Ten million years ago a star died silently. Her core imploded, a shockwave sending the outer layers into space. What was left was compressed into a black hole. The light from that explosion has traveled through the cosmos ever since – until tonight, ten million light years away, it lands on a telescope mirror in Chile. Minutes later, the same image appears on your phone screen. You are the first person on earth to see it – and record it.
This is possible with the new Black Hole Finder app: a so-called citizen science project (citizen science), in which everyone can help in the search for so-called transients: cosmic outbursts that become visible as a sudden point of light on a telescope image. Think of colliding stars and supernovae – exploding massive stars. Some such events can eventually lead to black holes. No flashy ones artist’s impressionsbut raw data – a spot of light here, a noise pattern there. Just as astronomers themselves see it.
Black Hole Finder was invented by astronomer Peter Jonker of Radboud University in Nijmegen. Jonker specializes in detecting and interpreting transients. The app links users directly to BlackGEM, a number of Dutch telescopes in Chile that continuously scan the sky for such signals.
Every observation is sent directly to the app. As a user, you receive three images of the same piece of sky: the new image, an older image and a difference image in which only changes are visible. So a puzzle picture, where the third image already gives away the solution. If a white spot appears, the question is: is this a real star explosion, or something else – a fault in the detector, a cosmic particle hitting the detector or a passing satellite? Give your opinion with a few taps on the screen. Astronomers will only determine what exactly the speck is later, after follow-up observations. But by allowing hundreds of people to watch at the same time, real astronomical phenomena are filtered out of the noise more quickly.
The app shows three images: the new recording, an older recording and a difference image.
You may think: nice occupational therapy for astronomy fans, but isn’t this a typical job for AI? Jonker sees it differently. “What humans remain good at, and I don’t know if AI will completely take over, is context,” he says. A dot of light on an image is not just a dot. Is it next to a galaxy? Then there is a good chance that it is part of that, and therefore interesting enough for follow-up observations. “A human can immediately see such associations. You can train AI for it, but that is difficult and possibly expensive.”
In addition to that context, reaction speed also plays a role. The images are received almost immediately after the observation. For some outbursts, every minute counts: you want to get there quickly before the signal fades. “We don’t have astronomers ready all over the world when it is night in Chile,” says Jonker. “You can bridge that with the app: someone is always awake somewhere who can watch.” This creates a continuous stream of eyes, spread across the world.
The app was built by the Dutch company Pocket Science, which specializes in citizen science. The principle is simple: many people together provide an enormous amount of observations. This creates a measuring network that is not feasible with a limited number of instruments.
Surprisingly hard science
This sometimes produces surprisingly hard science. With the iSPEX app, thousands of Dutch people mapped air pollution with their smartphones. The results were in good agreement with satellite measurements and professional measuring stations, and also achieved a much finer resolution.
In astronomy, a science that relies on observations, citizen science has long proven itself. At Galaxy Zoo, volunteers classified vast numbers of galaxies, helping to reveal patterns. In Planet Hunters, users discovered exoplanets that computers had missed. In Loss of the Night, tens of thousands of observations showed that stars disappear from view faster than thought – a result that in a Sciencepublication ended up. Black Hole Finder fits in that list.
Before users see anything in the app, a lot of work has already been done. PhD candidate Daniëlle Pieterse from Jonker’s group developed software to filter foreground objects from the data stream. These are mainly nearby asteroids that have nothing to do with star explosions. “You don’t want the app to fill up with flying rocks from the solar system,” she says. Her algorithm links observations to known orbits of space rocks and removes those signals before they appear in the app. What remains is a cleaner selection, with more chance of real events.
Astronomer Steven Bloemen, also from Radboud University, is project manager of BlackGEM. Side by side with Jonker, he developed the ‘astronomy back end’ of the app. “Transients have already been picked up via the app before we had seen them ourselves,” he says. In a field where minutes can make a difference, that is not a detail.
You can automate some things, but that is not always necessary – participating adds something
That speed is crucial because the app is not the end point, but the start of a chain. Once a promising signal is identified, follow-up observations follow with larger telescopes, often in multiple wavelengths simultaneously. “You have to be there on time,” says Bloemen. “Otherwise you can no longer take those spectra and you will therefore never know what it was. An important motivation is to detect the light associated with gravitational waves – ripples in spacetime that arise from collisions of compact stars.”
Bloemen is already looking at the next step. Artificial intelligence already plays a role in pre-selecting images, and will likely take over even more work in the future. But that does not mean that humans disappear. On the contrary, he says. The role shifts. Where a model can quickly classify, a user can help with doubts, context, or deciding whether something is worth follow-up observations. And if that too becomes automated, a new task will arise. “You will always be left with something that requires human interpretation,” is his expectation.
Even if AI can do this just as well in the long term, the question remains whether you should want that. “You can remove leaves with a leaf blower,” says Jonker, “but it can also be done with a rake. With a rake you work quieter and more efficiently, you see better what you are doing and can make adjustments where necessary. You can automate some things, but that is not always necessary – participating adds something.”
For Jonker, Black Hole Finder provides an important by-catch: involvement. “Our team wants to show how the scientific method works, and that it is a joint effort. Apart from the scientific harvest, that is an important motivation for us.”
“You also have to feed the mind,” he says. “People don’t live on bread alone.” Astronomy offers room for wonder – the realization that you are looking at something that is happening far outside your own world, and yet suddenly appears on your screen. “That amazement,” says Jonker, “I want to share.”
