Ruth Millikan (for instance) is right that given the nature of design constraints, it is unlikely in the extreme that there could be different ways of skinning the cat that left two radically different, globally indeterminate, tied-for-first-place interpretations. Indeterminacy of radical translation is truly negligible in practice. Still, the principle survives. The reason we don’t have indeterminacy of radical translation is not because, as a matter of metaphysical fact, there are “real meanings” in there, in the head (what Quine called the “museum myth” of meaning, his chief target). The reason we don’t have indeterminacy in the actual world is that with so many independent constraints to satisfy, the cryptographer’s maxim assures us that it is a vanishingly small worry. When indeterminacy threatens in the real world, it is always just more “behavioral” or “dispositional” facts —more of the same—that save the day for a determinate reading, not some mysterious “causal power” or “intrinsic semanticity.” Intentional interpretation almost always arrives in the limit at a single interpretation, but in the imaginable catastrophic case in which dual interpretations survived all tests, there would be no deeper facts to settle which was “right.” Facts do settle interpretations, but it is always “shallow” facts that do the job.
Philosopher Daniel Dennett's Book Intuition Pumps