Hey will - do you think it would be useful to make public a cost / revenue model that people could contribute to given existing usage and parameters (e.g. no of validators). You've started something similar here.

Dumb question here, but is there an evolution of this 'performance' model for collators where there is usage based income, rather than just producing empty blocks?

Is this another case of paying for uptime, but not necessarily for demand?

Curious as to your views...

Hey Rich,

The baseline we’re using comes from the approved RFC-0007 (System Collator Selection), which was informed by broader stakeholder feedback and includes input from SR Labs auditors. (paritytech.github.io)

RFC-0007 suggests a rule-of-thumb target of ~20 collators per system chain (often described as a split of ~15 invulnerables + ~5 bond-elected/permissionless slots) to balance liveness, censorship resistance, and operational scalability. (paritytech.github.io)

In practice, we have been operating below these thresholds across the system chains. However, given the increased operational importance of AssetHub, we have moved to tighten the gap by increasing its collator set to 18 total (invulnerable + permissionless), while still remaining under the RFC baseline. (paritytech.github.io)

The question of the “ideal” number of collators comes up regularly, and it can quickly become subjective. For system chains, the safest anchor is the research and previously approved direction that prioritizes liveness and censorship resistance as non-negotiable requirements. I’m open to adjusting collator counts downward, but any proposal to reduce them should be backed by clear, testable evidence that these guarantees are preserved at lower numbers—along with a concrete operational plan showing how the network would detect, respond to, and recover from degraded performance or coordinated disruption.

Best,
Will