So there I was reading a jolly little "silly season" BBC Online story about the man who says he’s cracked a core problem with the Rubik’s Cube – when I came upon a sentence that slightly chilled me, and also makes me wonder yet again what the hell Google is up to: "Google stepped forward and offered to run the computation… We still don’t know what machinery they used."
Whatever array the search behemoth did use – it was powerful enough to collapse 56 million sets of 20 billion combinations of Rubik configurations into a form where calculations could be performed to satisfy a hypothesis that of no less than 100 million starting positions of a possible 43 quintillion (i.e. has 18 zeros) any can be solved in exactly 20 moves.
Or you, me and your CEO would say… blimey.
I am deliberately throwing in all these massive numbers to underline the power of the Google computational resource. Supercomputers come and go, as we know, but Google seems to be adding resources to an already impressive amount of chained computers to get ever deeper compute-power.
If I was a bit daft or paranoid, I’d now mutter darkly about a Singularity event or suggest we all bomb the nearest Google data centre lest the computers take over the nuclear bombs like they do in Terminator movies and wipe us all out.
The idea here is that at some point computers will become so complex and self-aware that they will leapfrog us in an order of magnitude of brain power and then the question is, would they keep us around as pets because we look cute or would they crush us like bugs as irrelevant to their Galactic consciousnesses?
Instead, I do feel that some really interesting basic computer science problems may now finally be crackable, if we continue down this path. Today we have also had what could be a major contribution to the P/NP problem. If your undergrad data structures and algorithms courses are not that fresh in your mind, the issue is basically if there are classes of problems computers just can’t ever crack (the Travelling Salesman case) other than by well, cheating (a bit like we humans do) (though we also benefit from embedded, very very fast biological computing in things like vision).
If an HP researcher is right – and he hopes he is, obviously, as he gets a million dollars maths prize if so – then the answer is yes, or rather no, P (polynomial time) does not equal NP (non-polynomial). I think; I had best stop here or risk 20 angry emails from Babylonian-bearded mathematicians, who I refer with haste to this site. (By the way, chappie may not get his $1m as other boffins don’t buy his solution yet.)
Probably time to get back to Earth here, though I don’t see any harm in now and again thinking blue-sky and long-term about the basic point of all this IT we are growing. Computers are getting faster, smarter, more powerful – which has to be good for business, society and one does rather hope – the species.