Quote from: DeathAndTaxes on November 27, 2013, 12:15:06 PM

On the efficiency aspect it is nothing compared to the prior difficulty increases.

A GPU is something like 300 J/GH. The "worst" ASIC is ~8 J/GH so you are talking almost a 40,000% improvement. Various different smaller process node designs dropped from the 8 J/GH to 5 J/GH to ~0.8 J/GH (Bitfury - reference clock)).

On paper we have the rest of the 28nm players showing ~0.8 J/GH and KNC estimating ~0.7 J/GH. Seeing a trend. Overnight we went from 300 J/GH to 8 J/GH and then over the next six months saw that fall by a factor of 10 to ~0.8 J/GH. That "might" go down to a staggering 0.7 J/GH by next Spring.

Quote

My point is that the hashrate will continue going up exponentially for an unexpectedly long amount of time.

Who is going to buy all that hashrate increases. $3 per GH works out to $3M per PH. So to double the network again in a month would "only" require $15M in sales. Ok no problem there. To double it again would require another $30M in sales the following month, $60M the next month, $120M the monh after that, a quarter billion the month after that ...

Will difficulty keep going up? Oh HELL YEAH! Is it possible from either a cashflow or energy standpoint to double continually every month for years? No.

No argument the rate of increase is unsustainable. Of course, $/GHs and J/GH will decrease. I haven't run the numbers but going back to CPU/GPU days it's pretty obvious that constant investment will still result in significant increase in difficulty. Mining ROI is painful.