From the disunity of science to anti-realism
The two most popular arguments against scientific realism today are (1) underdetermination arguments and (2) the pessimistic induction. There is another kind of anti-realistic argument besides these two, which does not get the attention today that (1) and (2) receive. (Note: I would appreciate it if any readers could point me to someone working today who has developed anything akin to this third line of argument in detail.)
I don't yet have a clean formulation of this third kind of argument for anti-realism that I find satisfactory. All I have so far is an analogy. We see an instance of this third kind of anti-realist argument in Andreas Osiander's (much-maligned) Preface to Copernicus' On the Revolution of the Heavenly Spheres. Osiander argues that Ptolemy's astronomy is not intended to be the literal truth about the structure of the solar system on the grounds that, if the epicycles Ptolemy ascribed to Venus literally described Venus's movement, then Venus's apparent size for an Earthly observer should vary 16-fold. But Venus's apparent size does not vary in this way. Thus, even though Ptolemy's astronomy predicts planetary positions within the margin of observable error, if it were the literal truth it would also get apparent planetary size right as well.
What does this have to do with present-day scientific anti-realism? This is sketchy, but perhaps the increasing sympathy for 'the disunity of science' and the related notion of 'pluralism in science' could be used to argue for some sort of anti-realism. Now, what exactly those two expressions mean varies greatly from one person to another (I think; I would be happy for better-informed readers to enlighten me), but I think I can illustrate my point by imagining what a 16th Century pluralist/ 'disunifier' would say about Osiander's point above.
Such a person could say: Different scientific theories explain different phenomena. Ptolemy's theory only aims to account for apparent planetary positions, not their apparent sizes (and/or brightnesses). To demand that Ptolemy's theory also account for apparent sizes/ brightnesses is to impose an ideal of unified science upon Ptolemaic astronomy.
However, we today (and Osiander 450 years ago) think that this problem with apparent sizes is good evidence against the truth of Ptolemaic astronomy. That is, the 16th century pluralist/ disunifier seems to 'let a bad theory off the hook': we have evidence the theory is not true, but that evidence is discounted or ignored on the grounds that the theory's target domain of explanation does not include the countervailing evidence.
Now of course the question for modern-day antirealism is: how good is this analogy? In particular, is the kind of pluralism or disunity that people are championing (or at least accepting) today relevantly similar to the Ptolemaic pluralism/ disunity described above?