I based that on the 2nm diameter blip.
Sure one can reduce the size in software to something impressive but does this yield real position accuracy? From a 2nm wide return it won't - other than by averaging many returns which one can't do (for position fixing purposes) with a moving target.
A decent IFR GPS will be accurate to well within say 10m, everywhere, virtually all the time.
I would hope the radar would be more reliable, if not more accurate, because one could continually check it by marking certain ground returns (whose location is accurately known) onto the screen and checking where they produce a blip. I wonder if this is done.