[ExI] mit's answer to the stanford ai class

[Adrian Tymes]

On Wed, Dec 21, 2011 at 7:16 PM, Mike Dougherty <msd001 at gmail.com> wrote:
> The BS/MS/PhD challenge might be solved by the trust network of
> fellows with whom you've worked along the way.  If I already
> (inherently) trust 3 of your colleagues and they highly vouch for your
> competence in your professed field

That type of solution does not seem to scale.  There is no end of tale
of savants who are extremely capable in some field, but not so good
at making friends - and, conversely, people whose only skill is making
friends and convincing people to vouch for them, regardless of the lack
of factual basis for said vouching (and getting them to make an
exception to their honesty in that person's one case).  While some of
these tales are no doubt exaggerated, there is a substantial enough
basis in reality to make trust networks unworkable for this in practice.

> This is not much
> different than how such things are done "in real life"

People keep saying that, but at least in my case, that has never been
the case.  I've rarely gotten a job based on friend referrals - and even
when I have, that only played a part in getting me considered; my
demonstrated expertise was necessary for the rest.

> the same amount of time/effort as it
> currently takes to call your 3 colleagues for a reference.

I know there are some companies that actually call references, and
some fraction of them actually consider what the references say, but
this seems to be a vanishingly small fraction.  Most places, you could
just make up references, and no one would ever care (or know, other
than you).