We, the geeks, are nothing if not solution oriented. We love solving problems, and we have a fabulous digital toolkit to build with.
We are a culture of Hammer-makers, and, by golly, the world is full of nails.
But sometimes it is just embarrassing to see our people rush in to “solve” the problems of the world, even if they are the wrong problems.
For example, this spring researchers at the École Polytechnique Fédérale de Lausanne (EPFL) report a handy app that analyses digital video and determines whether people are maintaining a safe social distance . This app is touted as useful to fight Covid-19, especially in public spaces such as transportation hubs.
However technically clever—and it is a neat bit of kit—this app seems misguided to me.
Whether or not social distancing is needed at this time isn’t the point. The point is, it is solving the wrong problem.
First of all, maintaining social distance isn’t difficult. People have easily managed to do so with very simple methods. (For example, if you can touch someone with outstretched arms, you are too close. Don’t need an app for that.)
Second, maintaining social distance is mainly a behavioral issue. Anyone and everyone can do it if they want to. It’s generally not a matter of information, and it’s really not something that can be enforced very easily by third parties. Even if the airport police had this information, what would they do with it?
But the thing that bothers me most is that the app is solving the wrong problem. The problem is not how to keep people visibly distant from each other. The problem is to keep people from breathing each other’s breath too much. What matters for this problem is good air circulation; social distancing is just a rough heuristic of “keeping our air apart”.
And I’ll note that the same journal alone has published a half dozen or more projects in the last year, all using digital technology to detect social distance in real time. I’m sure other journals have published even more examples.
It is safe to say that we have the ability to solve this problem! Unfortunately, it’s not really the key problem, is it.
What we really need an app to do, and this would be really cool, is to accurately visualize the flow of our breath, so we can stay away from dangerous air. This is a difficult problem to measure and model. It involves the air flow in a space and the behavior of the people (e.g., masks, level of activity). The physics and phsyiology is difficult and the data we need isn’t easily measured in imagery or with simple sensors.
I have seen complicated visualizations based on simulations that show this, e.g., for an airplane cabin. But I’d love to have a real time visualization for any space.
To be fair, the EPFL app is technically quite neat. They deploy technology developed for autonomous vehicle navigation to rapidly infer 3D scene from imagery. Determining the distance between people is actually a lot simpler than trying to drive a car in a city, so the tech works pretty well.
They also are well aware of the real problems here, and the app tries to be somewhat sophisticated about behaviors. If you read carefully, it is clear they are well aware that this app is only a small step, useful only in certain situations.
Honestly, I understand that this is mainly a public relations exercise. Researchers everywhere and their sponsors want to be seen to be helping out in a difficult time. As the EPFL researchers comment, “We publicly share the source code towards an open science mission.” (, p. 1)
- L. Bertoni, S. Kreiss, and A. Alahi, Perceiving Humans: From Monocular 3D Localization to Social Distancing. IEEE Transactions on Intelligent Transportation Systems:1-18, 2021. https://ieeexplore.ieee.org/document/9406903
- Sandrine Perroud, 3D detectors measure social distancing to help fight Covid-19, in EPFL – News, May 7, 2021. https://actu.epfl.ch/news/3d-detectors-measure-social-distancing-to-help-fig/