I hadn't read your earlier post on the topic. That was an enlightening read :o)
As far as I understand PAL it seems to have a discrete curation and reward function. This formulation of the marginal reward relies on the smoothness of the curation function which is not satisfied.
You mean decimal precision? That's coming. It doesn't really detract from these points though. It just means a lot of truncation is happening (and anyway truncation has to happen, but at an appropriate scale... Right now the scale is not right, which is why they are adding decimals back)
But all these functions are smooth in the relevant domain. What specifically do you mean?
The functions you present are smooth so that is fine. The scale at which PAL is now suggests that it should be modelled discretely. But since they will change that (which I didn't know :P ) implies that your points are valid for the new system :o)
I would say a discrete model would anyway need this continuous approximation. It may be less accurate, yes, but I believe even the computations behind the scene are doing the same computations and truncating right before payout assignment. But yeah, because of that, for example, the minimum payout per rshare is basically 0 with current parameters, so I guess that is a very relevant point. (Unless they have a large amount of rshares to assign, which I think needs to be like 10k/100k ish)
Posted using Partiko Android