I think the point Max makes here is pretty solid (albeit he's selling his product!), but I find your point much less clear. What's the relationship between anonymity set being clear, and inefficiency?
It's one thing to not consider the anon set easily calculable, and still think the system is viable as an anonymizing one, but if anything it's dangerous to claim it's better if you can't quantify it.
Probably you mean something quite different.
Whirlpool txs are quite small, with consistent sizes, so it's easy to calculate the anonymity set with a dead simple algorithm.
With Wasabi's huge txs with many different values those simple algorithms aren't computationally feasible because there's so many possible ways the inputs could lead to the outputs. Meanwhile I've seen Samourai fans claims this is a bad thing, because you can't use some external tools to easily figure out the possible input/output anonymity sets.
The Wasabi wallet of course has a good idea of what the anonymity set could be, as it knows what outputs it created. So it can easily calculate a conservative k-anonymity set.
OK that, yes. You're really saying that a larger anonset in one coinjoin is better despite being less objectively calculable. I agree (though, the counterargument isn't 100% wrong).
