In some ways reading someone’s blog is like creating an artificial intelligence. I have never met “Matthew Yglesias” for example. I do not know this person (assuming he exists) and presumably never will. What I have done is sampled perhaps hundreds of examples of his thought processes and beliefs, by reading his blog posts. Over time, this has allowed me to build a fairly accurate picture of what “Matthew Yglesias” thinks. In a sense, I have been, through trial and error, building an AI version of “Matthew Yglesias” in my head.
For a time, I was a fairly avid reader of his blog. This has trickled away to the point where I basically don’t read him any more. Why? Because there’s no point. Through practice and observation, the “Matthew Yglesias AI” in my head has gotten so accurate that I can basically predict what he will think and say about virtually anything with 99.9% accuracy. So, there’s no longer any reason to read his actual posts: they no longer give me new information. For a while I did try to go back from time to time but it became so rare to encounter a “Matthew Yglesias” post that contradicted what my head-version of “Matthew Yglesias” already thought that it just wasn’t worth my time, and I started wondering why I was doing it. So I stopped.
According to this model of blog-reading, one will read a blog more if the blogger says interesting and fresh things, and less if the blogger continues to say the same thing over and over again. After all, the latter type of blogging is easy to model as an AI in your own head. Once that AI gets sufficiently good, you’d naturally lose interest in the actual blog.
So to keep readers it would seem above all else that a blogger needs to stay fresh and unpredictable. Note this is different from the usual interpretation of blogging which is that blogs are just ‘echo chambers’ people visit to have their opinions affirmed back at them. That may be the case for some blogs, but not for the sorts of blogs I tend to like. Even if I agree with a blogger about stuff, I’ll still lose interest if he becomes too predictable – too easy to model in my head as an AI.
In a sense the blogger’s challenge is to pass something like the “Turing test”, or Voight-Kampff machine, to prove to readers that he is not an AI. When a blogger becomes too AI-like, readers will just learn to build their own AI-versions of the blogger. Then why read? Just consult the AI. I can no longer tell the difference between Matthew Yglesias and the AI “Matthew Yglesias” I’ve built in my own head, so I no longer use the former.
On the flip side, one might take a more optimistic view and just say that Matthew Yglesias has succeeded in implanting his thought-programming in others, including myself. Before, I had no “Matthew Yglesias” AI in my head, then I started reading him, and now I do. If I ever want to know what Matthew Yglesias might think about something, my head-”Yglesias” will tell me (and will probably be right). I’ve been infected with “Yglesias”, it seems. And maybe that’s the goal of blogging, to infect others with your thoughts. Maybe being easy-to-model-as-an-AI is a feature not a bug, because people learn faster what you’re about.
I’ve read Tyler Cowen for years and years, after all, and I still can’t quite figure out what the heck he’s about. Which is probably why I keep reading.