The Intelligence of the Blogosphere

Glenn Reynolds of Instapundit linked to this Harvard blogger who included Instapundit in his research on the interconnectedness of blogs, as shown in this chart:

The new science of artificial "neural networks" suggests that one can model the blogosphere (or any human community) as a neural network. Each blogger has a set of regular inputs (like the dendrites on a neuron) and a single output (like the axon). "Learning" occurs as synapses form and/or break between dendrites and neurons (and/or strengthen and weaken), with the "intelligence" of the system emerging as a result of interconnected nodes.

There's no real question whether the blogosphere (and other networks of communicators, such as the traditional media) is a neural network--the more interesting question is "how intelligent is it?"

That turns out to be a remarkably easy question to explore. Politics divides the blogosphere neatly into left and right sides, and both sides of the blogosphere are constantly making predictions. Those predictions get tested every election cycle. That means researchers should be able to:

1. map out the general structure of the neural network on both sides,


2. collect pre-election predictions,


3. compare those to public opinion polls (as a "baseline"), and


4. compare the "spread" between predictions and polls to the actual results.

A neural network built around the axis of the DailyKos/DemocraticUnderground/FireDogLake blogs might prove to be more "intelligent" than the competing net that includes Drudge/Instapundit/RedState, or it might not. As a matter of pure science, I'd love to know which is better.

And, as a political junkie, I'd love to add all the "intelligence" I can find to my own side!