That's the one, though I prefer this format:
http://herpolhode.com/rob/utah2000.pdf
I don't agree strictly with everything he says. It is a polemic rather than a careful argument. But still, he has a point and the data about OS research is strong (even beyond his data).
From googling around, might you be referring to:
"Systems Software Research Is Irrelevant"
http://doc.cat-v.org/bell_labs/utah2000/
And there's a little slashdot discussion:
http://interviews.slashdot.org/story/04/10/18/1153211/rob-pi...