I like that we're moving forward in the discussion, it seems. The original compromise that was the reason for me to start the thread now looks a bit dated.

I would support Lauda's maximum cap idea, as it's true that there could be circumstances where such a flexible system could be gamed.

The challenge is now to find a number for this cap. I had done some very rogue calculations that a 1 TB/year blockchain (that would be equivalent to approximately 20 MB blocks) would enable 160 million people to do about 1-3 transactions (depending on the TX size) per month. That would be just enough for this user base if we assume that Lightning Network and similar systems can manage smaller pauments. 1 TB/year seems pretty high, but I think it's manageable in the near future (~5 years from now).

Obviously if we want the 7 billion people on earth to be able to use Bitcoin on-chain the limit would be much higher, but I think even the most extreme BU advocates don't see that as a goal.