A non-scarce good is one that does not require allocation because the available supply at a price of zero exceeds the maximum achievable demand at that price.
No good that requires time or energy to deliver can be non-scarce.
Including transactions in a block will always require both time and energy, therefore the space in a block will be scarce.
Because space in a block is scarce, miners will need to allocate the inclusion of transactions into a block, and there exists a price below which they will not do so.
We can't calculate ahead of time what the equilibrium price of a transaction will be in the future, because that depends on the future actions and preferences of millions of other people.
This is a bit unhinged. The _inherent_ costs of transactions is roughly size_of_data * decenteralization_level (actually there is a quadratic component in a decentralized network too, but lets ignore that; good design can make it small). In a free market for transaction capacity based purely on the inherent cost optimal competition can drive decentralization down to lower costs. With a completely centralized system the cost of almost any imaginable scale is basically nothing (e.g. a single <$2000 host on a sub-gigabit network connection is able to process a hundred thousand transactions per second).
So effectively one can replace the fee market with a market that favors the most centralization as they have the lowest costs (as thats all network income would pay for). This may be true, but it's not interesting-- since if a highly centralized system were desirable there are more efficient and secure ways to achieve one.
I believe you're making a false comparison. None of the market participants have a way to express their preference for a decentralized network except by defining Bitcoin to be one though the rules of the system. Absent that someone who doesn't care and just wants to maximize their short term income can turn the decentralization knob all the way down (as we've seen with the enormous amount of centralization in mining pools) and maximize their income-- regardless of what the owners of bitcoins or the people making the transactions prefer. You could just as well argue that miners should be able to freely print more Bitcoins without limit and magically, if the invisible-pink-hand decides it doesn't want bitcoin to inflate, "the market" will somehow prevent it (in a way that doesn't involve just defining it out of the system).
Of course, 28 minutes is still long. That is based on 2013 data.
This data is massively outdated... it's before signature caching and ultra-prune, each were easily an order of magnitude (or two) improvements in the transaction dependent parts of propagation delay. It's also prior to block relay network, not to mention the further optimizations proposed but not written yet.
I don't actually think hosts are faster, actually I'd take a bet that they were slower on average, since performance improvements have made it possible to run nodes on smaller hosts than were viable before (e.g. crazy people with Bitcoind on rpi). But we've had software improvements which massively eclipsed anything you would have gotten from hardware improvements. Repeating that level of software improvement is likely impossible, though there is still some room to improve.
There are risks around massively increasing orphan rates in the short term with larger blocks (though far far lower than what those numbers suggest), indeed... thats one of the unaddressed things in current larger block advocacy, though block relay network (and the possibility of efficient set reconciliation) more or less shows that the issues there are not very fundamental though maybe practically important.
