“The press is a machine, it doesn’t have any freedom. Freedom belongs to the people, they have a right to make choices.” — Dali Mpofu, CEO of the state-owned South African Broadcasting Corporation, quoted in the Mail & Guardian.

More funnies in the same story, from Snuki Zikalala i’Afrika, the government spin-doctor turned SABC news director who has undisclosed information on the dipso klepto health minister that differs from everyone else’s, and Ronald Suresh Roberts, the presidential hagiographer who thinks journalists harbour subversive fantasies involving baked beans and trespassing on the body of said minister.

This is somewhat surprising. In this picture, you might see expensive toys. Dr Gaurav Khanna saw the makings of a cheap supercomputer.

Instead of paying the government $5,000 a session to rent time on its supercomputers, this enterprising physicist simply clustered eight Sony PlayStation 3 games consoles, and gets the same power, indefinitely, for the grand total of $3,200. And the machines were donated by Sony, so it cost him nothing. With a headline you’d expect to see in The Onion, Wired has more on the story: Astrophysicist Replaces Supercomputer with Eight PlayStation 3s.

There is, of course, a reason Khanna doesn’t use, say, Microsoft Xbox 360 gaming consoles. Nice machines are useless if you can’t hack ‘em. Khanna runs Linux on the PS3 consoles, and discovered something remarkable about the Cell processors at their core:

Overall, a single PS3 performs better than the highest-end desktops available and compares to as many as 25 nodes of an IBM Blue Gene supercomputer. And there is still tremendous scope left for extracting more performance through further optimization.

Khanna uses the parallel computing power of the eight-unit PS3 cluster to research fascinating but very complex arcana such as the gravity waves produced when two black holes merge.

I think this story makes a fairly good point about the notion of governments and their capabilities. Often, when I argue that the private sector is inherently better suited to provide products and services than the public sector, I hear examples of government innovation and research that the private sector didn’t, couldn’t or wouldn’t engage in. And to be sure, such innovations exist, especially when the research is directed to the needs of tax collection or the defence sector.

In response, I point out that a government is worse at innovation not because it is inherently less capable, innovative, or efficient than any one company (though that is often the case). It’s because a government funds one project, to find one solution, and when it fails, can only really repeat this process in series. If something succeeds, the innovation stops. By contrast, the private sector consists of multiple actors committing risk capital to trying multiple solutions in parallel. They all compete with each other on performance, features, cost, and time-to-market. So not only are the chances of rapid success that much higher, but even those that succeed face competition from other, possibly better, solutions.

When a government does develop a solution, which it often does by excluding potential private sector competitors either just by tapping the almost bottomless well of taxpayer funds for capital, or by explicitly legislating to forbid competition, who’s to say whether the solution it comes up with is actually cost-effective? There’s usually no market to benchmark it against. Until someone like Dr Khanna comes along.

Is it any wonder, then, that the government, as a large end-user with diverse and advanced needs, does sometimes manage to innovate in its own interests, to meet its own needs, but that even when it succeeds, those innovations are soon made obsolete by the private sector? That the government developed the first electronic computers, but that it took IBM to turn the super-expensive behemoths into efficient computing machines for general use? That the government developed the embryo of the internet, but that it took thousands of private sector companies to bring its costs to within reach of ordinary consumers, and develop its full utility by developing new features, better usability and compelling content?

Is it any wonder that the US government runs expensive national supercomputer facilities, but it takes one researcher with a begging bowl and a brain to build a dead simple, dirt cheap lab hack that thrashes them all?

So next time your kid asks you for a PS3, know that he could deploy powerful tools to investigate the ripples binary black hole systems create in the fabric of space-time. Literally.