Remarkably accurate for glottochronology, which made every known mistake and invented a few new ones (though I failed one part of my comps for saying so). Dropping first and second person usually makes sense -- though I would miss them -- but third, with the necessity to actually repeat a lot of the time, is less desirable, I think. Still, be nice to see how it works out. It might even give tp some pointers. Oh, several of the problems with glottochron were in the compiling of the list, which proved to be a less than universally applicable tool (to put it mildly -- it didn't work for the second pair of languages it was tried on).