Charles Townes: It presupposes a targeted place, and it presupposes you know what the attenuation is between you and them. That's not so easy, but one could make some approximation to it. I think it can cause a little trouble. I don't think it's a very serious problem for initial efforts, at least. Any further comments? Well, I guess this makes it my turn.

One aspect I would like to bring up is the following: I am primarily working in astronomy now. In astronomy, there is so much interest and so much drive in just finding out things about stars and about planets and so on, that I would certainly expect that within the next two decades that we will have located the stars in our immediate vicinity which have substantial planets. The kind of distances we are talking about are about a hundred or a thousand light years. Presently, systems are being planned which should be able to do that. They're tough, they're difficult; it's beyond what we have done yet, but nevertheless I think they are practical. And so, we will have located planets.

Now the question is; how much further can an intelligent community go? Would they be able to look at the planets and say that "well now, this planet has this kind of an atmosphere, and temperatures" and so on. To what extent could they find out still more information and thereby be very selective in deciding which planets they would be interested in communicating with? I think there is a substantial possibility there and that will progress regardless what SETI does. A lot of information will come in over the next few decades; a great deal of information. That, too, will change the problem for us, and may well have changed the problem for a more advanced civilization because they will know a great deal more when they pick out, let's say, the one star in a thousand that looks really interesting and leave the rest of them alone.

The other point I make at this time is that several people have talked about and asked about this "quantum noise". I'm sure it's clear to some of you, but not everyone, that quantum noise is necessary because of the uncertainty principle. The uncertainly principle implies that you cannot measure both phase of a wave and the number of photons in the wave accurately at the same time. This is why, if you use heterodyne detection which is linear, it reproduces the phase of the wave to some extent, at least. Whenever you heterodyne-detect, you have the possibility of measuring phase and hence you automatically must introduce noise. You are introducing noise in the heterodyne process itself. That gives you a noise, and typically for a normal heterodyne detection, that's one quantum per unit bandwidth per second. Which is a lot of noise as you get to high frequencies. In the radio range, it's relatively negligible.

The way around that is to use photon detection, where you simply let a photon generate an electron in solid such as a photodiode, and you detect a quantum. You then know nothing at all about phase. There is no need to have any uncertainty whether there is a quantum there or not. This is why at the higher frequencies the quantum detection is enormously better. You do not produce the noise in your detection the way that linear detection or heterodyne detection would do. Hence there's no reason to have any background noise. Thereby you can detect single photons, even have very wide bandwidths if you can make the background sufficiently low. The background of a distant star is pretty low and you can also blank out the star to a certain extent. So, one can go a long distance in that direction. There is still the quantum fluctuations in the strength of the signal which people have pointed out. That's there, but simply for detection itself one can generate situations and do work which would then allow almost no background at all. Heterodyne is a great thing for radio astronomy, but it is very hazardous for the optical region.

Barney Oliver: That noise on detection is important too for some extent isn't it, Charlie? If your expectation is much less than a photon per sampling time, or per observation time, then that noise itself is high. You would say "no, there is nothing there" when there really might have been!

Charles Townes: If you were looking for something and you don't find it in one minute, and then three minutes later you find something, then you say "OK, I've detected a signal".

Barney Oliver: In three centuries, it's different! [laughter]

Charles Townes: Sure - even in three centuries in principle, if there is absolutely no background then you'd say "Wait a minute! I've got something! I've got a photon!". You can integrate for a long time and in principle you can integrate over a very large bandwidth if the background really is sufficiently low.

Let's go on to Fred. Would you like to say something, and then we will have general comments and questions?