Professor of pseudoscience philosophy Massimo Pigliucci summarizes, but I should warn you that he is one of those leftist ideologue anti-sciece reality-denying philosophers.
Here is his latest rant, on the denial of human races:
Realism, antirealism and conventionalism are technical philosophical terms usually deployed in discussions of philosophy of science, philosophy of mathematics, and ethics. Say we are talking about the existence of mathematical objects (or of moral truths, which in many respects is an analogous concept). If one is a realist about these objects one is saying that there is a ontologically thick sense in which, say, numbers “exist.” Ontologically thick here means that numbers exist in a mind-independent way, though not in the physical sense that you can point a telescope somewhere and see them. More along the lines of “if there are any other intelligent beings in the cosmos they will independently ‘discover’ the concept of numbers.”These terms are somewhat confusing, as the quantum mechanics literature uses them in a manner that may be opposite to what you expect. In that, the realists are the Einstein-Bohm-Bell followers who believe that QM is wrong and must be replaced by a theory of hidden variables, which are unseen and unknown to modern science. The anti-realists are those who follow the Copenhagen interpretation or something similar, and believe that the observations are what is real.
Being antirealist about numbers (or moral truths) means, of course, exactly the opposite: the antirealist doesn’t deny that numbers, once defined in a certain way, have certain objective properties. But she denies that it makes sense to think of any such definition in a mind-independent fashion.
The conventionalist, then, provides one possible antirealist account of numbers (or moral truths) to counter the realist one: numbers, like all mathematical objects, are human inventions, which are constructed in certain ways but could have been constructed differently. They are not “discovered,” they are invented. ...
Kaplan and Winther conclude (and, I think, are obviously correct) that the most sensible positions concerning race are: conventionalism about bio-genomic clusters, antirealism about biological races, and realism about social races.
Conventionalism is associated with Poincare, who believed that the aether was a useful convention, and you could believe or not believe in it, depending on convenience.
Applying these ideas to numbers is a little strange. Saying 2+2=4 is a universal truth that is independent of mind and convention.
Applying them to human races requires putting on ideological blinders. A commenter points to objective scientific genetic differences between the races, and he is just called a "racialist" who is undermining progressive work towards a post-racial society. Pigliucci says, "I can tell you that the majority of human population biologists don’t think that races exist in anything like the folk concept." He only admits that the genetic differences are useful for medical purposes.
Update: Pigliucci defends his political activism, because he wants to fight the creationist, and I guess, to attack those who believe in human races. This is funny because his racial denialism is just another form of creationism. Tney both deny human evolution and scientific evidence.