His criticisms are still perfectly accurate. What this underscores is that for some things, sort of working is often good enough.

Adaptive optics needs a wavefront reference, usually a bright star right behind what you're looking at. Only 1% of the sky has adequate references. There are ways to put a fake one up (shine something up that will reflect off of a known depth in the atmosphere, and look at the reflection), but those do not work as well as natural ones. Besides, the technique introduces errors that need to be corrected (in visible wavelengths they can be corrected for for about 50% of the sky, in infrared you can correct for 90% of the sky).

Secondly it is frequency dependent. If you're looking at a range of wavelengths, the one that you're using for the guide will be dead on (at least very close to the wavefront reference) - and the rest will smear. The cause is the same thing that causes a prism to scatter a rainbow - different frequencies do different things in the atmosphere.

Thirdly you can't predict the turbulence. Although this point of his is really an explanation for the first, you still can't predict the turbulence. You can merely calculate how fast it is likely to change, and then you know how quickly you need to measure and adapt. The wavefront reference is what you measure from, and the speed of adaption needs to be on the order of 0.001 s.

For a pinpoint image of something like a planet, this is great. For squeezing extra performance out of a telescope this is also great. The technique is definitely worthwhile.

However Ross is dead right. If you could take the same telescope and put it above the atmosphere, you'd get far better results than you every could with adaptive optics. Furthermore you can get those results in many wavelengths at once, over a wide field of view.

Take the Hubble. It has a 2.4m mirror. The Keck telescope has a 10m mirror. From straight optical principles you'd expect the Keck telescope to be able to see things that are 16x the detail that the Hubble can see. (That is images whose height and width are each 1/4 of what Hubble can make out.) But no. Instead we are astounded that the Keck can focus on a single point in the sky with results that are somewhat better than the Hubble, in one wavelength at a time.

I don't mean to say that it isn't an impressive accomplishment. It is. It isn't simple and it takes a lot of work. There are applications where it can be useful. But just imagine what the Keck telescope could do if we could just place it where the Hubble is! Conversely things like the full color shots of nebulae that the Hubble is famous for will never be duplicated from the ground. (Unless they build that telescope in Antarctica...)

Cheers,
Ben

Let's get actual figures.

According to [link|http://www.noao.edu/system/tsip/keck_cost.html|http://www.noao.edu/...ip/keck_cost.html], the cost of Keck is $25 million per year counting the telescope depreciating over 20 years, and operational costs. So we're talking $500 million. In the last 2 years they averaged 544 observing nights per year (they have 2 telescopes). For a night of observing that comes out to $47,400/night. If you get 10 hours per night, that is 4,740/hour.

According to [link|http://hubble.nasa.gov/faq.html|http://hubble.nasa.gov/faq.html], Hubble cost $1.5 billion to launch, and $230-250 million/year. It launched in 1990, and will continue until later this decade. So its total cost is about $6 billion or so. That is spread out over about 20 years, so that is $300 million/year, or $34,209/hour.

So anything that the Hubble can do which can be done by the Keck should be done by the Keck instead.

However if the price of space flight can be brought down significantly (and I think that it probably can), then at some point space-based observatories will replace adaptive optics. Of course that point is not now, nor is it in the near future.

And in the meantime there are things that can only be done from space. Two of those things are deep-sky surveys of background radiation, and scans of random locations for very, very distant objects. I mention this because those surveys are of critical importance to cosmology, which ties in to fundamental physics, which is (of course) Danny's interest.

So while adaptive optics can do some things, it is of no use for the things that affect Danny. Worse yet, it replaces some of what space-based observatories are good for, which lessens the chance of getting better space-based observatories up there. So this technology worsens the future of astronomy from his point of view. Which I suspect is a reason that he is so strongly opposed to the technology.

Cheers,
Ben