2005-08-08

Schrödinger's Web

This is a follow-up of the previous post at Inbetween, providing yet more exciting thoughts about co-existence of many inconsistent descriptions of the same thing as a native feature of the Semantic Web.
It strikes me that if inconsistency is fundamental then it should be treated as such, not something to be avoided.
Follows the idea that maybe we need something like the logic of Quantum Physics for the Semantic Web. In such a framework, subjects would be seen, as quantic objects are, as superposition of mutually incompatible states, each one with a given probability. Pushing this concept further needs to define the notion of interaction. When you interact with a quantic object through an experiment, you get the very peculiar behavior known as wavefunction collapse in which the probability distribution changes suddenly in such a way that one particular state is actually "observed". Very long debates and crucial experiments eventually turned out to be rather in favor of the strictly probalistic interpretation, which some famous Quantum Physics founders (including Einstein) would not have been happy with.
I wonder if we are ready to go this far. Seems to me people will have hard time to, but eventually accept to live with the notion of subjects being by essence superposition of mutually inconsistent states, but going further to admit that observed properties of a subject in a given representation context are probabilistically determined would certainly prove at least as difficult as it has been in Physics. Took about a century there.

[Added] Thinking about it, the case certainly differs from Physics. The "semantic collapse" leading to specific representation properties is certainly not completely random, but rather likely to use some hidden variables depending on the representation context.

[2013-08-20] : A typical observation of the subject identified by a URI is a HTTP GET request on this URI. The state of the resource you observe depends on server and client conditions, plus state of the network, plus content negotiation parameters, caching etc. All those are mostly hidden variables indeed for the regular user, but the important point is that most of the time you can't define the state of the resource independently of a specific interaction.