- From: John F. Sowa <sowa@bestweb.net>
- Date: Wed, 29 Mar 2006 17:35:14 -0800
- To: Paul S Prueitt <psp@virtualTaos.net>
- CC: 'Hans Teijgeler' <hans.teijgeler@quicknet.nl>, Service-Oriented Architecture CoP <soa-forum@colab.cim3.net>, semantic-web@w3.org
Paul,
> On page 2 you talk about the lack of progress in building
> general purpose intelligent systems.... and provide three
> possible hypothesis...
>
> Penrose actually suggests something different from any one
> of these.
The issues about simulating true human intelligence are very
complex, and I'm not holding my breath waiting. But in any
case, it is fairly clear that some systems are easier to use
and satisfy a broader range of requirements than others.
My recommendation is very simple: it helps to keep your
systems flexible and modular. That makes it much easier to
try out new options without scrapping the whole framework.
The example I gave in the paper was Unix (with the unstated
counterexample of Windows). The extreme flexibility and
modularity of Unix makes it much easier to implement, modify,
and build on than Windows. After the spaghetti code that
lies beneath the old Macintosh became unmaintainable, Steve J.
moved to a Unix base -- which enabled him to build a system
that has the features of Microsoft's future Vista, but at
a fraction of the development time and expense.
The IBM FS was started around the same time as Unix, but it
was much bigger and it wasn't flexible or modular, and it
failed miserably:
http://www.jfsowa.com/computer/
That is a topic that began this series of notes.
John
Received on Thursday, 30 March 2006 01:37:14 UTC