Is Knowledge Paradoxical?
Joe Firestone,
All Life Is Problem Solving,
Apr 23, 2008
I'm still really sick today, so I'll keep this newsletter short. This item links to an ongoing discussion on the ActKM mailing list between Dave Snowden and his critics on the nature of knowledge and (therefore) knowledge management.
Personally, I don't think that the critic - Joe Firestone - is really getting what Snowden is saying. Snowden, for example, says "If you think in categories, then the world is presented in categories or a failure to categorize." Firestone responds with a long list of categories used by Snowden in his criticism. But that scarcely establishes that categories exist! Much less that Snowden is talking about them (think about it: if I talk about unicorns, does that establish that they exist? No. Nor am I in fact talking about unicorns when I am "talking about unicorns" (because unicorns don't exist) - the words misrepresent what I am representing.
*sniffle* *sniffle*
Let me explain. Two statements are (crucially) true:
1. Language represents less than what we know. This is what people like Polanyi meant when they said that some knowledge (personal knowledge) is ineffable. It quite literally cannot be expressed in words. It is, said Polanyi, a knowing how rather than a knowing that. I have sometimes tried to characterize this knowledge as resulting from a process of recognition.
2. Language represents more than we can know. When Chomsky talked about the "poverty of the stimulus", he assumed that knowledge expressed things that we could not know, and therefore must be innate. But it's not innate - rather, our language misleads us. We simply don't know it. What don't we know? Well, a variety of things - but specifically, causal generalizations, laws of natures, infinities, universals, and (most saliently) categories.
Categories? Yes. Traditionally, a category is defined as a set of necessary and sufficient conditions. So if you identify something that violates a condition - an x that is not P - you have undermined the category. Pundits throw categories back and forth at each other, trying to find exceptions, and thereby find the Nature of the World.
But categtories - even the things that seem really certain - are not like that. They are, as Wittgenstein said, 'family resemblances'. They are produced not by identification of necessary and sufficient conditions, but by (a computational process of) clustering. This is important, because (a) clusters are recognized, they are not inherent in nature, and (b) clusters change membership depending on what properties (vectors) are salient, from the point of view of the observer.
So you may say, but a cat is still a cat. But what constitutes a cat - at which point does a cat cease to be a cat? If it is dead, is it a cat? If it is dismembered, is it a cat? If a cat's head is sewn onto a dog's body, is it still a cat? If the cat's DNA is altered, is it still a cat? It all depends on what you think is important about being a cat - and that, my friends, is a property of the observer, not the cat.
All of our knowledge is like that. All of our knowledge is a 'zeroing in' on a fuzzy probability space, a space that shifts every time we shift out point of view ( a 'strange attractor', to use the jargon). If we think we have more precise - or more universal - knowledge than that, it's an accident of language, not evidence of some special ability on our own part. It's not that we can't know that things are true or false, it's that "truth" and "falsehood" are properties of sentences, and don't really apply here. To know (to paraphrase Hume very loosely) is to 'not be able to recognize as not P'. Or in the words of Mark Pilgrim, "I know I'm a writer because I can't not write."
**sniffle**
What does this even matter to me? Well, it's like the old lady said: it's clusters all the way down. The way we currently structure classes, the way we currently organize subjects, the way we currently manage collaborations and companies, are all on the model of causal generalizations, laws of natures, infinities, universals, and (most saliently) categories. The 'groups versus networks' is logically isomorphic with the current discussion. Our own representation of ourselves is - quite literally - "people of the word."
Well. Words are a wonderful facet, and I would never want to do without them. But we are, both internally and externally, much more complex, much more interesting, much more beautiful than mere words can describe. Maybe, when you're sick, even if it's just a cold, you see this.
Thanks for yoru patience, and it's back to the usual newsletter tomorrow.
Personally, I don't think that the critic - Joe Firestone - is really getting what Snowden is saying. Snowden, for example, says "If you think in categories, then the world is presented in categories or a failure to categorize." Firestone responds with a long list of categories used by Snowden in his criticism. But that scarcely establishes that categories exist! Much less that Snowden is talking about them (think about it: if I talk about unicorns, does that establish that they exist? No. Nor am I in fact talking about unicorns when I am "talking about unicorns" (because unicorns don't exist) - the words misrepresent what I am representing.
*sniffle* *sniffle*
Let me explain. Two statements are (crucially) true:
1. Language represents less than what we know. This is what people like Polanyi meant when they said that some knowledge (personal knowledge) is ineffable. It quite literally cannot be expressed in words. It is, said Polanyi, a knowing how rather than a knowing that. I have sometimes tried to characterize this knowledge as resulting from a process of recognition.
2. Language represents more than we can know. When Chomsky talked about the "poverty of the stimulus", he assumed that knowledge expressed things that we could not know, and therefore must be innate. But it's not innate - rather, our language misleads us. We simply don't know it. What don't we know? Well, a variety of things - but specifically, causal generalizations, laws of natures, infinities, universals, and (most saliently) categories.
Categories? Yes. Traditionally, a category is defined as a set of necessary and sufficient conditions. So if you identify something that violates a condition - an x that is not P - you have undermined the category. Pundits throw categories back and forth at each other, trying to find exceptions, and thereby find the Nature of the World.
But categtories - even the things that seem really certain - are not like that. They are, as Wittgenstein said, 'family resemblances'. They are produced not by identification of necessary and sufficient conditions, but by (a computational process of) clustering. This is important, because (a) clusters are recognized, they are not inherent in nature, and (b) clusters change membership depending on what properties (vectors) are salient, from the point of view of the observer.
So you may say, but a cat is still a cat. But what constitutes a cat - at which point does a cat cease to be a cat? If it is dead, is it a cat? If it is dismembered, is it a cat? If a cat's head is sewn onto a dog's body, is it still a cat? If the cat's DNA is altered, is it still a cat? It all depends on what you think is important about being a cat - and that, my friends, is a property of the observer, not the cat.
All of our knowledge is like that. All of our knowledge is a 'zeroing in' on a fuzzy probability space, a space that shifts every time we shift out point of view ( a 'strange attractor', to use the jargon). If we think we have more precise - or more universal - knowledge than that, it's an accident of language, not evidence of some special ability on our own part. It's not that we can't know that things are true or false, it's that "truth" and "falsehood" are properties of sentences, and don't really apply here. To know (to paraphrase Hume very loosely) is to 'not be able to recognize as not P'. Or in the words of Mark Pilgrim, "I know I'm a writer because I can't not write."
**sniffle**
What does this even matter to me? Well, it's like the old lady said: it's clusters all the way down. The way we currently structure classes, the way we currently organize subjects, the way we currently manage collaborations and companies, are all on the model of causal generalizations, laws of natures, infinities, universals, and (most saliently) categories. The 'groups versus networks' is logically isomorphic with the current discussion. Our own representation of ourselves is - quite literally - "people of the word."
Well. Words are a wonderful facet, and I would never want to do without them. But we are, both internally and externally, much more complex, much more interesting, much more beautiful than mere words can describe. Maybe, when you're sick, even if it's just a cold, you see this.
Thanks for yoru patience, and it's back to the usual newsletter tomorrow.
Today: 1 Total: 98 [Direct link] [Share]
