@manuel @david I think # would be better if unspecified assumptions were discovered and specified. One of the best ways to do that is for someone to reimplement it (not necessarily as part of the GS projec…