prevent exploiting adjustment with value/time adjustment on upwards adjustment

[ghost]

One element of the game theory of the 9 way parallel not considered before is that the adjustment of big blocks is proportional and thus the approximate amount of them you can get is equal, given a current hashrate and a given amount of hashpower. So with this adjustment instead the bigger the reward the more reactive it is to attempts to jump on it.

It may not need to be linear to the ratio, possibly some other derivative ratio will work ok without being too heavy handed. Grabbing one is no big deal but getting several of the top ones in a row is not something that is fair to the other players.

The adjustment only increases upwards adjustment, downwards adjustment is the same. If a sharp upwards really should be followed by a sharp downward as the differential is small that's an unfortunate cost - the big blocks might thus tend to lag a little behind the small ones. So perhaps a half-way coefficient that sharpens down and up adjustment on the big blocks. Otherwise it would definitely tend to time dilate as there is always someone who tries the trick that gets one small win again. This way the next block after would rapidly move it back down.

Also, another thought, since all hash functions are really just one and some blocks need more than others, there can be a hashrate computed out of this that is smoother than even a round robin would produce. Also shouldn't the order be shuffled every cycle of the list? probably already is.

Anyway, it's essentially a resonance filter, blocking out one easy opportunity to lever the system.

[l0k18]

Not sure about this one. I think there is something in the algorithm that already escalates proportionally. We must check this before scratching it, but later.