I'd be interested to see a comparison of something like "dollars of revenue at ten years of age per dollar of initial investment" between Silicon Valley startups and newly formed companies of other kinds. That seems like a reasonable measure of the actual worth of a company to me, and so a decent test of whether Silicon Valley works or not.
Looking at valuations or even market caps is much less useful, because that tells you more about what's going on in investors' minds than in reality.
You have a point, but in this context necessarily
works
has to mean making money for the limited partners
(LPs) and, there, more money than the alternatives,
e.g., an index fund of the NYSE. Well, follow the
link I gave to AVC.com and Fred Wilson's table of
data there and have to conclude that on average information technology
venture capital, and, thus, nearly necessarily
also Silicon Valley, doesn't work very well.
I did mention NSF, etc.: There, Ike wanted pictures.
The U-2 was not getting it done. So, one day
Lockheed's Kelly Johnson walked in with a pile
of papers and proposed a new plane, fly
at 80,000+ feet, at Mach 3+, for 2000+ miles
without refueling, and essentially impossible
to shoot down. The project was approved;
the plane flew as planned; and we got the pictures.
So, Kelly had a very ambitious start-up,
and delivered as planned, with the project
approved from work essentially just on paper.
Such work can be done. Other cases include the
F-117, GPS, passive sonar, Keyhole (essentially
a Hubble but before Hubble and aimed at
earth instead of space).
It really is possible to do accurate evaluations
of ambitious projects just on paper and, then,
go forward with a quite high batting average.
And this fact does partly respond to your point.
Well, NSF, NIH, DARPA, CIA, DoE, etc. can work
this way, but Silicon Valley just flatly refuses to.
Instead, they want another SnapChat with
traction and f'get about anything more,
and that means that they can't do the really
valuable projects. And that's why, on average, they don't make the big bucks.
Looking at valuations or even market caps is much less useful, because that tells you more about what's going on in investors' minds than in reality.