Scarcity in terms of space in blocks for transactions to be confirmed is necessary to cultivate a healthy transaction fee market.
This argument presumes the value of the transactions will justify exorbitant fees. That will end Bitcoin as a currency. I can see their point, but it's premature. Bitcoin isn't ready to replace gold as a store of value, but it's already losing power in confirming transactions. Bitcoin should work towards staying relevant as a currency until the market deems it's better as something else. While I agree with Justus that the market can help find the best equilibrium between fees and block sizes, I think it's best to allow network science to determine that ratio rather than market exploiters.
So you prefer central planning over selfregulation and free market when it comes to transactionfees?
You either have a transaction rate which induces transaction fees or a transaction rate which does not. Either option is a 'central plan' just as was the 21x10^6 maximum circulation.
Fact is that it doesn't matter what the transaction fees are or the coinbase reward is. If you don't factor in alternative monetization Bitcoin is economically broken at the fundamental level with respect to support reward for POW. This because it will always approach net zero no matter what. It will just do it faster when the volatility increases the price and mining is more profitable (then fall below zero when the price reverses.) The situation is aggravated because Bitcoin is horribly inefficient making it very non-competitive as and exchange currency for buying skittles. That's why the 1MB block size has been fine and looks like it will be for a while yet.
So, let's consider alternative monetization streams, shall we? They come in two basic flavors:
1) Gathering user intelligence data. Very high value in the case of a monetary system. Most currently gathered data is just a proxy for who spends what where while transactions data is the real McCoy. Google-class players play this game and are much better at it than anyone reading this will ever be. Uncorking the transaction rate will merely ensure that they have no real competition (though they may choose to isolate themselves by buying the data from a third-party anyway.)
2) Subsidization from entities who need Bitcoin to be autonomous, reliable, and trusted. That would be entities running sidechains.