One way is to propose something that's mathematically equivalent to existing theory. If you have no testable predictions then you're not really doing science. You're doing philosophy.
I'm not sure if that's the objection here. They might also be calling it too vague to be a theory, or something else.
Verlinde does not build on MOND. An explanation and some extracts from his paper are in an earlier comment of mine:
https://news.ycombinator.com/item?id=13198573
(specifically at the paragraph half-way along in the comment, starting with "Sure Milgrom", if you want to zoom right in.)