On February 5, 2023 1:11:35 AM GMT+01:00, Russell O'Connor via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
>Since bytes in the witness are cheaper than bytes in the script pubkey,
>there is a crossover point in data size where it will simply be cheaper to
>use witness data.  Where that crossover point is depends on the finer
>details of the overhead of the two methods, but you could make some
>reasonable assumptions.  Such a calculation could form the basis of a
>reasonable OP_RETURN proposal.  I don't know if it would be persuasive, but
>it would at least be coherent.

I don't think it's worth the technical complexity trying to carefully argue a specific limit. Let users decide for themselves how they want to use OpReturn.