Big Data is watching you image

UX Insights (that we cannot get right now)

After the criticism in the last post about the limitations of KUserFeedback (KUF) for doing data-driven UX work — let’s get more detailed and constructive:

What insights do we as KDE UX people need to do even better than we are currently doing?

Let us start with what we already get from KUF. We get usage data, like how many people are using Wayland vs. X11. But we only get usage data according to our telemetry policy. So we do not get any deeper insight into how users configure their sessions when using Wayland compared to X11. But this is the kind of information we would need to do proper data-driven UX. What settings are users changing? How many users have icons on their desktop, and which ones? Are people manually mounting network drives? Which System Tray icons are interacted with the most? And so on.

But while this information is already impossible to gather with our current approach, we’re only scratching the surface. We need even deeper UX insights, like understanding where people click. And where they click next (in terms of Markov chains). That way we can understand if people are using Plasma the way we intended when we designed it. Or, how long does it take them to get from point A to point B? Are they taking detours because we’ve laid out paths that users don’t understand in the way we intended?

None of these questions can be answered with our current approach to telemetry.

The basic problem is that we currently send all the raw data to the KDE servers to get the answers we need. And the data we need to collect in order to get the above described desired user insights could of course be used to “identify a specific user” – which is not allowed by our telemetry policy for good reason.

And yet we need even more data. We want to target all users, or only users who exhibit certain behaviors. We want them to fill out questionnaires to better understand why they behave the way they do, to understand their goals and intentions. This would be extremely helpful in understanding bug reports. Or to support our design discussions with relevant data from real users.

All of this can only be achieved with a fundamental change in the way we do telemetry.

Existing alternatives, such as the opt-out Endless OS metrics system, also do not allow enough user insights and share the problem that the data leaves the property of the data owners, the users. That is why we have been working on the privact ecosystem, which allows all the insights described above, while fully preserving users’ privacy. And because of that, we can not only ask for more intimate data, but we can also make participation opt-out and so get data from substantially more people. And why is that? Because with the privact ecosystem, there is no technical possibility that any individual’s personal data can ever be shared remotely. Never. But it would finally enable good user-data-driven UX work. For the sake of KDE and our users.

Please also join the discussion about this issue on invent.kde.org.


Kommentare

Leave a Reply

Your email address will not be published. Required fields are marked *