Sunday, May 16, 2004

Phenotropic Computing

I've been thinking about this -- approximate "surface-based" interfaces. Intuitively, it seems like you would want it to work like molecular receptors, 3-d lock and key fashion. That seems hard to model in any current computer methodologies.

In terms of what's out there now, it seems like two software agents trying to communicate would 1st have to negotiate ontologies -- i.e., do we both speak the same language or know about the same things such that we have a topic of communication. W3C has published their ontology language for the semantic web: OWL, the Web Ontology Language, presumably everybody would speak that. Clearly easiest would be exact ontology match. If not, maybe a subset ontology matcher?

Speaking of agents, haven't seen much press on intelligent agents lately ...

So, what would these agents be, a new flavor of Web Service? If so, then we probably need PWSDL -- Phenotropic Web Service Description Language. Seems like it would not look like WSDL, with its "send me this, I will give you this". Rather, it seems like it would be more of a language syntax type thing. Simplest sentence: verb noun, both pulled from the ontology, such as "Create Object" or "Fetch Object". Rather than an input argument list like WSDL, you would instead populate the verb and noun with appropriate properties, has-a instances, etc.

Presumably your organic behavior would come from the subset ontology matcher, and you would also want synonym matching on the syntax elements, and defaults on everything. I wonder if this winds up being stuff you can do with XPATH or XQUERY. I'm not up on either of these, nor on XSLT -- the ugliest programming language since RPG. I guess tho, that the XML-based stuff is ugly because it is very LCD. Kind of like LISP as Lots of Stupid Insipid Parenthesis, maybe XML as eXtremely Many Left-angle-brackets.

I also wonder if, analagous to the cell incorporating mitochondria and spirochetes, you could have services that grew by ingesting unknown ontologies. Sounds like it'd be lots of fun to code.

Had an interesting night Friday. We have a QA person who is in his mid-50's and Japanese. He came to the US 10 years ago to study vibrophone under Gary Burton at the Berklee School of Music in Boston. He and his wife bought a house and had an open house. I took my guitar to play. The other musician there was a French-Canadian support person who played keyboards -- but all show tunes, 50's standards, etc. A couple of times the Japanese gentlemen and I were getting a jam going on a tasty riff, then from the piano would some "Volare". A little frustrating.

The best part was some world class cognitive dissonance. There was a QA person there, Chinese from Beijing in her mid-30s, in the US 10 years. She had a beautiful, clear soprano -- and beautifully sang 19th century and earlier American standards -- "Red River Valley", "Beautiful Dreamer", "Yankee Doodle Dandy" -- in Chinese ?!?!? She said they were popular in China when she was growing up ?!?!?

Then, my 21 year old baby daughter was having a party that we were invited to go to. So we took the Chinese woman and her husband and 6 year old son and another friend with us. "Invasion of the Parents". It was kind of fun.

No comments: