Information and Attention Scarcity

Friends sometimes ask how it is that I consume the volume of information that I do. But what is important is not so much what I choose to read, but when and how I choose to stop paying attention. Old models for valuing information (is the source trustworthy?) will cause you to discount good, early information of high value. Like a whale (or clam, for that matter), I flush the ocean through me, retaining the bits that catch in my filters.

In particular, this system is well-suited to blog comments, which are discounted as being essentially worthless, and indeed most are. But in the straw, one finds bits of needle, as well. I believe that a system capable of integrating and filtering this type of information effectively will be able to move significantly ahead of the curve.

A working example of my filtering process can be seen through the comments on this post on the economic situation at US News's website. The post says, 'the bailout is needed to prevent a much more costly Depression-scale event.' In this case, my goal is to see whether the comments reinforce or undercut the author's argument. Here's how I read the comments:

1. It's not our depression. It's the next generation and the one after that and so on that is stuck with this bill. Emotional reaction orthogonal to the question at hand.
2. The assumption here is that Paulson (and by extension the Bush administration) and $700 billion will fix all problems. Sorry, I don't buy it. Years of winding leverage and the adoption of faulty business models cannot be fixed by throwing (newly printed) money around. Failures and economic contraction are part of the cycle. Spaghetti and shibboleths. While the initial skpeticism is intriguing, what follows calls the merit of the commenter into question.
3. You've compared the highest possible costs -- well, half the highest and the highest -- of not acting with the lowest possible cost of acting (just the $700 bil). Let's see you cost out acting ($700b to $1tril, depending on estimates) plus the likelihood of more action being necessary, plus the recession that is coming anyway. Then make the comparison. Mostly factual statements, or short reaches. This is a good comment.
4. We are founded on a prinicple where no one branch of government is able to dominate in the republic. Allowing one person, in any capacity anywhere in this republic, to have such bizarre authority would cripple any opposition to that person's will. Axe grinding. Like the first, basically an emotional reaction that calls the observer's evaluation of events into account.
5. Isn't 'Bailout' Bush the same guy who hurried us into Iraq? Axe grinding.
6. There is nothing in the bailout package that will prevent this from happening again. Non-legal residents should NEVER be able to avail themselves of tax-payer backed mortgages.... Turtles, all the way down. This commenter looks for ways to link any issue to immigration. Maybe good for a marketing/PR flack but not an executive decision scenario.
7. Next, if we are going to pay the bill, we need to see the deal. The bailout of Fannie Mae and Freddie Mac were done in a dark closet. Anyone know the details of that $300 Billion yet? When I go to a bank for a loan, I get that loan based on their terms. Reduction to personal experience. At the margin, gut intuition by non-experts can be drastically wrong.

Backing out from these, we see a few rules emerge for cases in which we can stop reading/discount the contents of a comment:

1. Strong partisan bias
2. Linking to an unrelated issue
3. Relating large questions to personal experience
4. Spaghetti arguments with red herring sauce
5. Predominantly emotional reactions
6. Arguments of principle divorced from the issue in question

Obviously, the guy howling about immigration may have a bit of a point, and the other fellow saying that this is all unconstitutional may well be right. Politics often track Andy Warhol's definition of art as "whatever you can get away with." But if one were to try and incorporate all these bits and pieces, you could spend hours doing the equivalent of talking to the drunks at the local pub. Entertaining, perhaps, but not especially informative. By aggressively pruning suspect leaves from the tree, the remainder become far more useful as a measure of quality.