Soon BTG wil go to a new DA. It just so happens that BTC Candy copied BTG’s difficulty algorithm (which is contains an error that was my fault). They have also copied and implemented BTG’s new DA which is my LWMA that is being used on a lot of Monero / forknote / Cryptonote clones. So, we might be able to get an idea of the improvement we can expect by looking at its before and after data. One caveat: BTC Candy is 120 minute solvetime which gives them an advantage in difficulty. But their difficulty performance is a lot worse because they are smaller and get much bigger hash attacks.
Simple moving average N=30 with unfortunate accidental limits left over from Digishield’s code:
(the title says “Digishield” but it’s not similar to Digishield because I made a major change to it)
New LWMA algorithm: (first plot starts with prevous algo above)
For comparison, here’s BTG’s current algorithm performance:
All the numbers above are pretty bad. LWMA typically sees < 5% and < 1% on the “blocks stolen” and “delays” metrics.
I have been able to greatly improve the LWMA by making it “dynamic” which means if last 5 blocks were solved in < 600 seconds, then it switches to an LWMA with 1/2 to 1/3 the averaging window to make it respond faster. Only the pseudo code is done and it maye take us a few days to get the code done. Sicne it’s not been tested in a live coin, it’s premature to try to get BTG to use it.