I was asked by JASON to look into adaptive optics about 20 years ago and at that time it was unclassified, it was an open technology which had been developed by Itek, an industrial firm. The idea is to correct the distortions produced in an optical image by the atmosphere. When you look at the sky through an ordinary telescope you don't get a sharp picture because the atmosphere is turbulent and the turbulent eddies in the atmosphere distort the waves, so you always get a certain smearing of the image on the eyepiece of the telescope. So the best possible optical resolution from the ground is something like half a second of arc or so, which is the smearing produced by even a quiet atmosphere. So the idea was that you could have a very fast acting optical sensor system which could actually sense the distortions of the wave fronts as they come down to the ground, and do a very, very fast servo system to correct a flexible mirror, which would then bend the mirror in such a way as to exactly compensate for the refraction, the variable refraction, in the atmosphere. You need to work at a rate of about a KHz, that is, you need to make about a thousand corrections per second in order to have good performance, and so you can actually design a system to do that, and Itek already had such a system 20 years ago, and the question was: What can you do with it? And we were asked in by the Defense Department to look into this and to see what the future of this technology would be. And so I set myself the question of... what are the real limits? If you look at active optics as a information processing problem, how well can you do in principle? And the limits are essentially the balance between atmospheric noise and photon noise. The light itself which is coming through consists of photons, so the information that the light carries is limited by the number of photons, and if only... if you have to do everything a thousand times a second, photons are very precious, you don't have enough photons to get very high signal-to-noise ratios, so that the photon noise is actually very important. And so I did a... I worked out the general theory of this, how you make an optimal system, given the... given the telescope and given the detecting equipment, so you know what you can actually measure; then how do you organise the optimal feedback circuit in order to produce the optimal correction? I actually solved that problem, which turned out to be, again, a beautiful piece of mathematics. And it connected with another of my hobbies, the inverse scattering problem in quantum mechanics, which is an old problem to which I was introduced by Res Jost, another of my physicist friends. So I had this background in the inverse scattering problem which is another very beautiful mathematical theory, which is reconstructing a quantum mechanical potential knowing the scattering that it produces. And it turns out the mathematics for that was identical with the mathematics for the active optics correction system, although it's not at all obvious why they should be the same: one's quantum mechanics, the other's classical - but anyway, they turned out to be the same. So I had this theoretical solution which I published then in the Journal of the Optical Society and it was a good piece of academic science. And it said that if you looked at it from a practical point of view, it said that adaptive optics only works well on astronomically bright sources. You should have... if you're thinking of focusing on a object in the sky, it has to be magnitude 14 or brighter, and magnitude 14 of course is way below what you can see with the naked eye, but from an astronomer's point of view it's still a fairly bright object. Most astronomical objects are fainter than that, so that this was of limited use for astronomy. On the other hand, it was very good for the military because they're looking at satellites which are floating by, illuminated by the sun, and so they're much brighter than magnitude 14, so it should work beautifully as far as looking at satellites coming by, wanting to see clearly what the Russian satellites are made of and so on. That was the object that the military had in view and, so we said, 'That's okay, that works.' But at that stage, after we said that, the military very unwisely declared the whole thing to be secret and that was a disaster for the whole field, so everything went into a black hole and disappeared. So I was in contact with the astronomers on the outside and with the military on the inside, but it was very difficult to... to do anything either inside or outside that made sense. The people on the inside were not interested really in the science at all, they just wanted to have an operational system; they didn't want to know anything more about the atmosphere; and... the people on the outside were totally turned off because they said, 'Well, since the military is spending billions and billions on this, it's pointless for us to spend our little hard-earned astronomical money on trying to compete.' So the astronomers stopped doing it because it was classified, and the people in the military stopped doing anything interesting because they weren't interested in science. So it's a typical example of what classification does. I mean, essentially it stops progress both inside and outside.