Quote from: thefunkybits on December 09, 2014, 08:41:52 PM

Quote from: TracerX on December 09, 2014, 02:49:07 PM

Quote from: FlyForFun on December 09, 2014, 02:46:15 PM

Hopefully the difficulty will last few more months. I am sick of seeing the difficulty halt for 2 weeks and jump even more then the last month added up.

It looks like it may remain low, or even decrease again in seven days. That said, there's no telling when a new mine will come online.

The last time difficulty decreased it slumped for about 6 months, Maybe if the price picks back up we'll get another boost in hashrate

Almost certainly. I forecast a significant slowdown in the hash rate expansion back in April, which to be fair wasn't as accurate as it should have been because I overestimated the cost of wholesale hashing and electricity (http://hashingit.com/analysis/24-megawatts-of-mining), but the model was essentially correct (I now have a much more accurate model). The hash rate expansion is pretty-much funded as a large fraction of the USD-denominated Bitcoin mining reward. This has dropped off significantly recently because the hash rate has slowed down (so we're finding nearer 144 blocks per day instead of > 160) and because of the BTC:USD price.

At a given BTC price there's a steady state where most of the money ends up going to hosting, maintenance, electricity costs - essentially all opex. That shifts when new hardware comes online that dramatically reduces operating costs and causes older equipment to be taken offline. 16 nm ASICs are probably the next such technology shift so I wouldn't expect to see much change until then if the BTC price remains constant.

The model predicted that when we hit a steady state where hardware margins had been tightly squeezed then we'd see about a 2x per year increase in hash rate. If we take that 2x per year then that works out at a mean difficulty increase of 2.6% each time for a year. Given the noise in the statistics for hash rate estimates then that would typically mean seeing difficulty changes of between -2.4% and 7.6% about 90% of the time. Now of course the real world won't work quite that predictably but the trend should be reasonably consistent.

What the model can't tell you is what will happen to the BTC price. If it jumps 2x then things will go mad for a few months until the hash rate catches up - in fact it would more than double; older, less power efficient hardware would be taken offline faster as people sought to use their available power more efficiently, but then new power and hashing would come online too. If the prices drops 2x then new investments cease and older equipment goes offline a little quicker but most of the hash rate is probably now more power efficient and so the hash rate wouldn't halve.