> Fee estimation algorithms being ‘hard’ is sorta a scapegoat, since they don’t incur a heavy time cost (only once every 10 / 2.5 minutes, only for the miner, for all miners)
Fee estimation algorithms are client-side, not miner-side. So all wallet implementations need to have them, and if/once proprietary implementations start getting developed that are reliably better, that’s a possible centralization vector.
> The network never has any ‘costs’
Ah, by «costs» I was referring to costs incurred by the participants and beneficiaries of the network, including cost of processing, cost of storage, and marginal increments to systemic risks involving network centralization. So not the limit itself, but the factors that led to the limit being instituted in the first place. The issue is that the risks don’t actually increase much if the blockchain sometimes has 7m gas per block and sometimes 9m, versus always having 8m gas per block, and a flexible schedule that allows usage to drop to 7m when fees are low and rise to 9m when fees are high make the blockchain more usable by allowing more people to participate when demand is highest.
> it could cause a contentious fork / more trouble than good … the improvements to the social economics look worthwhile, but probably only work in practice on a new blockchain
Oh sure. If the Zcash community values stability of protocol, or not entering into weird unknown unknowns, enough that this is too contentious, then yes that’s a reason not to adopt it. That said, from my observations I don’t feel like the Zcash community is _that_ conservative; it’s willing to adopt radical changes to its underlying cryptography with a less than year-long development cycle, use weird spooky crypto with trusted setups, etc. I feel that the one thing that Zcash should definitely absolutely not take risks on is its core value proposition, which is giving people privacy. Possibly the 21m limit as well if that’s the community’s wish. I think transaction fee economics is more like a technical feature that can and should be optimized, and not some fundamental immutable feature that must be safeguarded in its current form at all costs. That said, it’s ultimately up to Zcash users to decide this.
(a moderate third alternative is of course to sit for two years and let ethereum be the guinea pig for changes to transaction fee economics first)