When I started my economics blog ranking, I caught some flack for choosing to exclude the excellent and popular Crooked Timber blog. It’s written by16 academics and intellectuals and does have some posts about economics, but many more posts about politics, academia, and other miscellaneous topics. So I decided that while it was a high-quality blog, its main focus wasn’t economics, and chose not to include it in my list.

Now I face a similar dilemma with Paul Krugman’s blog. Krugman is a very famous and gifted economist. He’s written many important and respected academic papers in economics and is a professor of economics at Princeton. But I’m having some doubts about whether his blog is really about economics. Looking over his recent posts, there are a few about economic issues, but many more about American politics and other American domestic issues, perhaps with a little bit of economics thrown in, such as the relationship between income and voting. So, is Krugman’s blog an economics blog or not?

I have a vague memory of my undergraduate statistics professor teaching us that “data mining” was bad. He was referring to the practice of conducting repeated statistical tests over and over on the same data. Since these tests are not perfect, if you do enough tests you’ll eventually find some kind of ’significant’ relationship or correlation or something when in fact none exists. However, data mining seems to have reformed itself since then. According to Wikipedia, “Data mining is the principle of sorting through large amounts of data and picking out relevant information”. Automated processes and techniques have been developed to perform these tasks in a sensible way.

Professor David Mease of Stanford University has what looks to be a very nice online course about data mining. The course website has slides and homework assignments. There are also videos of a version of the course that David taught at Google. Happy mining!

Further confirmation that marketing is the new finance comes from today’s news that Microsoft has paid US$240 million for a 1.6% ownership of Facebook. That puts the total value of Facebook at US$15 billion, which is a suitably impressive and large number. As the news article I linked explains, it’s all about marketing. Facebook is collecting loads of data about people, their preferences, their friends’ preferences, and so on. This is valuable stuff for marketers. As a tiny example, you can put in your Facebook profile a list of your favourite movies. So can your friends. Facebook knows your favourite movies, your friends’ favourite movies, and who your friends are. Assuming that you and your friends have similar tastes, they can use this data to target ads to you for movies that you might like but may not have seen.

This is valuable because the big problem about marketing is randomness. Different people have different tastes, and ads are more valuable if they give people information that is relevant to them. The old approach to this problem was ’shock and awe’ tactics, ie, widespread mass-media advertising. The trouble with this is that it’s inefficient — you pay for delivering ads to a lot of people who aren’t interested in what you’re advertising. If you can use some information to target those ads to people who are interested (’surgical strikes’, to continue the military analogy), then you’ll achieve a similar result by spending less money (less ‘collateral damage’ …). In economic terms, it’s a more efficient production process for advertising.

Of course the 15 billion dollar question is exactly how to do this. As always, the devil is in the details. How do you take that mass of information collected by Facebook and process it and use it to decide to show ad X to customer A but not customer B? The more effectively you can solve this problem, the more valuable your algorithm for targeting ads will be and the more profit your company will make. If I were Facebook, I would be employing lots of smart people to work on this kind of matching and information processing problem. If I were smart enough, I’d start my own company to work on these problems and develop customised advertising targeting algorithms myself.

This video gives some insight into the sophisticated mathematics and algorithms that can be used to attack this kind of problem. (Warning: extremely high nerd factor, but worth watching if only because complicated math problems sound so cool when explained in slightly broken english by a guy with a Russian accent).

Felix Salmon shows that data on number of RSS feed subscribers that Google Reader has recently started publishing about blogs can be used to form a ranking. Since Google Reader is the most popular RSS feed reader, this ranking should be pretty representative.

I’m very impressed that Freakonomics and Marginal Revolution score more than 70,000 subscribers each — I think that’s a viable number in terms of running a blog as a business. On the other hand, Felix’s numbers indicate that there’s a big difference between the two most popular blogs and the rest — his “second tier” bloggers such as Greg Mankiw and Brad DeLong have less than 2,000 subscribers each. Yet another “long tail”, I guess.

So how does my blog fare? According to Google Reader I have 55 subscribers, although Feedburner reports around 150 in total. Hey, it’s a start …

Following my earlier post about the Mankiw effect, the guys at the Environmental Economics blog were wondering whether a link from their blog would have a similar effect on my site’s visitor stats. Here’s the answer — not quite such a big effect as Mankiw, but a nice little bounce to offset Mankiw decay.