VC's don't need to pay for tech research. They're in the
enviable position of having the entire population do their own
research and only come to them after they have a good product or idea.
That's the "better mousetrap" model of research. Someone designs a
better mousetrap, and (assuming we know that there is or can be demand
for better mousetraps) the economic puzzle is immediately reduced to a
well understood production, marketing, sales, and distribution
problem. That's a relatively easy puzzle to solve because, for
example, manufacturing processes for non-precision, small, simple
mechanical devices made from non-exotic materials are mature
technology. Competition in that arena seems to me to be mostly over
very slight differences in quality and cost.
For the most part, computing systems don't really work that way.
Software, especially, is an immature technology: new techniques and
new foundational components are easy to come by (though not for free)
and can drastically alter the landscape of possible products. There
is tons of head-room for improving product production pipelines. Even
the marketing and sales problems are wide open for aggressive
exploration: How, for example, will software be delivered to consumers
in 10 or 15 years? That question is tied deeply to question of how
computing systems will be architected in 10 or 15 years -- itself an
open question. The whole space is still, mostly, wild territory.
Currently, the software industry is mostly concerned with plucking
the (perceived) low-hanging fruit. There is little investment in R&D
aimed at building new foundational components: new "basis sets" of
software from which families of products can be constructed. There is
little investment in improving engineering processes. There is little
investment in exploring the nature of spontaneous demand for new kinds
of computing system. Consequently, numerous opportunities languish
and a lot of investment redundantly chases fads (start counting "set
top box" companies, for example).
I think the cheap workstation and Internet explosions have
something to do with this. Computing rapidly became cheap and
popular. A lot of things we already knew how to do suddenly became
valuable. It made sense to put up a shack with cubicles in Silicon
Valley and crank out boilerplate systems -- often quality wasn't
terribly important; engineering for the future wasn't terribly
important: quickly deploying computing everywhere, without much need
for advancement in technique or functionality, was the game.
In that period of growth, I suspect, a lot of investors came to
(incorrectly) regard software construction as about as well understood
as, say, house construction. You draft a plan, hire foremen, they
hire journeymen and everyone gets to work in the usual way. To build
the house, it goes without stating they'll work with standard cuts of
lumber, standard plumbing fixtures, and the usual components for
But in fact, the software industry is ripe for revolutionary
changes in process, technique, and foundational building blocks.
Those are the areas in which VC-driven R&D investment is both needed
and likely to pay off.
It isn't necessary, by far, for VCs to match the R&D budgets of the
IBMs, Suns and Microsofts of the world. Indeed, the R&D organizations
hosted by those organizations are mostly working on a different
problem entirely: extending the life of the legacy systems which are
their source of funding. There is ample room for a few dozen software
engineers to upset the board, invent a new future, and spin off plenty
of immensely profitable businesses. It is a good bet that if a few
hundred or thousand engineers are put to the task, that at least some
of them will succeed.
The problem with research is that you're never guaranteed anything
useful. VC's can sit around waiting for a useful idea at no cost to
them, and then pick and choose which ones they want to invest in.
It's true that no particular line of practical research is
guaranteed to produce anything useful. You might also want to note
that, given the current state of product engineering, a VC armed with
a perfectly viable product idea and more than enough money is
also not guaranteed to produce anything useful. In fact,
VCs in that circumstance have what ought to be regarded as a dismal
At the beginning of, say, the automobile industry -- it made very
good sense for capitalists to invest not simply in new models of cars,
but in the entire process of designing and producing cars. Todays
computing system VCs, if you believe what they write in their
"investment focus" or "strategy" web pages, are 100% focused on the
next model of car. They need to expand their horizons.
[ Parent ]