Social Design
I have used my share of software tools with terrible user interfaces, and I
have read even more complaints from other people about such tools. Why can’t
the designers just put themselves in the shoes of the users, mentally, and
think about what kind of interface would be the most convenient and useful?
Those stupid autistic programmers must be completely devoid of any trace of
empathy to produce such designs, surely. If we could only make sure such design
work is done by reasonable people thinking about others, it surely would be
must better!
Well, turns out that this is not so simple, but it’s a little bit involved to
explain. I will try anyways.
The short version of it is that what you think is simple and obvious, is not so
simple and obvious to others. It only seems to be so, because you are
mis-communicating. You are using the same words as the others, and the others
agree with what they hear, but those words mean different things to different
people. This is invisible to you until someone takes what you said and turns it
into a concrete example. Chances are that when you look at it, you will say
they completely mis-understood you, and that is not what you meant at all. But
that is what they understood, and what they agreed on. If you make each person
who has previously so nicely agreed with you turn your words into a concrete
example, each one of them will produce some interpretation of your words that
is most likely completely different from what you actually meant.
If you took that horrible user interface, and asked all those people who
completely agreed that it is horrible, to tell you what exactly is wrong with
them, you would hear completely inconsistent and wildly different opinions. And
that’s not just because most people are very bad at analyzing user interfaces.
It’s also because each of those people probably wanted something different. But
as long as they talk between themselves, they will generally all agree, because
each of them will choose the interpretation of the words other said that
matches their own ideas.
Now, programming doesn’t work that way. Programming is one of those disciplines
where you can’t solve things by using a different word to describe the same
thing, until everyone agrees. It doesn’t work on consensus, it works on
physical reality, and unlike consensus, physical reality is unique. There is at
any point only one truth, even when there can be multiple interpretations, and
hypotheses that fit the experimental data. There is a necessary assumption in
there, that cannot be avoided, that all those interpretations, and hypotheses
are in fact describing a single underlying reality. And to make computers work
you have to believe in that single truth, you have to work within the confines
of the physical reality, because the computers don’t care about what you mean,
they just do their computation, in more or less predictable ways, and it’s up
to you to arrange them in a way that does something useful.
You might say: that is completely fine, they have to be those hyper-honest,
literally-minded autistics when they commune with their machines, but can’t
they be a little bit more human when they speak to users, or imagine what the
users want, so that they can then do their magic and make the machines do what
we want them to do? Isn’t this their entire job, to translate between the
normal people and the machines? Shouldn’t they put more effort into thinking
normally?
And that’s the entire problem. There isn’t just one “normally” out there. You
think there is, because you live in a society, and you have learned to exist in
it without too much conflict, and to reinterpret things others are saying to
match with your own opinions and experiences. With the physical reality, there
is no problem: there is only one truth, and if you make enough experiments,
collect enough data, and think about it hard enough, you can eventually get to
an interpretation of your data that is close enough to that truth for things to
work most of the time. This doesn’t work with “normal” people. They seem to all
be saying the same things, but they mean different things, and often they
don’t even mean anything at all, just repeat something they’ve heard without
understanding it, and will claim it means something completely different as
soon as you try to pinpoint it. They will also change their minds, and claim
contradictory things. And every translation to a concrete, formal description
that you might attempt will be wrong to most of them. You see, you simply don’t
understand.
So are we doomed to unusable software product with interfaces designed for
people who don’t understand the basic needs of the users? I believe good
software tools are possible, but the path to them does not lead through making
the designers more human and making them listen to “normal” people. On the
contrary, I think it’s best when they don’t listen to what the users say at
all, and instead just look at what the users do. Words lie. While it is
possible to mislead with actions as well, especially if you are trying to guess
the intentions, they are at least much less ambiguous. And they are repeatable
and rooted in physical reality. We can do science on them, and we can figure
things out. And then we can translate them into concrete, working things.
So next time you are about to complain that user interface designers don’t
listen to their users, keep in mind that listening to users is not how you
make good user interfaces.
dopieralski.pl
Social Design¶
I have used my share of software tools with terrible user interfaces, and I have read even more complaints from other people about such tools. Why can’t the designers just put themselves in the shoes of the users, mentally, and think about what kind of interface would be the most convenient and useful? Those stupid autistic programmers must be completely devoid of any trace of empathy to produce such designs, surely. If we could only make sure such design work is done by reasonable people thinking about others, it surely would be must better!
Well, turns out that this is not so simple, but it’s a little bit involved to explain. I will try anyways.
The short version of it is that what you think is simple and obvious, is not so simple and obvious to others. It only seems to be so, because you are mis-communicating. You are using the same words as the others, and the others agree with what they hear, but those words mean different things to different people. This is invisible to you until someone takes what you said and turns it into a concrete example. Chances are that when you look at it, you will say they completely mis-understood you, and that is not what you meant at all. But that is what they understood, and what they agreed on. If you make each person who has previously so nicely agreed with you turn your words into a concrete example, each one of them will produce some interpretation of your words that is most likely completely different from what you actually meant.
If you took that horrible user interface, and asked all those people who completely agreed that it is horrible, to tell you what exactly is wrong with them, you would hear completely inconsistent and wildly different opinions. And that’s not just because most people are very bad at analyzing user interfaces. It’s also because each of those people probably wanted something different. But as long as they talk between themselves, they will generally all agree, because each of them will choose the interpretation of the words other said that matches their own ideas.
Now, programming doesn’t work that way. Programming is one of those disciplines where you can’t solve things by using a different word to describe the same thing, until everyone agrees. It doesn’t work on consensus, it works on physical reality, and unlike consensus, physical reality is unique. There is at any point only one truth, even when there can be multiple interpretations, and hypotheses that fit the experimental data. There is a necessary assumption in there, that cannot be avoided, that all those interpretations, and hypotheses are in fact describing a single underlying reality. And to make computers work you have to believe in that single truth, you have to work within the confines of the physical reality, because the computers don’t care about what you mean, they just do their computation, in more or less predictable ways, and it’s up to you to arrange them in a way that does something useful.
You might say: that is completely fine, they have to be those hyper-honest, literally-minded autistics when they commune with their machines, but can’t they be a little bit more human when they speak to users, or imagine what the users want, so that they can then do their magic and make the machines do what we want them to do? Isn’t this their entire job, to translate between the normal people and the machines? Shouldn’t they put more effort into thinking normally?
And that’s the entire problem. There isn’t just one “normally” out there. You think there is, because you live in a society, and you have learned to exist in it without too much conflict, and to reinterpret things others are saying to match with your own opinions and experiences. With the physical reality, there is no problem: there is only one truth, and if you make enough experiments, collect enough data, and think about it hard enough, you can eventually get to an interpretation of your data that is close enough to that truth for things to work most of the time. This doesn’t work with “normal” people. They seem to all be saying the same things, but they mean different things, and often they don’t even mean anything at all, just repeat something they’ve heard without understanding it, and will claim it means something completely different as soon as you try to pinpoint it. They will also change their minds, and claim contradictory things. And every translation to a concrete, formal description that you might attempt will be wrong to most of them. You see, you simply don’t understand.
So are we doomed to unusable software product with interfaces designed for people who don’t understand the basic needs of the users? I believe good software tools are possible, but the path to them does not lead through making the designers more human and making them listen to “normal” people. On the contrary, I think it’s best when they don’t listen to what the users say at all, and instead just look at what the users do. Words lie. While it is possible to mislead with actions as well, especially if you are trying to guess the intentions, they are at least much less ambiguous. And they are repeatable and rooted in physical reality. We can do science on them, and we can figure things out. And then we can translate them into concrete, working things.
So next time you are about to complain that user interface designers don’t listen to their users, keep in mind that listening to users is not how you make good user interfaces.