Last week Jody Azzouni was here to give a pair of talks: one about scientific theories, another about his view that English is inconsistent in a pretty radical way: Every sentence is both true and false. They were both a lot of fun, and Jody is a great interlocutor -- he kept both presentations relatively short and to the point to there'd be more time for questions and clarifications. I also have a soft spot for arguments defending unpopular ideas -- though I usually side with the orthodoxy, incredible ideas are often a bit more interesting to think about.
In the philosophy of science talk, Jody was building on his work on what he calls "thick epistemic access." His argument was that we should not (contra Quinean orthodoxy) have existential committment to all the posits of our current best scientific theory, but rather only those posits to which we have thick epistemic access. (See e.g. Kenny's post here for a quick but accurate description of thick v. thin v. ultrathin posits.)
I was wondering, however, whether the Quinean orthodoxy could be undermined in a more direct way that does not involve developing a whole epistemological apparatus to distinguish when we really do have strong evidence that such-and-such thing exists. (Such a question is certainly philosophically interesting and worthwhile, but it is likely to be complex and contentious in places.) Rather, I thought a simpler argument against the Quinean orthodoxy could go as follows:
Science is rife with idealizations -- some of which are ineliminable/ indispensible. But no one should be committed to such idealizations, since they are (almost by definition) deliberate and conscious falsifications in our theoretical account of the world. So existential commitment does not follow our best theories as well as Quine would like.
I realize that (1) there may sometimes be a legitimate question about whether a given bit of a theory is an idealization or not, but that just shows the term 'idealization' is vague -- all parties agree there is some idealization in science, even if they don't agree on every case. Also, (2) most examples of idealizations are not entities, but rather inaccurate properties (e.g., treating some body that we know to exist, like a point particle: we give an inaccurate description of the thing's dimensions). So maybe pointing out the widespread use of idealization will not create widespread problems for the Quinean orthodoxy.