Quick thread on UI/UX design, ask vs guess culture, & a certain widespread attitude in commercial UX design circles that I’ve come to refer to as “UX paternalism” – exemplified by the assertion that “we (designers) understand what users want better than they do”.
This is a sentiment I’ve come across time & time again in UI & UX design conversations. In the UX classes I took in undergrad, it was literally part of the curriculum. And it’s widespread for a reason – there’s an extent to which it’s true, especially of novice users.
But as a totalizing design philosophy, “users don’t know what they want” is a self-fulfilling prophecy. If users are only exposed to interfaces that assume they’re clueless, they’ll give up on even trying to learn more sophisticated usage patterns. A kind of learned helplessness.
UX paternalism is often justified w/a rhetorical bait & switch: designers make a change that users say they don’t want, then use increased “engagement” numbers to argue users really *did* want the change after all.
(Hidden presumption: engagement up == better UX. This is false.)
(Hidden presumption: engagement up == better UX. This is false.)
UX paternalism intersects w/the rollout of opaque recommendation engines, algorithmic feeds, etc in a massively user-hostile way. We’re spending so much time & energy trying to guess what users want, at the expense of opportunities for users to explicitly state their preferences.
IME, users often know perfectly well what they want, but the UI doesn’t give them any way to express it! How can you possibly say users don’t know what they want if you never let them *tell you* what they want – if you keep trying to guess their intent without ever asking for it?
This ask-vs-guess tension isn’t unique to UI design. In fact, the idea of “ask culture” & “guess culture” as distinct *things* originally arose in the context of asking for favors from other people: ask.metafilter.com/55153/Whats-th…
IMO the ask/guess distinction works pretty well for attitudes about *preferences* as well as favors (do you ask people what they’d prefer explicitly, or try to guess using other cues?), so I’ll be using ask/guess terminology in that sense from here on out. Anyway…
When people & preferences are diverse, ask culture is objectively better at satisfying people’s needs. It means you don’t have to model everyone perfectly (a task which rapidly approaches impossibility as group size + diversity increases) to get things right in relevant ways.
Plus, guess culture can act as a shield for powerful people who like to run roughshod over others’ preferences. It enables them to do whatever they want, then say they assumed you’d appreciate it – they had no way of knowing you wouldn’t – you really should’ve said something!
In this way, UX paternalism is economically beneficial to ad-funded tech companies. They use it as cover to shove whatever changes they want at you, then claim that actually their *data* shows you’re using the product more as a result, so really it must’ve been Good UX all along.
And when a guess is wrong in a harmful way, the company defers responsibility for the bad call – “the algorithm did it” – then claims they’re improving the algorithm so it’ll make a better guess next time. Dodging around the fact they shouldn’t be guessing in the first place!
Reminds me of conversations I’ve had with cis people who are used to guessing gender pronouns from appearance, & can’t figure out how to make these guesses for people like me. To which I can only say: stop guessing! It’s not a thing you can reliably guess given the info you have!
To close this off: there’s a real need for innovative UI patterns that enable users to express more sophisticated preferences in human-friendly ways. Deferring design decisions haphazardly to the user by adding more & more switches over time is still bad design.
The end result, I suspect, will look a lot more like a *language* for talking to the computer than a menu full of buttons & switches. But to even see the development of such a language as a worthwhile project, you have to first accept that users are *worth listening to*.
