Quote from: ArticMine on July 16, 2016, 01:18:48 AM

( snip )
There are two very distinct scenarios I see for setting fees in Monero.

The first scenario is when the median of the blocksize over the last N blocks, M_N < effective median of the blocksize over the last N blocks, M₀ = 60000 bytes. This is the current scenario and not the scenario I discuss above, since M_N is replaced above by M₀ eliminating the penalty for a blocksize increase. The current M_N is less than 300 bytes.

The second scenario is where M_N > M₀ and the penalty applies. In this scenario per KB fees are proportional to the base reward divided by the median of the blocksize over the last N blocks, R_base/M_N. The objective here is to stimulate an efficient market between blocksize scaling and fees. One way to achieve this is to set 5 fee levels corresponding to transaction confirmation priorities. minimal, low, normal, high, very high. The per KB fees are calculated using the known values of the block reward, and, the median of the block size. One places minimal and low below the penalty boundary, normal at the penalty boundary and high and very high above the penalty boundary.

The prior discussion in this thread can in reality only be possibly applied to the first scenario for the minimal fee and the low fee. The rest of the fees would have to be set as above but using M₀ rather than M_N.

Edit: One important consideration is that there is a 200x increase in transaction volume and possible corresponding increase in price before we trigger the second scenario, so we will have to scale down the fees with no trigger of the penalty. It is here where my original idea as edited and modified above may be a possibility to set the low minimal and low fees.

I'm confused, though I think I'm putting it together. I see you're threading the needle of making the fee adaptive while giving the fee the ability to serve its intended purpose of preventing blocksize expansion. I think you were on to something before with using the difficulty as an on-chain surrogate of external value. I think the need for that factor will exist at any stage of the chain's life - during initial distribution curve and during the tail emission. Your second scenario above seems focused on the tail emission portion of the coins existence, which doesn't happen for another ... however many years.