Re:Timelock: time-release encryption incentivised by Bitcoins
by shasta on Wednesday June 04, 2014 @09:05PM
(#7849568)
Why is this based on a symmetric problem (encoder solving the same problem as the decoder) requiring two days of crunching on a 16 core machine? Is it because all known asymmetric problems (factoring primes, etc.) allow large speed-ups with parallelism?
[ Parent
| on HN ]
Re:Timelock: time-release encryption incentivised by Bitcoins
by mrb on Wednesday June 04, 2014 @09:32PM
(#7849659)
Yes, because asymmetric problems can be parallelized.
Re:Timelock: time-release encryption incentivised by Bitcoins
by stromgo on Thursday June 05, 2014 @01:45PM
(#7853134)
It's not so clear.
Here, as long as you can afford a near-optimal SHA256 processor, more computing power doesn't help.
In the "successive squaring" method (see the essay posted by gwern), as long as you can afford a near-optimal modular multiplier, more computing power doesn't help.
In both approaches there is a constant cost of the near-optimal hardware, and a confidence level that the hardware is actually near-optimal. In SHA256 the cost is probably lower and the confidence level probably higher, but the difference is quantitative, and not qualitative.