21.49, Sunday 17 Aug 2003

The key feature of the conduit metaphor (Heckler & Coch one, two) is the separation of things-that-resolve-into-meaning from things-that-perform-resolution.

It doesn't say anything about how the resolution takes place, about whether meaning is entirely contained, about how things-that-perform-resolution come to do it in similar ways, about what else is taken into account in negotiation of meaning (Grice's implicature, etc [via sylloge post]): it's transport layer.

Is this separation valid? (It's non-obvious, and all other features can be challenges. But this dichotomy is fundamental to it.)

It's coherent with metaphors that say meaning is containable in vessels; that meaning, once contained, can be handled independently of the transport mechanism; that things-that-resolve-into-meaning are passive and act as stimulus on things-that-perform-resolution (people); that communication between people is always mediated by passive things-that-resolve-into-meaning.

Within limits these are useful; employing them colours our understanding of other systems.

The industrial mindset, in which responsibility is "contained" and orders (to be resolved into actions) are given; Fordism and the production line in which objects (and people) are bodies given a stimulus for a predictable response (self-fulfilling, this one); the Semantic Web in which meaning is wholly contained; nouns/verb, nodes/arc, objects/method, commodity/value, signified/signifier; computer technology, and the telegraph itself -- all of these share the conduit metaphor.

I feel the conduit metaphor is misleading for the sort of systems we're trying to deal with. This leads into [my notes on]:

  1. Conduit metaphor as folk systems theory which is all about how this ties together
  2. Distance, shape and expectations which is searching for an alternative vocabulary (see also my post Distance is the half-life of causality)

For a more ranty version, read what my dictionary of the intertextual fabric is about. For a much, much better conception of what shape means in the software domain, read about emit/accept in biological computing at ETCON 2002, or far better is Jaron Lanier's Phenotropics [Powerpoint]. Mindblowing.

(On the subject of where word meaning comes from: understanding is having our expectation-shape filled; encoding information is just altering our extended body shape in meaning-space, just as in physical space we learn to manipulate our vocal cord muscles without thinking as we grow up. The method of learning these meanings is the same method as any expectation is "learned". On whether I think the Chinese Room produces intelligence (explanation)... I think it's the same question as M&M patterns being conscious (or not). They're like the intertextual dictionary that only uses the semantic content of words. Just as the dictionary also need order, implicature, an understanding of the author and the reader, context in all its forms (pragmatics?), consciousness as we have it is a climax state system and makes use of not just physical properties, but the property of where it's situated in the abstraction layers, of social context, of continuity in time/space and so on. If there's another way of having consciousness, fine, but it won't be the same as this one. Simularly, I'd say the one way to simulate the universe correctly is with the universe itself. So no solipsism either. On whether this is all nonsense: Of course. But read the Phenotropics presentation anyway, it's spot on.)

Follow-up posts: