When Dr Chris Lintott, a researcher in the department of physics at the University of Oxford, first considered launching a website to ask the public to help classify photographs of 1m galaxies, he assumed it would probably take three or four years to complete. Galaxy Zoo (galaxyzoo.org), launched in July 2007, was supposed to be a side project; instead it has turned into the biggest citizen-science experiment on the web.

Galaxies can be classified as spiral, elliptical or merging (when two come together). The Sloan Digital Sky Survey, or SDSS (www.sdss.org), has images of nearly a million galaxies; what those images don't have in their raw form is the information about what class of galaxy is pictured.

Lintott had hoped that each image would get 10 classifications, or "clicks", and that the public would prove able to classify galaxies accurately according to their features. Three weeks and 10m clicks later, he was proved right: the public is capable of classifying galaxies as well as, or even better than, professional astronomers. And quickly, too. "It was like being hit by a tidal wave," says Lintott.

The images have now had more than 70m clicks in total, allowing Lintott and his team to go beyond simply sorting galaxies into spiral, elliptical or merging categories, into further research. "You can have confidence, as we can say, '100% of people think that's a spiral galaxy, so it's really, really spirally'," says Lintott.

Human zoo

His team was allocated precious time on the IRAM 30m radio telescope in southern Spain, and was easily able to select just 40 galaxies from the original sample to study, safe in the knowledge that the scientists had exactly what they needed.

After Galaxy Zoo's initial success, Lintott wanted to take a closer look at merging galaxies. At 5pm one Tuesday he posted a spreadsheet listing a selection to the forums and asked members to take a look and email him the best. Then he went to the pub.

One "Zooite", Richard Proctor, a telecoms consultant, spotted the spreadsheet and thought: "I can build a webpage to do that!" By the time Lintott logged back in, the web interface was already in use.

"What was going to be a quick little study turned into a much larger study of about 45,000 images," says Proctor. Each image has now had more than 25 clicks, with four particularly committed souls having seen every single one. The first academic paper about those merging galaxies has been submitted, and Proctor is listed as a co-author.

It's not the first time collaborative astronomy has proved a hit online. The Search for Extraterrestrial Intelligence (Seti) project created the most powerful distributed computer ever with its Seti@home project launched in May 1999, which used spare processing power on home computers to process data collected from a specific wavelength to see if aliens were trying to get in touch. It was used by 5.2 million people and offered, at its peak, 265 teraflops (a trillion calculations a second) of processing power.

Similarly, Nasa's Clickworkers project in 2000-01, found that ordinary people were just as good as astronomers at identifying craters on Mars in photos.

The biggest influence on Galaxy Zoo, however, is Stardust@home, where users search for tiny interstellar dust particles within images returned by the 2006 Stardust mission. Thousands of users have logged on to the site.

"Stardust@home was a real inspiration for us," says Lintott. "We thought, 'If 20,000 people will look for dust grains in their spare time, surely they will look at beautiful pictures of galaxies?'"

But unlike those projects, where the task is set from the top down, the Galaxy Zoo community has its own ideas about what can be done with the SDSS data.

"Up until now," says Proctor, "the professional astronomers have come up with things that want classifying and we've classified them." But there are thousands of irregular galaxies - neither spiral nor elliptical - that the professionals don't have time to examine. So the community decided to do it themselves, drafting a list of questions and building a web interface to make classification easier.

"Once we've got some results," promises Proctor, "we'll publish them for everybody to use, and then see if we can find out anything useful. Given that the biggest study of irregular galaxies to date looked at around 150, and we've got 9,000, we must find something!"

Lintott is supportive of the irregulars project: "They are doing everything professional astronomers would do," he says. "It's up there with any work I've done."

It's not just the volume of research that can be done by collaborating with the public that's important to Lintott, but also the opening up of scientific research to anyone with a browser and a little time.

"By making the data available to everyone," he says, "whatever stage you are in your learning, you can do research."

And a surprising cross-section of people are doing just that. Hanny van Arkel is a Dutch primary school teacher who discovered a strange object, now called Hanny's Voorwerp (Dutch for "object"), near a spiral galaxy. "I thought it was fun," she says of Galaxy Zoo, "but I didn't expect that it would become such a big part of my life. I've learned a lot and it's a good way of participating in science."

Elisabeth Baeten, a Belgian secretary, is one of only four people who has seen every image in the galaxy merger project. "I also classify the irregulars," she says, "and in between I cruise the universe looking for asteroids and gravitational lenses."

In future, it will be much easier for both scientists and enthusiasts to take part in such projects. Not only are Lintott and his colleagues finalising Galaxy Zoo 2, which will examine galaxies in much more detail, they are also building a platform that will allow any scientist to upload data and tap into the vast potential of the internet.

"Our users are clamouring for stuff to do," says Lintott. "The problem of having too much data to pay close attention to is not just an astronomical problem. It's astronomical in scale, but it's not just us."

The new project will allow scientists to upload their videos of elephants, pictures of galaxies, or chemical structures, then specify what they want done with them. They can either use the provided templates, or customise their project. Lintott describes it as the scientific version of the popular blogging software WordPress.

And there are many uses of such a platform. Nasa has high-quality images from the Mars Reconnaissance Orbiter, most of which are simply filed. And study of the pictures produced by this year's Lunar Reconnaissance Orbiter will allow scientists to build a history of asteroid bombardment of the inner solar system.

"It's like Nasa does the map and we'll write the guide book," says Lintott.

Indeed, it would be possible for anyone to upload data, including amateur astronomers. "More than a million separate observations a year are recorded by amateurs," says Lintott. "That's a huge pile of data that, at the minute, professionals slowly sort through, but we can hand that back to the amateurs to analyse.