Filed under: %a la mod
More Shirky on information.
There is a key point that gets lost when optimistic tech enthusiasts ejaculate over the glories of “alternate organizational systems” which “like the Web itself,” “let individuals create value for one another,” “often without realizing it.” Shirky is complex enough to both make every facet of this point and to lost it in the same essay. The point is that ontologies are all about seek times, reliability, and parallelism.
A good ontology is self-similar; when you reflect upon it, the ontology itself reminds you of truths you know about interrelationships between different concepts and different aspects of the world. The best ontologies have a non-null learning curve; you get more out of it with appreciation and practice. A good ontology is largely orthogonal; it creates deep and meaningful divisions in the unbroken flesh of raw thought.
If I tag everything I see as quickly as possible — free association, tempered by habit — I will be far from the ideal ontology, both for myself and for reuse by (or the emergent enlightenment of) others.
Yes we should collectively listen to the casual ways masses of anonymous users classify things. Better still, we should teach them ways to improve their personal classifications so that they will scale and age better. But this does not mean we should let these masses dictate what the best classification/ontology/search-algorithm looks like.
My armchair proclamation: the best systems [for finding information] are patiently considered, organically informed but not dictated by large bodies of users, and steadily improved in ways that teach users how to effectively form the questions they didn’t know they were asking. These systems should provide answer-sets that expand searchers’ concepts of what they were looking for, and should preempt clarification when possible.
When I look for “Georgias” I should discover, in separate taxonomically-contextualized sections, results for the US state, the Eurasian nation, the woman’s name, and the ancient Greek sophist (see Gorgias). Each of these should be well-identified by its place in at least one (and preferably a few named and referenced) well-conceived, self-similar ontologies.
I have nothing against Shirky, btw. It is the very excellence of his writing that makes it such a pleasure to take issue with it.
good to know you don’t have very many requirements for your organizing constructs 😉
Any general classification scheme, designed to be all things to all people, is liable to be the worst thing for all people (no matter how hard you try) because of how differently people think about the world. OTOH there’s not much else we can do other than try and make such systems better — better search, better tagging, better options, better deliniation of language. It’s a bit of a dilemma, but apparently a rather sexy one to be working on at the moment.
Comment by phoebe 08.11.05 @ 7:18 amwhat I meant to say was this:
Comment by phoebe 08.14.05 @ 5:51 pmhttp://www.livejournal.com/users/silencius/4058.html
A dynamic classification scheme could keep up. Providing classifications in as many schemas as are available, allowing schema authors to modify schemas (but retaining schema-versioning), offering schema-weighting based on a combination of personal and personal-network preferences… all of these would be major improvements on the present systems.
And the present systems would remain available as a limit of the above… so any transition could be made quite painless.
Comment by sj 08.18.05 @ 3:00 pm