assume is that he actually wanted to create some panic sells to accumulate cheap. (It looks like he got exactly what he wanted.)

Having this bot active is not strictly a bad thing, however.  Although I'm personally kicking myself for not having found the time to finish a competitive bot first (hehe) having this bot come online does solve the "no hash-rate" problem we were facing.

Although I'm still mulling over all of the factors involved, I think right now that our best course of action is probably to simply do nothing.  We fully expected that bots would come online again at some point, and even fully expected them to have the capacity to dominate at low difficulty.  Because this bot is still somewhat reliant on a map filter we can expect it to be hit pretty hard by the anti-warp work kicking in.  Because the operator has decided to remove any throttle, we can expect that anti-warp to ramp up quickly.  I say we let it run the course, and take the opportunity to see a pragmatic test of the anti-warp in action.

Of course this creates some short term inconvenience for human miners, followed by a potentially longer-term period of slower map generation as the warp debt is repaid to the network.  I will move straight away onto background map generation work, which should alleviate much of the inconvenience of the slowdown.

If anyone thinks that more drastic measures should be taken more quickly (such as increasingly stricter map criteria, a larger hashing work requirement, etc) let's discuss the potential implications and consider it.  If anyone has other suggestions for better re-balancing work requirements, let's hear them!