|
| |
|
Saturday 4 May 2002
|
The Feyerabend Project
Patrick Logan
mentioned the Feyerabend Project yesterday.
I had read some about
Paul Feyerabend
in John Horgan's
The End of Science, which I hope to finish
reading one of these months.
Anyway,
the Feyerabend Project
is a project mostly inspired by Richard Gabriel's
ideas -- he describes it as his
"attempt to repair the arena of software development and practice".
I poked around a little, finding the following
links to be among the best summaries and pointers
to what it's all about.
Very briefly, my summary would be:
let's reset to zero and rethink everything
we know about computing.
Some of this seems quite related to some of the
systems integration
work that my brother Chris and Kirstie Bellman have
been talking and writing about for several years,
e.g., at
http://www.scs.org/confernc/wmc00/text/vwsim.html
and
http://nestroy.wi-inf.uni-essen.de/workshops/WETICE99/wetice99-report/
And the folks running Sun's "N1" thing probably oughta be
monitoring this stuff, too.
I have some reservations about the Feyerabend Project.
The languages we use in working with computers, even
at or below the level of assembly language, instruction
sets, and machine language, are ultimately still human
languages. Success, for human languages, depends on building
and maintaining a user community.
Radical departures from "popular" solutions
make it difficult to build community.
So I just wonder if this project might come up with
something beautifully elegant (like
the constructed language Lojban)
but unable to build a critical mass of users.
11:03:22 AM
|
|
New Sun Strategy: N-O-N-E
John Markoff's NY Times article
"Unfazed by Defectors, Sun's Chief Charts Next Era"
includes this:
To combat these, Sun is preparing to introduce a strategy, known
internally as N1, a combination of hardware and software that will
in effect combine the entire computing resources of a company --
from desktop PC's running Windows, to I.B.M. mainframes, Sun servers,
inexpensive Linux computers and Cisco routers -- to work as one
vast computer.
Actually, this sounds like an interesting direction
for Sun to go, and they may have some luck with it if
they lose a bit of their C bias (where "C" means "C/C++/Java").
However, it's very hard for me to picture Sun doing anything
like this that couldn't be done
cheaper, more easily, more efficiently, and more effectively
with Linux boxes.
My guess is that Sun will stick with their C bias
and thus won't be able
to make the best use of design approaches like Erlang's excellent
interprocess communications primitives, or the tuple
spaces pioneered by the Linda language/system.
But the real trigger for this posting was their code name "N1".
Spelled out, it's "N-O-N-E".
I guess most people don't care about stuff like that:
e.g., Exxon still has that double-cross right in the middle
of their name; and the Grocery Outlet still has my favorite
corporate abbreviation: we call it the "Gross-Out".
8:59:39 AM
|
|
More words than you need
On the other hand, if I take the set of English phonemes
that I distinguish in mlang, I have 25 consonants and 15
vowels. So without even looking at diphthongs, I end up
with 19.77+ billion four-syllable (eight-letter) words, and
7 and a half trillion five-syllable ones. That oughta last us
awhile.
Here's the Mlang table. In Ken Thompson's immortal words,
"You are not expected to understand this."
| p | t | k | u | o | 4 |
| b | d | g | q | 2 | w |
| f | 3 | 5 | a | e | r |
| v | 6 | 7 | i | y | l |
| h | s | z | 1 | 0 | 8 |
| m | n | 9 | c | j | x |
12:21:48 AM
|
|
|
|
© Copyright
2007
Doug Landauer
.
Last update:
07/2/6; 12:06:19
. |
|
|
|
|
Random radio Weblog