Digifesto

The machine learning research blog hunch.net humbly describes itself like this:

This is an experiment in the application of a blog to academic research in machine learning and learning theory by John Langford. Exactly where this experiment takes us and how the blog will turn out to be useful (or not) is one of those prediction problems we so dearly love in machine learning.

So Langford‘s introspective post about his own experience with scientific “blogophobia,” as documented by Scientific American is a poignant one. SA and Langford both evidence that the problem is that blogging flies in the face of some crucial scientific virtues. Published scientific works are, at their best, humble, meticulous, and precise. Blogs are, more often than not, self-promoting, off-hand, and gestural.

Nevertheless, the gains to be had by openness and connectivity in scientific research are the same to had by everyone else on the internet: faster and more efficient processing of information. As Langford writes,

…the real power of a blog in research is that it can be used to confer with many people, and that just makes research work better.

What’s interesting is that the social sciences appear to be moving towards blogs much faster than the hard sciences. Perhaps it’s the accessibility of the material, or perhaps its the tradition of social scientists as public intellectuals, but these folks seem to have no qualms about putting out their latest musings. Alex Tabarrok‘s 6-year old reports that blogs are all his dad’s friends talk about at parties. Meanwhile, rockstar blogging economist Tyler Cowen is so optimistic about the role of blogs in academia that he believes that they will replace the literally ancient concept of “schools of thought”:

Overall I don’t believe in schools of thought for modern economics. Think of the notion of a school of thought as a brand. The whole point of the internet is to break down branding into the evaluation smaller units, including individuals and their very particular ideas, even doing cite counts paper by paper. Why move toward more macro branding in that kind of environment? You can think of trustworthy bloggers as another means of branding and also as substitutes for schools of thought.

Of course, these social science blogs aren’t about research–they are fountains of insightful abduction, not earnest induction. But they represent an encroachment of the web into academia which I think can only help the cause. Research has a lot to gain from the openness that the internet facilitates. Blogging academics are an important step in that direction.