How about as a first test measure "FRR = (average of all current swaps/85)*100"? This would mean that the 15% fee of Bitfinex is added statically on top of the FRR, as currently if you lend at FRR, you receive only 85% of the average rate in reality.
This would mean the FRR rises the more it gets used and no fancy programming would need to happen (unlike "FRR +/- X%" scenarios, that can also end up cancelling each other out).
Are you proposing including the FRR swaps in the calculation and not just the fixed rate swaps?
I don't see how your calculation will raise the rate due to more FFR swaps being taken. I think the FRR would get a one time boost when the new formula is implemented and then will proceed to drop just as it does now.
Taken as written, without including FRR swaps in the average (setting a variable rate to the average of a bucket of rates including itself just seems too self-referential to work), it would at least move the wall up to "slightly above average", allowing swaps in the slender middle-ground between "average" and "FRR" to be taken, and hence allowing the FRR to move upward
without the wall first being eaten.
It's not quite what I had in mind but I'm not sure how my thought could be implemented mathematically - I'd want the wall to move upwards as it gets consumed, regardless of what fixed-rate swaps are doing.