Across a decade, 73,000 enthusiasts joined forces to classify galaxies in their spare time
The mysteries of outer space have long fascinated scientists and academics, but an innovative research project has put ordinary people at the helm of exploring our cosmos.
To celebrate World Space Week, Professor Bob Nicol and Professor Daniel Thomas joined the Life Solved podcast to talk about Galaxy Zoo, a citizen-science project part-funded by the University of Portsmouth.
What is Galaxy Zoo?
Galaxy Zoo saw more than 73,000 space enthusiasts joining forces to help classify galaxies and contribute to our understanding of the universe. This decade-long research project was born out of a need to accurately and quickly process the vast amounts of imagery created by the Sloan Digital Sky Survey.
Professor Bob Nicol had previously been working on the Sloan project and was proud when the reams of digital camera imaging it generated were archived and made publicly available. That presented an enormous mine of data but also a problem. Daniel explained:
If you say to somebody, look at 50000 images and classify them into just one type only type galaxy and another spiral galaxy, that seems crazy.
And yet, that’s exactly what they were able to do – and more – by creating an online portal where members of the public could register and classify galaxies in their free time.
An overnight success
This joint project with colleagues at Oxford University was featured in a spot on national television news when it launched, only for the sheer number of interactions to crash the server! The public’s unprecedented enthusiasm to take part caused the team to quickly upscale operations, and before long, communities were forming over the exciting work to be done.
Considering all this took place a decade ago before social media communities were a regular part of public life, the Galaxy Zoo phenomenon foretold a revolution for public engagement in science and research.
In fact, Galaxy Zoo quickly led to Zooniverse: now the world’s largest portal for people-powered research. Right now, 2.3 million people are taking part in work that informs breakthroughs and furthers research and ideas across the scientific world. And the subjects are as diverse as their community.
How a community informed artificial intelligence
Galaxy Zoo itself began with a simple tutorial that walked users through the process of classifying a galaxy into either a rugby ball-shaped structure, or a spiral arm structure, something that AI and computer tech of the time was not able to do as reliably as the human brain.
The human brain remains the most powerful computer we can ever dream of having
Nevertheless, in addition to the classifications Galaxy Zoo generated, the process of identification has fed into the development of more sophisticated AI and machine learning to aid with such tasks in future.
So, what about the scientific outputs of the data classification? Well, it was then possible to combine the findings with data from the Hubble project to understand more about where different types of galaxies were located. Their proximities to one another also suggested that galaxies that have collided lose their spiral arm structures, giving insight and ideas into the formation and life-span of our cosmos.
What’s more, communities blossomed, and other strands of analysis spun off this original study, even in the fields of psychology!
More to explore
To Bob and Daniel, many questions remain to be explored, and thanks to the hard data and algorithms generated by the project, the sky’s the limit for the answers we might find.
How you can find out more about citizen science projects
You can find out more about Galaxy Zoo here on the Zooniverse portal.
To listen to Bob and Dan talking about their work, you can listen to the podcast on any app or desktop device, just search 'Life Solved'.
Anna Rose: Welcome to Life Solved. This week, it's World Space Week and to celebrate, we're diving into a citizen science story that puts everyday people at the heart of understanding our cosmos.
Daniel Thomas: It seemed like a good and crazy idea, certainly worth trying, but I think none of us had an idea where it would lead to.
Anna Rose: In this podcast, we share the amazing research that's happening here at the University of Portsmouth, and how it's impacting our lives. But this time, we hear how everyday people, not scientists or academics, but citizens with passion and interest are revolutionising our understanding of galaxies.
Bob Nichol: I think Galaxy Zoo was the first-ever citizen science project that truly went to the normal, ordinary public who, you know, just suddenly engaged in this material and knew they could make a difference.
Anna Rose: What's more, the project we'll hear about today has created insights for citizen-led research across the disciplines and even powered the development of sophisticated artificial intelligence. We're meeting the Portsmouth researchers who joined forces with another university in the public to revolutionise our understanding and analysis of data from outer space. John Worsey spoke to some of the minds behind Galaxy Zoo. Every day, 2.3 million amateur scientists and enthusiasts are combining forces to drive our understanding of diverse subjects forward through this community. The work they do informs papers, classifications and breakthroughs across the scientific world. And its original project was Galaxy Zoo.
Daniel Thomas: I was very much interested in analysing galaxies, looking at their properties and seeing so galaxies that look different, how are they different in their formation histories? And I've been used to working with, you know, tens or hundreds of galaxies at most. And then you can look at them at images or whatever data you have and you can work with those.
Anna Rose: That's Daniel Thomas, Professor of Astrophysics and Head of the School of Mathematics and Physics here at the University of Portsmouth. Two decades ago, he was wondering how to classify vast amounts of imagery from outer space in order to analyse and better understand galaxies. An enormous turning point in our exploration of the cosmos came in 1992 when the Sloan Digital Sky Survey launched. This revolutionary project used digital camera imaging to map the night sky in incredible detail. Professor Bob Nicol is today the Pro Vice-Chancellor of Research and Innovation here at Portsmouth. At the time, he was working with Sloan. Bob explained why the huge amount of data generated by Sloan wasn't the only cutting edge thing about it.
Bob Nichol: I'm very proud that my colleagues within Sloan said, we're going to make this data available to everyone, and they put it on digital archives that anybody could access. And that was quite revolutionary in astronomy at the time. You kind of held onto your data. Sloan was the first time we said, we've got too much of this data, we can't look at it all, we should give it back to the public. And we did. We gave it to everybody.
Daniel Thomas: At the time, that seems a crazy thing to do. If you say to somebody, look at 50,000 images and classify them into just one type only type galaxy and another spiral galaxy. That seems crazy, but we did it. There was another colleague, Chris Lintott in Oxford at the time as well, and at some point, we were chatting about this and we're saying, why not ask the public to help? If we want to look at hundreds of thousands of images that something like SDSS the Sloan project provides us with. We can't do this alone.
Anna Rose: Bob and Daniel joined forces with colleagues from the University of Oxford to workshop an idea that put the public at the heart of space science.
Daniel Thomas: And that's how this very early team formed between Chris and Kevin in Oxford, myself and Bob in Portsmouth. We started back in 2007 having these discussions, how can we actually do this? And that's also the time the Galaxy Zoo came about, which then evolved into Zooniverse.
Bob Nichol: The first I heard of it, so when I was brought on, Daniel and Chris were talking to me in Patrick Moore's Garden.
John Worsey: Oh, lovely.
Bob Nichol: So that's the first I heard of it, and it was sort of an event there. And I remember at the time, you know, they said, well, why don't we just get the public to look at it for us? And it was like, it was one of those light bulb moments. It was like that, wow, yeah, why not?
Anna Rose: Patrick Moore's garden, eh? The presence of astronomer royalty put a rocket under the idea. Galaxy Zoo launched with a humble website, anticipating years and years of volunteer data crunching ahead. But none of the team were prepared for what happened next.
Daniel Thomas: The website was publicised on BBC Six O'Clock News and just when it went live, and that clearly helped to spread the word and get it out. And I remember how we were then in the evening hectically on the phone in crisis sitting because the server crashed. So we had so many people interested in it that actually the server that we had set up couldn't even deal with the demand. I think we had 10,000 clicks straight away, quickly going into 100,000. And none of us had anticipated this and it was a it was a very exciting time, really like so many clicks. How is that possible? And then, of course, then we knew we had to step up our game and to very quickly launch the system into something that can deal with the demand and we immediately understood. Wow, we really hit a gold mine here, and that's something that really can grow to something big. And of course, it did.
Bob Nichol: I think we did underestimate the desire of the public to get involved with this type of subject. I mean, or this type of project. I mean, I think it was the combination of sort of a digital project you could do in your spare time while looking at some wonderful things in the universe. And I don't think we truly understood that this was going to go as fast and as crazy as it did. So, as Daniel said, we were not in the first couple of days ready for the interest.
John Worsey: The numbers are staggering. Seventy thousand classifications in one hour on day 2. Eight million classifications submitted in the first ten days and in the first year you got 100,000 volunteers expecting more than 900,000 galaxies. It's phenomenal. I mean, has citizen science ever been done on that scale before? Or is this really a kind of a step forward in the whole notion of citizen science?
Daniel Thomas: Yeah, I think it really was. It's the new era in citizen science clearly. I mean, I think you had these city projects when people from home were contributing in search of extraterrestrial life, but it was much low key. I think it was much more based on also spreading wide, you know, across the globe – some people here and there clicking somewhere. Galaxy Zoo was a completely different pace. It was, obviously, it was not actually confined to the U.K. it was international. But of course, there were lots of users in the U.K. and just contributing enormously to it. And then, of course, if you have 10,000 users with this rate of classification, you could use loads of classification straight away. It was absolutely amazing.
Anna Rose: Although space projects had sought to use citizen input for science before, Galaxy Zoo took things to a whole new level, stretching across international boundaries to the final frontier. So what was it that drove so many members of the public to play their part?
Bob Nichol: Whenever we've done surveys, whenever we've gone and asked the Zooniverse or the Galaxy Zoo public and said, Why are you doing this? The number one reason is to make a difference. You know, they really want to engage in the science and make a difference to science. And I think that surprised us as well. You know, there is a lot of people out there that will give their time to advance science. And I think that's one of the lovely stories to come out of Galaxy Zoo.
John Worsey: So given that this is something that someone can log on to the website and they can get classifying these images from a point of really zero-knowledge, how does that actually work? What is it that they're presented with? How do you explain what you want them to do?
Daniel Thomas: We have a tutorial, so obviously, that has evolved from early days in Galaxy Zoo. It was relatively simple. We took care of explaining well in simple terms with example pictures and exactly what they have to do. It became a little more complex later on. But at the beginning, we're really, really simple. Actually, the very first question when we did Galaxy Zoo, it was only just is this, you know, a round-ish galaxy or does it have spiral arms? And it's just that sometimes the images, you know, sometimes it was a bit tricky to see. So you need to look at it carefully. You're shown a pattern, and all you have to do is to recognise it. And I think that's what Bob said, that obviously, people learnt about it so in that way they became experts and they even read about it afterwards. So starting off from very simple pattern recognition, in fact, they started to become, you know, astrophysicists. What was also amazing I think in this context was the users developed their own platform. So we had then provided a forum and they started to discuss their images. So they completely almost, I don't want to say Decoupled, but they created their own space where they were actually having discussions semi-scientific about what's going on there? And there were quite a few interesting objects also that were discovered that way that looked strange. And we wrote a proper science paper on things that, you know, armchair astronomers have discovered from their living room. So that was really amazing.
John Worsey: That's brilliant.
Bob Nichol: People started researching on their own. They went off and started reading stuff. As Daniel said, they made up their own forum and they started talking to each other. I mean, it was amazing to watch this so that suddenly this whole community sort of blossomed in front of us.
Anna Rose: In the pre-social media age, space provided armchair astronomers and academics alike with a common language and purpose, and Galaxy Zoo continues to have a vibrant social media community today. It's @ Galaxy Zoo, by the way. Today, you might think a computer or A.I. programme would be better adapted to the sorts of tasks involved in classifying the shapes of galaxies. But when Galaxy Zoo launched, this was far from the case, Bob explained.
John Worsey: Computer-based algorithms struggle to visually inspect digital images of galaxies as accurately as people. Why is that?
Bob Nichol: It's tough. It's hard. You know, if I was to show you a picture of an elliptical, it would look like a rugby ball. And if I was to show you a picture of a spiral arm or spiral galaxy, it would look like a fried egg with lines drawn on the yolk of the egg. Now today, you know, you could probably give this to DeepMind in Google, and it would figure it out. But, you know, 20 years ago, it couldn't. I mean, the processing power, we didn't have the processing power. We didn't have the algorithms. We have to remember what life was like before everybody had computing power at their fingertips. And you know, that was – we were dealing with the lack of resource to do this using the machine, and the algorithms at the time you wouldn't trust to do it anyway. This to us as a human is a trivial – does it have a spiral arm? Does it not? Does it look like a rugby ball shape or does it not? Asking a computer to do that to the same accuracy as you and I could visually do it was impossible back in 20 years ago.
Daniel Thomas: There's a lot of artificial intelligence and approaches that have been also been formed by Galaxy Zoo classification. So then you start to make those communicate. But one of the reasons why Galaxy Zoo is so powerful is this idea of using the human brain, of course, the human brain remains the most powerful computer we can ever dream of having. The kind of patterns we can recognise and we can process with our brain is far superior over any kind of computer that we can produce. And the other thing is the computer, as much as artificial intelligence, machine learning is advanced, they can only be as good as you trained them. So they can only recognise what you tell them to recognise. When our brain is recognising things it didn't know before. People found very peculiar objects that triggered scientific applications for lots of really interesting discussion in and outside the scientific community. Machine learning would have never picked it up. It would have put it somewhere and it would have been lost. So I'm really glad we insisted on using the human brain.
Bob Nichol: I think we have helped in having a more mature conversation around the labelling of data that will be used to advance the next generation of machine learning algorithms.
Anna Rose: The insights from Galaxy Zoo not only help generate data for understanding our universe, it also helped with the development of programmes, strategies and more subtle A.I. to fit with the human interface. It's pretty remarkable to think how much technology and programming has progressed in the past couple of decades through this process of human-computer feedback. So when you've classified bundles and bundles of data, what do you do with it?
Bob Nichol: We actually made our Galaxy Zoo data available, and that sparked other generations of people working with our data. One of the very early things and I'm very proud I think Portsmouth's was at the vanguard of this and this is the work Daniel, I and other colleagues at Portsmouth did, is we started saying to ourselves, Well, actually, there's a quite interesting problem here. It had been known before, but the wealth of data that we were given both by the zoo and also by the Sloan was that it was clear there was sort of this almost dichotomy of galaxies. You could either separate them with colour, so you can either say blue, red or you could separate them spiral arm, rugby ball shape. And the general consensus at the time was that those two dichotomies with this roughly the same. So blue spiral arms, red rugby balls. Well, it quite a lot of work that I think both Daniel and I are very proud of that we've done from the Galaxy Zoo has been to demonstrate that while in general, that's right, there is a significant fraction sort of 15 to 20 per cent where that doesn't apply. So you could have things like red spirals and you can have things called blue rugby ball shapes and those, interestingly, I think Daniel would agree, where the Zoo really highlighted and where we've spent quite a lot of our scientific time figuring out now what is colour telling us and why is it different from the morphology of the galaxy? And those are the sorts of questions that are still being asked today. We also in Galaxy Zoo, eventually, we got around to putting Hubble data in. So, you know, the famous Hubble, which gave us the same sort of detail about these galaxies as we had sort of in the nearby universe, we suddenly had all that information about the faraway universe, and there was a huge debate at the time about the emergence of these spiral structures in galaxies. When did these disks appear? When did the spiral structures in the history of the universe, when did they appear? And the work that we've done in that regard clearly shows that as you go back in time, the number of galaxies with these spiral disk-like features disappears. I think that was certainly one of those sort of hidden things where we could see that in the data, and I would say again, we could see it in the data. All right. The humans could see it. And the fact that we could see it as when we combined Hubble with Sloan and with our community, actually, I think that's one of those hidden results that really doesn't get the publicity it deserves.
Daniel Thomas: The other really key addition that Sloan gives us is that because we have so many, we can actually look at questions like where in the universe are these galaxies located? And that actually matters a lot in our quest to understand how galaxies form and evolve. And we call this galaxy environment. So is it a galaxy cluster with many others around? Is it a galaxy isolated. The studies we have made of galaxies have in fact showed things like these roundish galaxies. They tend to sit more in environments where they have lots of other galaxies around them. We are sure it is connected to the fact that galaxies collide. That makes them round, so they lose this spiral arm structure.
Anna Rose: So classifying those enormous amounts of data from the Sloan images and combining them with those of Hubble has given enormous insights into the formation of galaxies and how they interrelate. Galaxy Zoo began to evolve into different strands of analysis as more and more communities formed. Soon enough, the approaches were applied beyond the field of astronomy. Professor Karen Masters was working at Portsmouth at the time and realised that the insights into how people could work together to gather data on mass would be invaluable for other large scale scientific research projects.
Bob Nichol: She was starting to, with colleagues at Oxford, starting to look at how you could take what we've learnt and I think key here, isn't just that we've learnt sort of the mechanics of doing it. So it wasn't just that, oh, we can take a bunch of images of penguins and get someone to click some buttons. It was actually we'd learnt about the psychology of how to engage with a community.
Daniel Thomas: We have some users who are just classifying a huge amount of data across all sorts of disciplines, and that means they're sitting for hours at the computer looking at images that are very similar and looking for the difference. And that is just amazing. It's wonderful how I think we are getting this focus back on people. So that's clearly some kind of change of behaviour, I think, compared to how social media is otherwise used at the moment. It is education. The links on the Zooniverse, the pages button, we have used it a lot in our outreach programmes and we've been going out to schools in particular where you talk about astronomy it is very welcome and kids and adults alike, they're incredibly fascinated by the subject. And of course, we have used Galaxy Zoo a lot there and Zooniverse later to get across our messages to publicise it. We're clearly inspiring young people to go through that and engage with science in that way.
Anna Rose: To Daniel and Bob, the big takeaway from Zooniverse is the way it engenders a spirit of curiosity in so many scientific minds today, and hopefully for many years to come.
Daniel Thomas: I think in terms of researchers behaviour or approach to science, where clearly something like a Galaxy Zoo in Zooniverse that contribute to, is to remind us of the very essence of what we are driven from and what our job is, which is curiosity and interest in detail and complexity. The essence of research is that every answer will trigger more questions. That's the whole point. As a researcher, you should never be satisfied by setting out a quest, having an answer, and then you go home and job done. You will always you always want more, of course, go deeper into.
Bob Nichol: I think, you know, we're in danger of thinking science has the answer to every question. Well, it doesn't. And I think the Zooniverse demonstrates across a whole variety of disciplines there's still stuff to discover. And if you're curious, you can go and look at it and you can go and find something.
Anna Rose: It just goes to show how collaboration between institutions, researchers and enormous amounts of public passion can even today drive forward some of the biggest innovations in science and remind us that we are all still here to ask questions. You can follow Zooniverse and maybe join a project yourself at Zooniverse.org. Thanks for joining us for Life Solved. If you want to find out more about research at the University of Portsmouth, go to the website port.ac.uk. We'll be back next Thursday with another story of how work that's happening here is changing all of our lives for good.
Jac Batey: I mean, many of these items are quite ephemeral in nature. So many of them are kind of would be made and gifted to a friend and maybe catch or may be lost. So it's kind of quite fleeting and quite a lot of stories are quite personal.
Anna Rose: Catch you then.