German Corona warning app starts: "Every installation helps"

The Corona warning app for Germany is ready and ready for installation. Henning Tillmann, self-employed software developer and co-chair of the digital association D64, explains what makes the app special, why it took so long to develop and whether it can be installed without hesitation, in an interview with ntv.de.

ntv.de: The Corona warning app for Germany was originally announced for April. It is only finished now. Is she late?

Henning Tillmann: Based on the promises made in April, it is too late. But they weren't realistic either. Overall, mid-June is certainly not too late. Because what was built there is anything but trivial. It is not an easy solution. Bluetooth is used for a certain purpose, that happens for the first time. Developing something like this takes time if it is to be good.

What is special about how Bluetooth technology is used here?

Bluetooth cannot actually measure distances, it is used, for example, to connect to headphones. Everyone knows this: If you move too far away from your smartphone, the signal will become weaker until the connection is broken. You want to use that here now. They say: If the signal strength is 100 or 80 percent, then you are relatively close to each other. If it is only 20 percent, you are probably relatively far apart.

Doesn't sound too complicated at first. What problems do developers have to deal with?

It's all about accuracy. Bluetooth works outdoors, much better outdoors. The signal strength can be very high, even though you are five or six meters apart and there is no risk of infection. An example: You are sitting on a park bench and someone is sitting on the bench across the street. You can't actually get infected, but the signal strength can be high enough. It may even be the case that thin walls do not shield sufficiently and the signal strength between two rooms is still good, even though you have no physical contact at all.

Does the built-in hardware also play a role, i.e. the technical quality of the Bluetooth components?

The Bluetooth components are very important for the signal strength measurement. There are clear differences between expensive and cheap smartphones. This has to be correctly calibrated and adjusted at the operating system level. Google and Apple take care of that, and that's a good thing, because they know their devices best. However, there will probably not be 100 percent accuracy. How well the distance measurement works remains to be seen. Fortunately, the Bluetooth signal strength is not the only factor in risk calculation. In addition, there is the length of time of the contact and the time of infection of the contact person.

Apple and Google play a crucial role in app development. Only its programming interfaces enable the app to operate as it is now.

Exactly. Corona apps in other countries often don't work very well. For example in Australia: The iPhone app has to run in the foreground permanently – this cannot work properly. This is not user-friendly at all. This means, for example, that you cannot close the app or switch to another app. These restrictions have only lifted the newly created interfaces from Google and Apple. Apps that do not use these interfaces can only fail. So it was important to wait for it. Germany decided late, but fortunately it chose the right path.

The right way also means that user data is only stored locally. Why is this approach so important?

There are three good reasons for this. First, data protection. Decentralized matching ensures that there is not so much data on a server. The second is data economy, also in terms of data security. A central server on which all data is stored can quickly become the target of the attack. And the third argument is simply the technical feasibility. This can only work properly with Apple's and Google's interfaces. If one had continued to pursue the central approach, one could not have used these interfaces. You would have had to access Bluetooth in a different way, and you can see that that doesn't work in other countries.

Are all privacy concerns cleared now?

It is correct and important that the app is completely open source. Everyone can look at the source code, and many experts are doing that too. This transparency and openness are of fundamental importance in order to create trust. Currently I would say: the app does no harm. Whoever installs them need not be afraid of being spied on.

How many people have to install the app so that it also works?

It would be optimal to use 60 percent or more of the population. Of course that is very, very much. The app can also be useful for fewer installations. For example, if ten people meet, the infected person installed the app and another person. That can help to discover clusters. Every installation helps.

What are the obstacles to a massive installation?

On the one hand, there is misinformation, warnings about the surveillance state. You can only refute this with complete transparency. The second point: You have to be active yourself and install the app. There will surely be advertising campaigns for it. But it will also be the case that the operating system manufacturers Apple and Google will point out regional apps with a message: "There is a corona tracking app for your region. Would you like to install it?" In addition, both date the supported operating systems back to 2015. Anyone who has bought a device from 2015 will most likely be able to install the app.

Another hurdle is user participation. Infected people must enter their infection in the app themselves.

Of course, the app stands and falls with the users. But this voluntariness is very important for acceptance. You have to work with openness, transparency and trust, and make it clear that nobody experiences disadvantages if he or she indicates their infection in the app. This is currently also ensured. Of course, you do not know what will happen in practice over the next few weeks and months. You have to be careful and maybe take countermeasures. If it should develop differently, this must be prevented. It is only possible with absolute voluntariness.

Are important functions missing that could make the app even better?

You have to weigh it up. There are certainly epidemiological reasons to collect more data, such as GPS location data, as is done in China. That can certainly help, but the question is: what is the price we want to pay for it? I am in favor of the transparent, open, data protection-friendly and data-saving way. Incidentally, there is currently no other way. Google and Apple say in their guidelines for the corona tracking apps: there can only be one app per country, and this app must not access the address book or collect GPS data. It's a good thing. Only the exchange of Bluetooth identifiers is permitted, which consist of randomly generated strings.

If everyone installs the app, can we go back to everyday life before the Corona crisis?

Tracing is elementary in the fight against the pandemic, the detection of infection clusters. The app can help, but how much, that remains to be seen. Because a technology is misused here, no one can give a reliable assessment. If it doesn't work through the app, we need more health office staff. But that everything automatically becomes more relaxed once the app is there is not true, because you put a flea in people's ears. I would say: we speak again in late summer or autumn and can then draw an interim conclusion on how well it all works. However, the app cannot do any harm, we should install and use it.

Johannes Wallat spoke with Henning Tillmann