Community support needed: We want to use our users personal data

What comes to your mind, when you ask yourself why people use LibreOffice? A “stunning user experience” is likely not in the top list. And unfortunately we share this problem with a lot of user facing Free Software.

And there is a common reason: We value the privacy of our users.

Empty walkway in green field

Photo by Erik Mclean at Pexels

While our proprietary competitors constantly (mis)use user data to improve the UX of their products, we are maneuvering in thick fog. And whenever we try to reach out to our users to decide about design directions, we mainly get feedback from white, male nerds [1]. You hopefully see the problems this implies.

The only solution is: We need to be able to evaluate usage information from everyone else too. While preserving the privacy of our users of course.

We need to be totally clear here, as we are touching common ground of Free Software Communities. Data protection and control over own data is a major driver for a large fraction of our community [2]. And it is for us too. Only when it comes to avoiding data in the first place, we do have to state, that we need aggregated personal data to able to do our work. We do not think creating data is bad per se.

Is there a third way?

The current data economy is set up to send out personal data to the algorithms. Unfortunately data can easily be copied, and once it leaves your property, you lose all control over it. So, today, you either surrender and give up on your personal data or have to avoid producing data in the first place, which is hard work.

But how about a third way, where all personal data is stored locally on users’ devices and only the algorithms are sent to the data to learn? Then we only have to assure that the algorithm is not harming the privacy in anyway, that the results from those algorithms cannot be de-anonymized anymore and we can use the aggregated data of our users without compromising the integrity of a single user’s privacy.

The good news: We believe that we have found a partner project, polypoly.coop that creates exactly the infrastructure for this third way.

With this post we ask you, the community for your help and support for collaboration. We need you as a critical partner to watch out all we do aligns with our values. We, the UX team need the results of this collaboration to understand and reflect the needs of our diverse community. And we need your technical excellence to make sure we keep the basic promise: Any single user’s privacy is preserved.

You can find a lot of information about polypoly on their website. We still want to highlight a few aspects that make us think polypoly is the right project to partner with:

  • polypoly is legally organized as a cooperative. Only individuals can become members. Each member has one vote, no matter how many shares are owned. This assures that polypoly will always work for and in the interest of the users and can never be bought by big money.
  • polypoly is Free Software, following a dual licencing model that allows the commercial world to use the infrastructure in proprietary products. You can find the source code here: https://github.com/polypoly-eu/polyPod
  • the dual licencing model is part of a business model and allows the cooperative to generate income, which makes it likely that the project is there to stay.
  • polypoly is actively interested in cooperating with us and is willing to invest into this project

Benefits for LibreOffice

From time to time we run surveys about existing features or planned enhancements but without direct contact we have to invite users via social media. With the effect that questions typically are completed by tech-affine people with some expertise.

As a consequence we base decisions like what kind of UI should be the default [3] on anecdotal reports and input from individuals who are more vocal than other. Another example: A question we will never be able to understand right now, is what function comes most often before Undo, from which we could learn where uncertainties exists.

How it works

The basic idea of polypoly is pretty simple:

  1. User data is stored in a mesh network of user owned devices, which is called the polyPod. Each person stores their own data on their own device only, and it is only accessible with their explicit consent. So we want to log the personal data of our users in there, e.g. interactions with LibreOffice or demographic data from e.g. questionnaires. Very different kinds of data can be stored in the polyPod and all data stored is fully transparent and can be modified by the user.
  2. The users can define via so-called “features” who is allowed to work with their data, so The Document Foundation needs to ask the users for permission. Again, the user can fully control who is allowed to work on which data.
  3. Finally we can send out algorithms that learn from all the users that allow us to. So if we are interested how often different LibreOffice applications are being used related to the location of the user, we have to create an algorithm that learns the answer from user’s data and hops from user to user until we have the aggregated answer and the algorithm comes back to us.

The details are of course more complex and look well thought through. You will find more information on the polypoly website.

Next steps

In a first step we want to create a prove of concept that allows us to store relatively uncritical personal data from LibreOffice in the polyPod and run evaluations upon this. The actual data set has not been defined yet. Everything needs to be opt-in and must not compromise the normal operation of LibreOffice.

We want to increase the amount of generated data and users participating gradually until we are able to actually do UX work based on representative user feedback.

But most important is the backing of our community. We know how important privacy for our community is and we want you to help us and polypoly to do things right. We are convinced that this third way of dealing with personal data is the only possibility for Free Software to be successful also on the UX side. Because you cannot do proper UX work without proper contact to the users.

We are asking you to back this collaboration. If you have questions, please go ahead. Not only now, but always. Please, be a critical partner and help us to do things right and succeed in the end.

Written by Bjoern Balazs

Comments
  1. 4 weeks ago
    • 4 weeks ago
  2. 4 weeks ago
  3. 4 weeks ago
    • 4 weeks ago
      • 4 weeks ago
  4. 4 weeks ago
  5. 4 weeks ago
  6. 4 weeks ago
  7. 4 weeks ago
    • 4 weeks ago
  8. 3 weeks ago
  9. 3 weeks ago
    • 3 weeks ago
      • 3 weeks ago
        • 3 weeks ago
  10. 3 weeks ago
  11. 3 weeks ago
  12. 3 weeks ago