RE: Design shift of view

On Monday, July 27, 1998 11:28 AM, Rob Dixon [SMTP:rob.dixon@erros.co.uk] 
wrote:
> Buck
>
> I am glad that you find looking at our industry from a broadened
> viewpoint
> to be of value.

Every place I've worked for has been somewhat isolationist; budgetary 
restraints lead to a more narrow view of things.  Without looking outward 
and onward, we'll never learn about more efficient ways to do things.  So, 
for me, even if I never use any of the things we're talking about in this 
thread, at least I'm listening and combatting the isolationist 
tendencies...

> I entirely sympathise with your view on new paradigms and your wish to
> highlight similarities between methods rather than differences.  I agree
> that this would make the sale easier.
>
> My first reaction however was that it was not possible since the
> connectionist methods of the Neural Database are so different from
> traditional computing methods.  Then it occurred to me that before
> computers, our method was of course to rely on the human brain.
> Connectionism is based on the brain, so we might describe its use in
> computing as a "return to basics" as much as a new paradigm.
>
> Do you think that this approach might help?

Yes, I do think this is a helpful concept.  I've learnt all too late that 
even the very best of ideas are subject to the political process: if you 
can't sell it, it's not going to happen.

AI has always been something of a hobby for me.  I had the honour of 
speaking with Marvin Minsky in 1986, and he's quite a character.  He 
refers to the human/brain combination as a "meat machine."  By which he 
means that with proper study, we should be able to disassemble, understand 
it and make it better (as the lever is better than the elbow.)  This 
school of thought is countered by the "mind is greater than the sum of 
it's parts" school; meaning that there is some intangible "essence" about 
it that cannot be captured, understood and bettered.

Rather than risk starting an unrelated philosophical debate, let me say 
that I think it's almost irrelevant which school is right; the question at 
hand is whether a *more* mind-like approach is better able to adapt to 
modern MIS needs than the *less* mind-like (mathematical) models we use 
now.

Offhand, I'd say that if connectionism is to advance from a small research 
effort to a mature product, there needs to be publicity (It's the 
marketing!) and actual applications software out in the field (demo 
copies?)  The trick is in quantifying how much better it is to solve a 
business problem with the connectionism model versus the relational model.

Buck Calabro
Commsoft, Albany, NY
mailto:mcalabro@commsoft.net

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---