Thursday, April 19, 2007

The Vagueness of George Siemens

I like George Siemens and he says a lot of good things, but he is often quite vague, an imprecision that can be frustrating. In this discussion of my work on connective knowledge, for example, he observes, "In this model, concepts are distributed entities, not centrally held or understood...and highly dependent on context. Simply, elements change when in connection with other elements." What does he mean by 'elements'? Concepts? Nodes in the network? Entities? You can't just throw a word in there; you need some continuity of reference.

Why is this important? Siemens dislikes the relativism that follows from the model. Fair enough; people disagreed with Kant about the noumenon too. But he writes, "I see a conflict with the fluid notions of subjectivity and that items are what they are only in line with our perceptions...and what items are when they connect based on defined characteristics (call them basic facts, if you will)" And I ask, what does he mean by 'in line' or 'defined characteristics... basic facts' (if they are defined, how can they be basic facts?)

Then he says, "I still see a role for many types of knowledge to hold value based on our recognition of what is there." Now I'm tearing my hair. "Hold value?" What can he mean... does he know? Does he mean "'Snow is white' is 'true' if and only if 'snow is white'?" Or is he simply kicking a chair and saying "Thus I refute Berkeley." In which case I can simply recommend On Certainty (one of my favorite books in the world) and move along.

He continues, "The networked view of knowledge may be more of an augmentation of previous categorizations, rather than a complete displacement." Now I'm quite sure that's not what he means. He is trying to say something like 'knowledge obtained through network semantics does not replace knowledge obtained by more traditional means, but merely augments it.' Fine - if he can give us a coherent account of the knowledge obtained through traditional means. But it is on exactly this point that the traditional theory of knowledge falters. We are left without certainty. You can't "augment" something that doesn't exist.

Here is his main criticism: "At this point, I think Stephen confuses the original meaning inherent in a knowledge element, and the changed meaning that occurs when we combine different knowledge elements in a network structure." Well I am certainly confused, but not, I think, as a result of philosophical error. What can Siemens possibly mean by 'knowledge element'. It's a catch-all term, that refers to whatever you want it to - a proposition, a concept, a system of categorization, an entity in a network. But these are very different things - statements about a 'knowledge element' appear true only because nobody knows what a 'knowledge element' is.

He writes, "Knowledge, in many instances, has clear, defined properties and its meaning is not exclusively derived from networks..." What? Huh? If he is referring to, say, propositions, or concepts, or categorizations, this is exactly not true - but the use of the fuzzy 'knowledge elements' serves to preclude any efforts to pin him down on this. And have I ever said "meaning is derived from networks"? No - I would never use a fuzzy statement like 'derived from' (which seems to suggest, but not entail, some notion of entailment).

He continues, "The meaning of knowledge can be partly a function of the way a network is formed..." Surely he means "the meaning of a item of knowledge," which in turn must mean... again, what? A proposition, etc? Then is he saying "The meaning of a proposition can be partly a function of the way a network is formed..." Well, no, because it's a short straight route to relativism from there (if the meaning of a proposition changes according to context, and if the truth of a proposition is a function of its meaning, then the truth of a proposition changes according to the way the network was form).

What is Siemens's theory of meaning? I'm sorry, but I haven't a clue. He writes, "The fact that the meaning of an entity changes based on how it's networked does not eliminate its original meaning. The aggregated meaning reflects the meaning held in individual knowledge entities." An entity - a node in a network? No.

He has to be saying something like this: for any given description of an event, Q, there is a 'fact of the matter', P, such that, however the meaning of Q changes as a consequence of its interaction with other descriptions D, it remains the case that Q is at least partially a function of P, and never exclusively of D. But if this is what he is saying, there is any number of ways it can be shown to be false, from the incidence of mirages and visions to neural failures to counterfactual statements to simple wishful thinking.

But of course Siemens doesn't have to deal with any of this because his position is never articulated any more clearly than 'Downes says there is no fact of the matter, there is a fact of the matter, thus Downes is wrong'. To which I reply, simply, show me the fact of the matter. Show me one proposition, one concept, one categorization, one anything, the truth (and meaning) of which is inherent in itself and not as a function of the network in which it is embedded.

Siemens says, introducing my work that I explore "many of the concepts I presented in Knowing Knowledge...and that others (notably Dave Snowden and Dave Weinberger) have long advocated - namely that the structured view of knowledge has given way to more diverse ways of organizing, categorizing, and knowing."

I don't think this is true. Siemens, Snowden and Weinberger may all be talking about "more diverse ways of knowing" - but I am not talking about their 'diverse ways of knowing' but rather - as I have been consistently and for decades - on how networks learn things, know things, and do things.

3 comments:

  1. "... I have been consistently and for decades - on how networks learn things, know things, and do things"

    That would be consistent Stephen with the perception that "networks" for want of a better clustering nomenclature are inextricably written for the age of ubiquity. Your term lacks personage and this allows for emotion free neutrality to flatten the ego out of the topic.

    Well done.

    You comment, I know. They know cause you pressed publish and now it's feed sits in my phone. She buzzed and he twittered.

    What annoys me about the connectivist diatribe is the lack of neutral clustering acknowledgement. It's all gooey human feelgood and dosent seem to acknowledge that a vast statistic of interaction is automated grid pings.

    Backl to the console. I've got a lot more keys to stroke and a it's all kind of qwerty. Qwerty. Qwerty. Qweerty.

    Try it sometime. Oh.....famous cinema.

    All work and no play makes Jack a dull boy. All work ......

    No machine would make any connections with that banter.

    ReplyDelete
  2. I agree. I also hate imprecision in the use of language. I read the Siemens post as well but I didn't have your energy to make sense of it.

    Was the "snow is white" example meant to be a Tarski/Davidson sentence: "Snow is white" if and only if snow is white?

    Sometimes I think that Pask probably had the right view on concepts and relations but I find him confusing as well, so I'm not sure.

    ReplyDelete
  3. Was the "snow is white" example meant to be a Tarski/Davidson sentence: "Snow is white" if and only if snow is white?

    Yes.

    ReplyDelete

I welcome your comments - I'm really sorry about the moderation, but Google's filters are basically ineffective.