How to Think about Data Privacy
If you’re working on a project that requires troves of data about people, concerns about violating their privacy loom large. Unfortunately, what privacy is and how to think about it in relation to particular projects is often unclear.
There are three aspects of privacy we need to tease apart.
First, there is the regulatory aspect. You want to make sure that how you collect, handle, and use data is compliant with, for instance, GDPR (the General Data Protection Regulation), which holds in the European Union, and CCPA (the California Consumer Privacy Act).
Second, there is the cybersecurity aspect. This speaks to ensuring that access to the data you use (and generate) is restricted to those who are authorized to access the data.
The third and arguably most under-discussed aspect of privacy is the ethical aspect, and it’s what we’ll focus on here.
It’s common to think about privacy, from an ethical perspective, about being anonymous, or being unable to be seen. If I have data about you but I don’t know it’s about you, or if I’m unable to acquire that data in the first place, then plausibly your privacy is respected. If that’s right, then you know how to develop a product that respects privacy: anonymize the data and/or make (some) data unable to be acquired in the first place.
This is not, however, the best way to think about privacy.
Suppose you’re in your bedroom, you want some privacy, and so you decide to close the door and draw the shades. Then someone you’re expecting knocks on the door and you let them in. Has that person violated your privacy? Of course not. That’s because you willingly gave them access to something that you were unwilling to give to others. Had someone else knocked on the door, you would have denied them entry.
The moral of the story is that privacy is not about a passive state of being: being unknown or inaccessible. People’s right to privacy is about the right to exercise an ability to determine who, if anyone, has access to you. And it is no different with data privacy. Respecting consumer data privacy is about giving people control over data about them: what is collected, how it is handled or shared, and how it is used (e.g. for training ML).
On your next project involving data about people, you should ensure that how you handle and use data is not only compliant with regulations and reasonably robust cybersecurity practices and procedures, but also that your team has thought about what data a) your users of your product and b) those people who are affected by use of that product should have control. You’ll then need to think carefully about the presence/absence of features that will enable them to exercise that control without the product putting an undue burden on them (e.g. by hiding the controls they need to access). The more control you can give them, the greater the extent to which you respect their privacy.