First, credit where it’s due. There’s serious work behind this EIP and disagreement shouldn’t obscure it. We want the same thing for Ethereum. My aim isn’t to be right — I’d be glad to change my mind.
On issuance buying security: I think this is precisely where the visions diverge. I’ve written several pieces on it, because the framing runs into contradictions the moment you do an honest financial review. It’s inherited from proof of work and carried into proof of stake without one.
Under PoW there’s a real input being bought. Hardware, power, depreciation. The budget is consumed, and as in any competitive sector providers push toward the lowest cost while preserving a thin margin to stay viable. That’s a genuine procurement market, and the rational thing for a protocol is to buy security as cheaply as it can.
Under PoS the infrastructure cost is marginal — a minimum setup per instance and little else. Measured as procurement, the margin makes no sense. A single provider could serve the entire network and collect billions at what’s paid today.
The reply will be that what’s bought isn’t the machine, it’s the bonded capital. Take that seriously and it leads somewhere else. You can’t be a provider without being a shareholder, and what you receive scales with your holding, not with anything you produce. A validator’s P&L depends far more on the price of the share than on the 2% it collects. That’s a dividend — one conditional on turning up to validate, closer to attending the shareholders’ meeting than to invoicing a client.
The market has in fact already separated the two prices. What an operator charges to run validation is the commission on rewards — competed down, observable, a fraction of the total. That’s the price of the service. Everything above it is a return on capital. And LST operators validate without owning the stake, so the separation isn’t theoretical.
Which is why the procurement framing collapses on its own optimum: buy security at the lowest cost and you converge on one provider. Nobody in this thread wants that, least of all anyone six tweets deep into concentration risk. When we all reject a model’s efficient outcome, none of us actually holds the model.
What’s left as a defence is that the rate isn’t compensating realised risk at all — it’s the price needed to attract enough capital that an attack is expensive. I’d accept that framing entirely. But a price paid to bring capital in and keep it committed is an interest rate, not an invoice.
Because underneath, I think staking is meant to do something simpler: oblige the shareholder to make the small effort of contributing security to the network. The reward isn’t what the network pays for a service. It’s that whoever doesn’t bother gets diluted. Delegation satisfies the letter of it — and it’s fair to say that delegated stake doesn’t contribute in the optimal form.
Which is my problem with the taper. It removes the motive for the effort while keeping the mechanism, and the participation that survives a falling rate is the one where the effort is nearly free at scale. That’s delegation to the largest operators — the least optimal version of the thing, and the outcome you’ve spent this thread warning about.