Reinventing Discovery - Michael Nielsen [30]
In programming, the information commons took off in the early 1990s, with broad adoption of the internet. But in a more primitive form the ideas of reuse and the information commons were pioneered centuries earlier, in science. When someone publishes a scientific discovery—say, Einstein’s famous paper containing the formula E = mc2—other scientists can reuse that result in their own papers, simply citing the original derivation. This allows scientists to build on the earlier work without having to repeat that work. The citation both credits the original discoverer, and provides a link in a chain of evidence. If someone wants to know why E = mc2, they merely need follow the citation to Einstein’s original paper. The result is that, as in modern programming, a great scientist isn’t merely a person capable of enormously penetrating insights into nature, but one who also has a mastery of the information commons—aady published scientific knowledge—and an ability to build on that knowledge. Science is, in this sense, one big collaboration, built on the information commons.
Science’s citation-based information commons is powerful, but cumbersome and slow when contrasted with, say, the rapid-fire pattern of reuse in a project such as Wikipedia or Linux. A scientist who used the Wikipedia and Linux pattern—reusing someone else’s text word for word, but making a few improvements here and there—would likely receive an indignant note (or worse) from the original author. Yet such improvements are the lifeblood of many online collaborations, enabling extremely rapid iterative improvement, with people focused solely on moving forward, not on rehashing what is already known. A moderately active Wikipedia article may be modified 20 or 30 times by a dozen different people in a week. To get the same cumulative buildup of ideas in many areas of science might take years. Projects such as the Polymath Project speed up the cumulative building process of conventional science, creating a shared space where scientists can rapidly build upon one another’s ideas. Citation is perhaps the most powerful technique for building an information commons that could be created with seventeenth-century technology. But as the Polymath Project shows, and as we’ll explore in more detail later, modern technologies now enable a better way.
The MathWorks Competition
In 1998, a software company called MathWorks began running a twice-annual computer programming contest that is open to anyone in the world. For each contest MathWorks poses an open-ended programming problem. To give you the flavor of the contest, consider the problem used in the first contest, in 1998, a problem called the CD packing problem: to write a program which, when given a long list of songs, picks out a sublist that comes as close as possible to filling the 74-minute length of a CD. For example, your program might be asked to pick out songs from Pink Floyd’s back catalog. You run your program, and it finds a list of songs from the catalog that leaves just 35 seconds of extra space on the CD. But if your program had a better way of selecting songs, you might find yourself with only 15 seconds left on the CD.
The CD packing problem seems artificial. Not too many people have a need to burn CDs that are as close to filled as possible. Despite this, the problem is exactly the kind of challenge many programmers enjoy. It’s a simple problem that’s easily understood, but can be attacked in many different ways. Like all the MathWorks competitions, the original competition was very popular, attracting more than 100 contestants from all over the world.
Every program entered in a MathWorks