Localized generalitiesPosted: October 2, 2005 at 10:42 am in cognition, socialsoftware
One thing that I noticed in the comments on my “designing for the collective” post was that I have been using Latour’s term “collective” in an extremely fuzzy way, where I change what I mean by it depending on the point I’m trying to make. This got me thinking as to whether this was an appropriate thing to do or not. Problems in communication are often due to imprecision of language, and I was wondering whether my lack of clarity was allowing me to avoid or ignore problems with my line of thought.
After tossing this around for a while, I realized that one of the issues is that “collective” is an extremely general term. Even as Latour used it, it can refer to a wide array of organizations. And yet there was meaning to be extracted from this general term, as he found ways to adapt it to specific situations that he constructed and discussed.
My theory of the moment is that generalizations are useful in part because they are somewhat vague on specifics. They can be adapted to the needs of the local situation. Admittedly, I’m biased since I tend to be a deductive thinker. But let’s consider the extreme opposite case, somebody who observes all of the details around them, but can’t separate the underlying commonalities from those details. They would not recognize that a traffic light is a traffic light whether it’s aligned vertically or horizontally. They would think that different fonts represented different languages. I’m being silly for the sake of argument, but you get the idea. Looking for the underlying general principles in specific situations means throwing away some of the specific details. In the case of traffic lights, it’s the colors and order of the lights that matter, not the orientation. In fonts, it’s the general shape of letters, not the specific details of serifs.
The really brilliant people, the deep thinkers, are the ones who are able to identify useful separations between the general and the specific. They can extract generalizations that apply to a variety of situations that had been heretofore thought completely separate. I’ve discussed the power of theories that apply across different areas before. It’s important to remember that such generalizations are only mental tools. They are not necessarily “right” or “wrong”, because they are both – they are “right” in the sense that they unify observations across several disparate situations, but “wrong” in the sense that they might not apply exactly in any of those particular situations. So they should be judged on whether they are useful or not.
Too many people adopt an all-or-nothing approach to generalizations of this type. In the “all” case, they take the results of the generalization and apply them indiscriminately without regard to specific examples. Prejudices are a good example, where the person will ignore the evidence in front of them in favor of their prejudice.
In the “nothing” case, they think the theory is completely useless because it does not explain all of the details; a lot of critiques of evolution fall along these lines, trying to disprove just one element of evolution (from carbon dating to the fossil record), then standing back and saying “If that one thing is wrong, then the whole thing must be bogus”. Of course, their disproof is often faulty, but that’s another rant.
The true use is somewhere in between, because generalizations are tools. They may not apply in all situations. The user has to take responsibility and decide whether they are appropriate or not, and if they use them, to adapt them to the local situation. This adaptation is like the premise of “Web 2.0″, taking the general tool and building a local adaptation. All of the details might not be right, but the general principle may provide insight that is not apparent from the observations of the specific situation. And it can work the other way – the specifics can help inform revisions of the generalization. One analogy I came up with is that it’s like being under a good manager, one who outlines the overall goals of the project, but leaves it to the employees to figure out how to accomplish those goals using their skills. The relationship works both ways – general goals flowing down, specific plans flowing up – to achieve success.
So this is my theory of why I think some of my posts and theories are sloppy in their language. I am trying to get to general principles and sometimes let the details slip. One of the things I need to work on is to try to figure out how to apply some of my theories to specific situations, going from a descriptive mode to a prescriptive mode. But this idea of the general being adapted to the local is a powerful one that I want to continue exploring.
P.S. In an amusing coincidence, I had sketched out the ideas in this post the morning before I read the “invariant representations” section of On Intelligence by Jeff Hawkins, which are much the same idea. He’s coming at it from a neuroscience perspective, I’m coming at it from the realm of ideas, but it’s the same idea (finding general representations and applying them to specific situations). So, as usual, that was both satisfying (because my ideas aren’t totally crackpot) and disheartening (he’s already published). I should have read the book long ago – my cognitive subroutines posts are pretty congruent to his theory of intelligence. More coming when I finish it and write a review.