It is not at all self-evident that excluding larger data blobs from a scarce data space is not more lucrative. Strictly because of the variable phi input, you can't just draw the conclusion that more equals more. I literally gave you an example of a miner making more money by excluding inscriptions and large data pushes.

Private in this context means hidden from public mempool gossip, not your definition of private property or whatever.

And my burger analogy, although a little cumbersome for the application, is still apt. Because no matter what someone has paid previously or will pay in the future, what I determine to be the fee can't be estimated. Example: if the previous block everyone paid a thousand sats per v-byte, that doesn't mean anything compared to what I'm willing to pay for my next transaction. Fee estimation is Keynesian psuedoscience.