Archived

Crowdsourcing fundamental physics

Photo: ImageEditor

The Large Hadron Collider (LHC) is an experimental facility of particle physics that spits out data in quantities unprecedented by any scientific experiment. Physicists use these to gain more understanding of the fundamental laws of the Universe, often by searching the data for deviations from what the theory predicts. Disagreement is always interesting: it indicates the limits of validity of the current theory and serves as a guide towards an improved one.

Even in today’s large international collaborations it is absolutely impossible to analyse the vast amount of data by hand. Instead, scientists need to write powerful computer algorithms to do this. But there are some tasks that are hard to teach to a computer, even though they’re easy for most humans to do. Speaking, for example, or recognising visual patterns. One possible solution is crowdsourcing.

Crowdsourcing as a solution

Lets take a look at an example of how crowdsourcing is used to solve a pattern recognition problem, starting with a bit of background information.

The innermost part of the big LHC detectors is called a tracker. It works a bit like a 3D digital camera, recording the tracks of charged particles that fly through it. (Unfortunately, it cannot detect electrically neutral particles, because they don’t interact with its material enough.) A point where several particle tracks meet is called a vertex. Vertices are where the particles leaving those tracks were created, so they mark the places where a collision or the decay of an unstable particle happened.

Many particles (including the famous Higgs boson) are so short-lived that they decay right after the collision in which they were produced, so their decay vertex and the collision vertex lie right on top of each other and can’t be distinguished.

But one of the many things that are being looked for at the LHC are new elementary particles that are sufficiently long-lived to fly for some macroscopic distance before they decay. When they do, they give rise to a secondary vertex. Finding these secondary vertices starting from all the various particle tracks in the event is a difficult thing to teach a computer to do, but humans are relatively good at spotting them.

Hunting for Higgs

The Higgs Hunters project makes use of this fact. In their web browser, volunteers can look at images of collisions that were pre-selected by a computer algorithm and pick whether they can see something that looks like a secondary vertex or not. Full-time scientists may then concentrate their efforts by taking a closer look only at the events picked by the volunteers. A community is forming around the project, with users commenting and discussing the more ambiguous collisions.

In some sense, this way of doing particle physics is going back to the “roots”. Back in the days, in the 1950s to ’70s to be precise, that science was done by analysing actual photographs of tanks of liquid in which charged particles had left traces of tiny bubbles along their path. This job was not done by volunteers at large, but physics students and assistants. Many of the assistants were women, of which there were terribly few in higher physics research positions at the time. Fortunately, humanity has been moving forward since, and largely realised that the ability to do science (or pretty much anything else does!) not depend on a person’s gender. Nowadays more and more women are taking leading positions in science. Just this past year, physicist Fabiola Gianotti became the first woman to be elected head of CERN.

Learning from citizen science

Photo: European Space Agency & NASA

Another cool example of citizen science comes from astrophysics. It included a classification of spiral galaxies into clockwise- and anticlockwise-turning ones. Spiral galaxies have “arms” curling like a spiral around a bright galactic centre. Volunteers looked at telescope photos of such galaxies and marked them according to the direction in which the arms point — again something that was difficult to automate on a computer. To the great surprise of everyone involved, this led to the discovery of a strange psychological bias: participants classified more galaxies as having an anticlockwise than a clockwise orientation, even though upon careful analysis both types turned out to be about equally frequent in the sample ¹. It turns out that the human brain is not great at determining which way something is spinning. This goes to show how we can set out to learn something about the Universe outside and end up learning something about our own minds! By the way, a new version of this project is ongoing.

In my opinion, citizen science is a great thing, because it promotes the enthusiastic public from being consumers of the results of science to being their creators, a bit like Youtube did with broadcast video. It’s not asking people to lend scientists their ears like it used to be, it’s inviting them to take part in the discoveries.

The end of a beginning
In the near future, many of the kinds of relatively simple and well-defined tasks that some citizen science projects involve today will surely be performed by computers. Other tasks will continue to be best performed by people and many new ones will only become accessible to crowdsourcing through advances in computing.

While it is true that this is more aiming at professional scientists, more and more collaborations are even publishing their raw data for anyone in the world to analyse, often along with open-source analysis software. This is what the Planck satellite did, showing beautifully that not even the sky is the limit to the scope of open science.

And although it has been around for many years, I have a feeling that citizen science is only just beginning to take off in terms of both breadth and depth.
If you want to get involved, you could start by checking out the incredibly beautiful projects on Zooniverse, a platform for crowdsourced science projects. Together, let’s make this a more curious, collaborative, and peaceful new year 2015!

^{1. Thanks to Gioacchino Accurso at University College London for pointing me to this.}

Publisher: University of Helsinki / LUMA Centre Finland
ISSN 1799-4780