Quantum codes, deep space and utter geek delight

Heard of Viterbi? There are two right answers: Andrew Viterbi, co-founder of Qualcomm and the inventor of Viterbi decoding, and the coding itself. He was part of a team whose mythology encompasses the first US satellite in space and the invention of the basic tools of digital radio communication. One of the first and most dramatic uses of Viterbi's ideas was their use on the Voyager II mission - as the probe coasted between Saturn and Uranus, the support team rebuilt its communications system to use Viterbi - and got back what remain the high points of the first wave of robot exploration, those superb pictures of the outer planets and their moons.

But I digress. Viterbi decoding, in conjunction with Reed-Solomon encoding, is one of the most undersung inventions of the 20th century. It's an exquisite mixture of logic, mathematics and engineering. I understood it once, for about three minutes.

Which is about two minutes and fifty-five seconds longer than I ever understood the ideas behind quantum communication. You've got this thing called a quantum bit, or a qubit, which is one and zero at the same time. It can be entangled with another qubit - so they both know what state the other's in - and then forcibly separated, but the entanglement is maintained until someone peeks - when it disappears instantly. That way, you know if someone's peeking - and can trust data that you know you're the first to unwrap.

But there are lots of things that can go wrong. Not least, quantum states are fragile things and can be disturbed by something as insubstantial as a puff of probability. The quantum world is a noisy world (which is why most quantum computing takes place in extremely cold devices, where much noise is hushed), and that places extreme limitations on what you can do with it outside the clean lines of mathematical formulas.

Until, it turns out, you add Viterbi. Viterbi works by observing the damage done to a noisy signal and calculating the most probable original state of that signal before it got damaged. Normally you just want the information in the signal - the details of what happened to it in transit are rarely important. But when you combine what Viterbi's found out about the channel with the maths of quantum communication, you end up with a much stronger, more reliable link.

There's something rather heartening in the way that ideas created for the practicalities of the early 1960s - when integrated circuits were as complex as the average transistor radio and cost rather more - seem to map so well to stuff at the very edge of what we can do in century 21. Think we're getting something right.


(PS - I did read a very good analogy of why entanglement doesn't break causality. Imagine you're two light years away from your friend Bernie, Exactly halfway between you is Charles, who has two balls - one black, one white -- and two light-speed spacecraft that can travel exactly at c. Charles puts one ball in one spacecraft, one in the other. He sends one spacecraft to Bernie, and one to you. A year after he does this, you receive your spacecraft, Bernie his. The instant you look at your ball, you'll know what colour Bernie has - despite the fact that you're two light years apart and only a year has passed since the balls were assigned. It looks like information has passed at twice the speed of light - but nothing has gone faster than c. Einstein is happy, even if Bernie probably thought it a wasted year. )