![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 03:02:03 AM |
|
Reading tons, absorbing lots, and learning gradually...
Question: has anyone considered implementing the difficulty increase as a feedback control system?
I understand block discovery is targeted for 10 minutes on average. I also understand we're seeing 8-9 minutes block time due to increasing global hash rate, putting us ahead of schedule on minting. Adding some kind of integral term or higher-order model control might be better suited to keeping it near 10 minutes.
e.g. modelling mining as a 1st order plant with desired block rate as the input, an integral term on the discovery rate error (i.e. difference between average block time and 10 minutes) might serve to drive it closer to 10 minutes. In fact this would suffice to drive the error to zero given a linear climb in hash rate, though an exponential rise would still see some steady state error, possibly unbounded.
Note: I'm not sure whether it's already using feedback like this, or whether we even care about keeping the schedule - more reading required for me.
PS: we may be seeing too-rapid inflation if the amount of money being produced is not actually required at this time by the surrounding Bitcoin economy. Can someone point me to discussion of how things are going in that sense, and whether anything could or should be done to correct it?
Cheers
|