[cap-talk] Wiki sympathy, incentive misalignments?

[Pierre THIERRY]

Scribit Jonathan S. Shapiro dies 31/03/2008 hora 17:01:
> But you asked about computer science. This has become a branch of
> mathematics, largely uninterested in practical concerns.

But I see many researchers and labs doing work on very practical
problems as well as very theoretical ones, even bridges between the two.
The issue is more a lack of application of the scientific knowledge by
the engineers.

Two friends of mine are engineers, one builds satellites (their solar
panels, actually) and the other searches various things in the ground
with fancy devices. Both seem to consider part of their work to keep an
eye on the scientific publications that might have an impact on their
own work.

Software engineers, on the other hand, seem to blissfully ignore the
very existence of research and publications in CS. And by their lack of
knowledge, they come up with obviously flawed designs or inefficient and
inadequate solutions for problems that we know the solution for years.

For example, how many regular expressions engines have an exponential
time complexity on some regular expressions because their authors didn't
know about deterministic finite automata, known practically as a perfect
vehicle for this task since the 70's?

I would not blame the engineers too much, though, as their lack of
curiosity is only part of the problem. It's an essential part of a
researcher's job to ensure dissemination of its results, and there are a
lot of CS labs where there's probably no efforts at all to try and reach
engineers.

In this regard, I like projects like Coyotos[1], BitC[2] or Why[3]: they
all have a pretty decent web site, where you can download and actually
test or use the practical results, as well as find the publications that
describe the theoretical aspects of these results...

  1. http://coyotos.org/
  2. http://bitc-lang.org/
  3. http://why.lri.fr/

Why is now even packaged in Debian, so the entry barrier to it is
incredibly low. Same goes for Isabelle, Coq or Proofgeneral. They are
the results of very theoretical scientific research, but you can easily
get them and use them.

Practically,
Pierre
-- 
nowhere.man at levallois.eu.org
OpenPGP 0xD9D50D8A
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://www.eros-os.org/pipermail/cap-talk/attachments/20080401/e2b47379/attachment.bin