“In twenty years time, it is likely that most astronomers will never go near a cutting-edge telescope,” says Ray Norris at the Commonwealth Scientific and Industrial Research Organisation in Epping, Australia. So begins a fascinating discussion about the future of humanity’s oldest science.

Norris paints an optimistic picture. For him, the future is filled with automation that will make astronomers’ jobs easier. He says, for example, that in twenty years time: “I expect to be able to click on an object in a paper, and see its image at all wavelengths.” This data will be provided more or less automatically by a new generation of smart telescopes that calibrate and edit data on the fly and then send it to a Virtual Observatory that anybody can access.

The job for astronomers will be to theorise about this data, to look for patterns within it and to see how it explains some problems and creates others. They might then suggest what other data to collect.

That should free up much of their time. Norris says the time not spent fiddling with equitorial mounts and lens cloths will allow them up to better engage with the public who pay their wages.

That’s certainly a reasonable change from what astronomers do today but has Norris gone far enough?

One thing he fails to take into account is the newfound ability of computers to analyse data in ways entirely inaccessible to humans.

Last year, Hod Lipson and pals at Cornell University developed a genetic algorithm capable of sifting through data looking for the laws of physics behind it.

And it seems to work. These guys generated a load of data by tracking the motion of things like simple harmonic oscillators and chaotic double-pendulums. They then set their algorithm loose on the raw data–not the manicured stuff but the warts’n’all measurements.

Their jaw-dropping result is that their algorithm derived Newton’s laws of motion from this data, without outside help. Since then, they’ve been inundated with requests to let their algorithm loose on other data sets. They’ve even set up a website where anybody can try it for themselves.

That’s quite an eye-opener. One problem is that the algorithm doesn’t always throw up well known results like Newton’s laws. And that leaves scientists puzzling over the mathematical relations that it reveals. What do they mean? How should they be interpreted? Are they important?

This should be of more than passing interest to astronomers. As Norris points out, astronomers are in the process of automating their work, to the point where the only task left to them is to analyse the data.

And yet, Lipson’s work at Cornell indicates that even this can be automated too.

What Norris has failed to take into account is what will happen when Lipson’s algorithm, or something like it, is set to work on the corpus of data in the Virtual Observatory.

The likelihood is that these algorithms will become powerful tools for discovering relationships in data that humans would find difficult to extract. That leaves astronomers with the task of puzzling over the results, sometimes understanding them but perhaps more often, not knowing what the newfound relations mean or why they hold.

This is a post singularity-type scenario, in which the machines make discoveries at a rate that humans cannot keep up with.

Of course, astronomers are not the only scientists with this fate in store. But as the ones who have more or less automated their jobs already, they’re likely to be the ones who come up against it first.

It’ll be interesting to watch how they cope. But by then, it’ll be too late to for the other sciences.

Ref: arxiv.org/abs/1009.6027: Next-Generation Astronomy