This paper seems to suggest that a possible counter attack is to create bogus edges and adjacencies--thus fucking with the utility of a (sparse) adjacency matrix or attempts at having useful window sizes for BFS or the like.

I suspect, though, that any sort of attempt at that would somehow be thwarted and just cause more pain on the attempting party.