A segwitv0 output is 31 vbytes, but a witness input is around 68. That's 99 vb, multiplied by 37.5 = 3712.5. Yes, it is less than I wrongly calculated, but it is still waste of money for such a small output.
What do you think of the dust problem in Whirlpool?
I'm good at quoting old posts as well:
As has been explained to Kruw dozens of times, the change output from Tx0s are sent to a separate account and deliberately segregated from your other UTXOs. There is no way to accidentally include them in a transaction. Any user consolidating their change output as has been done in the transaction he has linked to above is doing so deliberately. I understand that Kruw gets angry when people spend their bitcoin in ways that he personally doesn't approve of, but there is no bug here, just Kruw either being deliberately misleading or simply not understanding what is happening.
Using a different derivation path didn't stop these Whirlpool addresses from being linked together.
Furthermore, I showed the proof of how I linked postmix Whirlpool outputs to the premix transaction that generated it, which no one ever addressed at all:
B-b-but, I thought this ethos was unacceptable.

I guess I can't argue against someone who takes the stance "Wasting your Bitcoin and ruining your privacy should be allowed."
I know, I know... Just another confusion in the name of privacy!

Okay, since you seem to think there's a problem, what is the solution to the problem you are describing? How should the coinjoin protocol be changed?