[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Is Systems Software Research Irrelevant?

In what Rob Pike writes, I see two major points:
1) Research results are not making it sufficiently fast into "mainstream"
(industry, products, projects, public awareness, ...).
2) Research is done in the wrong direction (microspecialization, no breadth,
no long-term, ...).

I tend to agree roughly with both observations, though with a different

ad 1)
I think you have to be aware that the adoption of research results is a case
of common learning. There is much more involved than only scientific merits
and advances. Besides economic factors (ROI, shareholder-value, ...) that
often compete with the introduction of innovation (often to the worse for
both), people on a broad range (technicians, engineers, managers, deciders,
...) need to learn the new things, before they can be introduced  in common
efforts, like an economy. Many good things are beaten to death, after they
were "hyped" into the sky, not deserving both of it. And there is an
irrational part to it:  A wise man said "There is nothing more powerful than
an idea who's time has come". "Innovations" during the last years, like Java
and the Web, showed that it is not about radical new technology. Their time
had simply come.

Now if you apply that to the amount of unused, not deployed research results
living in academic books, papers and heads, there's a lot more already
existing that waits for its time to come. Just remember AI in the eighties.
I'm living now in a world where you cannot mention AI in a serious situation
without losing your credibility instantly. Yet its concepts are valid (mind:
I'm not saying "omnipotential" ;-), and they are making their way back,
under cover: In Web services, search engines and customer relationship
management tools, in image processing and agent technology, not being labled
as AI.

ad 2)
I think the evolution of technology runs through phases. In the first phase,
where everything is new, there is a lot of diversity. Then things settle
down to a couple of concepts and strategies, which are explored in greater
depth. When a certain amount of saturation or exploitation has been reached,
entirely new things are looked for and paradigm shifts occur. - With
computer technology, I think we are at the beginning of the second phase
(automobile industry is probably somewhere towards the end of it). I'm very
happy that things start to settle down, so they can be explored in greater
depth. This brings research also closer to practice (mention caches,
networks, distribution, ...). And I think it is unfair to call this not
innovative / not creative. It takes a lot of ingenuity to spell out the
consequences, details and implication of an initial approach. Comparison is
a valid and very necessary source of insight. And exploring the depths is
just what we need now in the current phase. Rob Pike is probably looking for
paradigm shifts that are not due.