The brick wall thing seems like a particularly silly analogy. People are already competing for block space with fees. Why are people doing that when block space is effectively infinite? Why are miners encouraging that when the fees are so meaningless in comparison to the subsidy? It seems like the car hit a wall of jello a couple of miles before hitting the brick wall.
The brick wall is analogy is accurate because It's the difference between being able to send transactions and not being able to send transactions. Right now all the users who want to send Bitcoins can do so because the total demand is lower than the protocol limit.
Once the blocks fill up we'll get to a situation that changes. If more than 5000 people want to send a transaction in the same 10 minute interval they won't all be allowed to do so, no matter how much they are all willing to pay in fees. The ability to use Bitcoin will be rationed to a fixed number of entities.
What I want to know is, how much cheaper? How well will people innovate? Are people willing to pay enough in fees to keep mining going as the subsidy declines? What is the actual marginal cost of including a transaction? How will that change over time?
There are a lot of question marks between where we are and where we want to be. It seems imprudent to assume that they will all be resolved to our liking.
It's complete economic and historical ignorance to assume the best way to answer any of those questions is with central planning in the form of transaction rationing.
Miners will mine if they can obtain a market price for their services that is acceptable to them. Users will initiate transactions if the price if sending transactions is acceptable to them. Not only is there no need to second guess price discovery by trying to calculate the marginal cost of handling a transactions and using that as the basis of a magic number in the protocol but also succeeding at such a strategy is
probably impossible.
I'm not sure why you think that there is some sort of free market for bitcoin transactions. I have a lot of confidence that someday there will be such a thing, but for the next couple of decades (and possibly for the rest of my life) there is not and will not be.
Right now, every single transaction (with a couple of exceptions caused mostly by programming errors) is mined as a charity. There is also some indirect benefit to the miners, in that the network actually working makes their holdings more valuable, but there is nothing even slightly resembling a balance between cost and reward.
To recap, we have no idea what it costs to process a transaction. Even without the block size limit, there is a limit to how many transactions can be processed by the network in a given time. There is a much smaller limit to how many can be processed economically. And there is a vastly smaller limit to the number that could be processed economically without the subsidy. We have no idea where these limits are, and they are all moving targets.
If we were starting at 100% adoption, Moore's law would let computing power keep ahead of the game, and probably even pull out far ahead. But we aren't at the top of the adoption curve, we are at the bottom. In the short run, the vertical slope looming is going to kick all of our asses. There will be a period when the system won't be able to keep up with demand for reasons other than an artificial block size limit. God help us all if we've evicted all of the small miners (less than, say, 100 million readily available capital) from the network before that happens.
I'm really not sure why I bothered to write this reply. Your want the network (people other than you, at least mostly) to have to provide a scarce resource in unlimited quantity, while you accuse me of advocating central planning for pointing out that the market that you need doesn't
actually exist.