Suppose we granted that punishing certain acts is an intrinsically good thing.
If the above it were true, it would seem hard to deny that punished crimes create less net-badness than otherwise identical unpunished crimes. This is because they are associated with an intrinsic good (punishment) that unpunished crimes are not.
Presumably, if there are two crimes, and one will create less net-badness than the other, we have a reason -a defeasible reason but a reason nonetheless- to prioritise preventing the crime that will create more net-badness.
But this means that we have a reason to be less concerned about stopping crimes that we know will be punished. But this is absurd, ergo we should reject the view that punishment is intrinsically good.
Keep in mind that we’re talking about intrinsic badness here. If we were arguing that it’s more important to stop crimes when we wouldn’t be able to catch the perpetrator afterwards because doing of, say, the importance of deterrence, that would be very plausible. But anything which suggests the mere fact of punishment makes it less of a priority to stop a wrong seems repugnant to me