[Idea Review]: Would Batch balancer apply to batched SnapDeal proofs?

[lucaniz]

Request Summary

We are in the process of evaluating the possibility batch SnapDeal proofs (i.e. SnapDeal proofs for different single sectors) via SnarkPack (similarly to what happens for ProveCommit today).

We went through the "What is the Batch Balancer supposed to be doing?" section of this doc

where it says

The ability to submit sector proofs as a more efficient batch was introduced in FIP 13. If all else stayed equal, such an improvement would reduce the demand for gas and base fee. The batch balancer mechanism imposes an additional BatchGasCharge to be paid whenever a set of proofs is batched. The consequence of this additional charge is that when the base fee is low, it becomes irrational to batch proofs, as it would be cheaper to send the proofs individually. This is a measure that protects against demand for gas lowing to very low levels.

and it would be nice to know is this would apply to batched SnapDeal proofs as well, or if BatchBalancer is meant to be restricted to ProveCommit only

cc @irenegia @nicola @cryptonemo

Audience / Consumer for this Review

Core Developers

Timeframe

Needed in 1 month

Rationale

It should be an easy answer on your side which does not require modeling/analysis (it is basically a yes/no question). Let us know if this is not the case.

Desired Deliverables

A yes/no answer on the question above

Additional Information

SnapDeal project

https://www.notion.so/pl-strflt/SnapDeals-Project-8e85ed1678e044d7bb514ec7038814c6

Contact Details (if Private Response Requested)

No response

[AxCortesCubero]

This is not really "technical" question in the sense that we would do some analysis to find what the "right answer" is, but more of a Business question, of what we should do as the Filecoin network. In a sense we should collectively decide if we will apply batch balancer logic to a new type of batching mechanism that is being introduced.

My opinion (and I emphasize given the preface above, this is just about opinions), is that is that the same rationale used to implement the batch balancer for PreCommit and ProveCommit, should apply here, there doesn't seem to be anything fundamentally different about this proposed batching mechanism, that would make the batching logic not apply. So my opinion is if we agree batch balancer mechanism should apply to PreCommit and ProveCommit, then it should be made to apply to SnapDeal proof (up to choosing appropriate new parameters). Unless you have some different feedback, or reason to believe there is something inherently different about the batching mechanism being proposed?