Yes the signal has noise. But it is not all noise. The solution
isn't to throw away the signal, but to find less noisy signals.
>I rarely even am as concerned about what someone knows as I am with
>their work habits and whether they can think.
But can you design a credential/test which strongly correlates with
these features? If you could, and if others employers value these
features like you do, you could make a *lot* of money creating a
credentially service.
A pretty-free market currently provides certain credentails, and the
market is pretty open to those who want to market new credentials. So
if you think you have a good idea for a product here, I encourage you
to create it. But in the absence of much serious effort here, I think
it is also reasonable for the rest of us to think that it is pretty
hard to beat the existing educational credentialing services.
I've always been curious why IQ scores don't appear on resumes. IQ is
a well documented and studied credential shown to have high
correlation with future success. Personal interviews, in contrast,
look really lousy in academic studies. Is it a signaling problem,
where anyone who would choose to put their high IQ on their resume is
likely a member of a small subset of the high IQ folks who you don't
want?
Robin D. Hanson hanson@hss.caltech.edu http://hss.caltech.edu/~hanson/