This is fun. A month ago, I asked whether the Netflix competition was over. After BellKor’s Pragmatic Chaos merged the solutions of many players to break the 10% barrier, I expected them to win. It turns out that another coalition was created—The Ensemble—and they have beaten the previous best score by 0.01%. There are only a short few hours before the final deadline, so it is still impossible to know who will win. However, if The Ensemble wins, it will be very frustrating to BellKor’s Pragmatic Chaos: The Ensemble is a last minute player.

Both teams broke the 10% barrier by using a diverse coalition, by merging several different ideas. As Peter Turney recently stated:

There are no whole-truths, but we can get by reasonably well with a large number of half-truths.

I am now more convinced than ever that science needs diverse explanations, techniques and opinions. We should actively reward creativity. Science is not merely about truth-seeking.

Should you hire Ph.D. graduates from top schools in your country? Maybe not:

The present analysis however dispels the notion that graduates of high-status doctoral programs in the discipline of information systems will become superior researchers. (…) The findings indicate that productive scholars were not heavily concentrated among a few elite universities with respect to their academic origins, and that graduates of middle-status doctoral programs were as productive as graduates of high-status programs in terms of both research quantity and quality, if not more so. (Long et al., Scientometrics 2009).

(Hint: There is no substitute for getting to know people.)

Further reading: Big schools are no longer giving researchers an edge? and Why are top universities losing their lead? (Andras, Charlton and Source: Science and Public Policy, 2009)

I sometimes get annoyed at Americans who seem to think that the rest of the world is modeled after them. Here are some things many American academics seem to take for granted:

• Professors are paid for 9 months, the rest of their salary comes from eventual research grants. At least in Canada, this is false. As a professor, I am employed for 12 months a year. I can take a second job as long as most of my time is dedicated to the University. I do not know a single Canadian professor who gets an extra salary directly out of a Canadian research grant. Thus, in Canada, getting large research grants does not necessarily translate into a higher salary.
• There are few top schools, and many lesser schools, that is, universities are distributed according to a power law. In Canada, there are some more prestigious schools (Waterloo, UBC, McGill, Toronto…). Yet, there is often little difference in tuition or admission rates between universities. There are salary differences between professors, but not always in the direction you would think: many smaller Universities offer better salaries and working conditions. In my province (Quebec), working conditions for University professors are almost the same throughout: the teaching load is standardized. Thus, a professor at a smaller University has no excuse not to spend less time on research than someone at a larger University. Yet, I hear that in smaller American colleges, the teaching load of professors can be substantially higher: there is no direct equivalent in Quebec. In this spirit, you might be surprised to learn that the Canadian school with the most important research licensing portfolio is a school you probably never heard about (Sherbrooke University). Yes, you could say that Canada is a little bit more socialist than the USA as far as education is concerned: Canadian governments try to make sure that all Universities (and thus, their students) have somewhat equal chances. Of course, even in Canada, if you plan on an academic career, you should probably attend a larger school, if only to have more choices. But it is not uncommon in Canada to find top researchers at smaller schools.
• All Universities teach calculus at the elementary level. At least in my Canadian province (Quebec), most University degrees last only three years, the last three years of an American degree. Community colleges—not universities!—offer the equivalent of the first year of an American  University education. Thus, University-worthy topics in the USA are often of “community college” level for us.

Don’t you feel sometimes like your brain is running out of storage space? Myself, I am very forgetful. I always seek new tools to extend my brain.

FriendFeed is a fantastic social networking site. It lets you integrate all of your activities from all over the Web into a single flow. You can browse mine. Unfortunately, it has failed to attract attention. Worse: many of its features have been copied by Facebook.

Instead of trying to compete with Facebook and Twitter, FriendFeed should offer lifestreaming: an indexed and exhaustive record of everything you have done, for all times. In effect, FriendFeed should be an external-memory extension to your brain.

Note: This blog post was inspired by a chat with my colleague and social-networking researcher Sébastien Paquet.

Organizing learning around courses implies the creation of groups and a tight control by professors. It is convenient to organize students into classes, and grade students by topics. Industry-based economies are similarly convenient. They are hierarchical with a clear reporting structure. However, it is slow-moving and boring.

Tim Bray is proposing to rearrange Computer Science degrees in a Web-centric manner.  Here are some arguments he could have used:

• Many prospective students are interested in Computer Science precisely because they think that the Web is cool! I am certain a Web-centric curriculum would attract more students.
• Web technology is an endless source of Computer Science applications. For example, Web sites presenting sorted lists will be faster when using Quicksort than when using Bubble sort, Google uses Graph theory to quantify the relative importance of Web sites and Amazon uses Machine Learning to recommend products.

Alas, most schools are simply unable to be so agile. Simply picking a Computer language can be a political issue. Changing the pedagogy of several core courses within a few years might be impossible. Thus, nearly 20 years after the invention of the Web, many Computer Science programs barely address the Web.

My solution? My dream? I wish I could work within open problem-based Computer Science programs. And the fact that professors spend so much time with graduate students tells me I am not alone. And do not tell me it is not possible at the undergraduate level: people have been doing it for years in biology and medicine. But it is inconvenient.