@liw This is a depressingly common metric in all kinds of analysis, particularly in economics.
I strongly suspect it's a case of "easy to measure" (tally up patent applications / assignments) trumping "meaningful measurement".
It's ... staggeringly close to measuring production by, oh say, "pieces of string produced" rather than measuring the length of the string(s).
The underlying problem is that measuring the value and/or significance of inventions is difficult. That said, the method you describe is really, really bad.
(I've spent way too much time looking for lists of / ways to assess "most significant inventions", and ... it's just a mess. That said ... yeah, stop with the bullshit.)
@dredmorbius @liw as we know, patents for software are negatively correlated with innovation thanks to patent trolls... https://web.archive.org/web/20120112152634/www.bu.edu/law/faculty/scholarship/workingpapers/documents/Bessen-Ford-Meurer-no-11-45rev.pdf
On the internet, everyone knows you're a cat — and that's totally okay.