I think there may be a good idea behind it though: view it as a cryptic appeal to Occam's Razor. Various moralists (e.g., Railton, Craig) were shown to be speaking of real things and properties, or of imaginary ones, with their moral language. Why not then hypothesize that all are - albeit less transparently than these two - and do away with the need of a special metaphysics or semantics (or both) for "ought" questions as "opposed" to "is" questions.