I’d say this is a tool among many others, and as I argue more in the other post on SSF, this is happening anyways: history expiry, zkEVM/statelessness, DAS all converge to making it extremely light to run a validating node. Designing mechanisms such as Light FOCIL is another way to tilt the power more towards the many than the few, and create groups with possibly competing enough incentives to obtain checks and balances, essentially.
That’s the defensive part, but I also think our focus can and should be on the attack. Scale matters to provide as much welfare as possible to our users, so spending time on trying to get as much juice out of our system is also a very worthy use of our time. Fortunately, the two are not in conflict in many places, e.g., zkEVMs make it extremely easy to verify (so high robustness) as well as providing huge scaling benefits. There is still a trade-off between maxing out gas and proving time, relying on fewer parties able to prove this, but at least the efficiency frontier is pushed way forward thanks to this new tech.
Where sensible and where we have more of these 1-of-N quality of service guarantees (i.e., we only need one party out there to be live and honest to give us good service), there is the opportunity. But it’s not a blank cheque, as detailed in the post above, and it still requires careful calibration. And yeah, this post by @Julian also discusses the question for state management.
My view is additive, we should first communicate as clearly as possible the opportunities and risks, gather more understanding and allies and critiques of these ideas, and I don’t imagine that a discrete one-off decision is made saying “ok now we pump it”, but I expect when the topic at hand is “how far can we go with blobs” (for instance), hopefully some of the discussion we had on higher-level principles filters through and is actionable enough to guide the current decision.
I’ll also link to this slide of the protocol research call #1, hoping to use these channels to keep the conversation going.