A question that belongs on any list devoted to proto-languages for any
family is the correct methodology for determining the significance of
"look-alikes" for valid comparison.
Some years ago, Don Ringe produced some statistical studies which purported
to establish a statistical basis for this determination.
His "studies" were eagerly seized upon by "mainstream" linguists.
I assert that they were a useless collection of theories that
1) misanalyzed the problem; and
2) created statistics that have no bearing on the problem.
Is that general enough for an "employed at a university" (cafeteria does
not count) linguists to want to refute?
***