Yes, the block subsidy is just a subsidy for transaction costs. By way of inflationary effects, the entire network is splitting the current costs of mining.
So if blockrewards are a subsidy for transaction costs, isn't it true that that same subsidy is whats causing the transaction/mining costs to be so high in the first place? That seems like a vicious circle to me? Even if the subsidy was doubled, the hashrate would double too, so the profitability would stay the same for miners, and all it does is increasing the tax (average transaction cost) for end-users even more. So that subsidy seems to be completely counter-productive, except for making sure the hashrate stays above an artifical baseline, whose level seems like overkill these days.
Is there any proof or logic that predicts the baseline that would be achieved by rewarding miners only with fees, would be too low for a secure network? Im just trying to wrap my around it.
Logical, self-serving miners who pay for their own electricity/cooling would take the strategy of "when there's over x fees, it's worth it for me to mine" and have their machines automatically start and stop accordingly. This will drastically change the "block every 10 minutes" scheme we've got now.
You may just have answered my question above. Indeed, this may be the logic behind behind the 'fixed' reward every block, to reduce miner variance.