Designing for Humans Before We Know the Ending

I’ve recently gotten into watching police bodycam videos on YouTube.

A lot of them.

Enough that I sometimes leave them running in the background while I work.

And somewhere along the way I noticed something weird.

I fast-forward through the arrest.

Someone runs? Skip.

There’s a five-minute wrestling match? Skip.

A bunch of cops are trying to get someone into handcuffs? Skip.

That’s supposedly the exciting part.

It’s certainly the part that gets the thumbnail.

But I’m waiting for everyone to start talking again.

I want the interaction before the arrest.

And I want the interaction after it.

Because I’m not really watching arrests.

I’m watching people.

And as a designer, that makes bodycam footage unexpectedly fascinating.

We Say We Design for Humans

Designers spend a lot of time trying to understand people.

We interview them.

We observe them.

We run usability tests.

We build personas and journey maps.

We study analytics.

We ask people what they want, what frustrates them, what they would do, why they did something.

All of this is useful.

All of it also has a problem.

The moment I sit someone down and say, “I’m conducting research,” I have changed the situation.

They know they’re being studied.

They know there is a designer sitting across from them.

They know we’re interested in what they have to say.

Put someone into a usability test and now they’re not simply using your product.

They’re using your product while knowing somebody is watching them use your product.

That doesn’t invalidate research.

It just means we should never confuse research conditions with life.

Humans are messier than our research sessions.

Reality TV Isn’t Reality

This problem isn’t unique to design.

Most of the human behaviour we see through media is mediated.

Movies and television are constructed.

Documentaries have filmmakers, subjects and narratives.

Social media is performance almost by definition.

Reality television might be the funniest example.

Put people into an artificial environment surrounded by cameras, producers and incentives to remain interesting enough to stay on television, and then call the result reality.

Bodycams are different in one important way.

The camera isn’t primarily there to create entertainment.

The interaction would have happened whether or not I eventually watched it on YouTube.

That’s rare.

It doesn’t make bodycam footage objective.

A police interaction is an incredibly specific environment. You’re disproportionately watching conflict, intoxication, fear, crime, mental distress, authority and people having very bad days.

Then a YouTube channel selects the footage, edits it, titles it and turns it into content.

There are filters everywhere.

So this isn’t a representative sample of humanity.

But it offers something I don’t encounter very often:

people dealing with a situation before they know what the story is.

For someone interested in human behaviour, that’s gold.

Everyone Is a Research Subject

At first, you watch the person who gets stopped.

Then you start watching the cop.

Then another officer arrives and suddenly you’re watching them too.

Eventually I realized I was studying everybody.

The driver.

The passenger.

The officer.

The witness.

The backup officer.

The supervisor who arrives twenty minutes later.

Every new person enters with a different amount of information.

And that’s where things get really interesting.

The first officer might know why they made the stop but almost nothing else.

The driver knows things the officer doesn’t.

The passenger might know something neither of them knows.

The second officer arrives knowing almost nothing and has to rapidly construct their own understanding from whatever they’re told.

Everybody is simultaneously observing everybody else.

Everyone has a model of what is happening.

Those models are rarely identical.

And then new information arrives.

Watch what happens.

The Research Stage

The beginning of a traffic stop sometimes looks remarkably like research.

Start with an observation.

Ask a question.

Form a hypothesis.

Look for evidence.

Ask another question.

Notice a contradiction.

Update the hypothesis.

Check another source.

Update again.

Eventually:

Act.

Observe → Question → Hypothesize → Test → Update → Decide → Act.

Sound familiar?

The difference is that the research subject is actively researching the researcher.

The person being questioned is trying to figure out what the officer knows.

Why did they ask that?

Did they see something?

Did my passenger say something different?

Am I free to leave?

Am I about to get arrested?

The observer changes the observed.

The observed changes the observer.

And both keep updating.

That’s much closer to real human systems than the clean separation between “researcher” and “participant” that our diagrams sometimes suggest.

Watch the Model Break

This might be the most useful part for me as a designer.

Someone forms an explanation of what’s happening.

Then reality contradicts it.

What happens next?

Some people update immediately.

Some resist.

Some reinterpret the new information so they can preserve their original belief.

Some become defensive.

Some become curious.

Some double down.

Police officers do it too.

An officer can approach a situation with an initial assumption and then encounter evidence that doesn’t fit.

The interesting question isn’t whether the original assumption was correct.

Nobody gets perfect information.

The interesting question is:

How quickly can you recognize that your model no longer matches reality?

That’s design.

We make models constantly.

We think we know what the problem is.

We think we know who the user is.

We think we know why they’re struggling.

We think we know what feature they need.

Then we put something in front of an actual human being.

Reality responds.

Sometimes designers update the model.

Sometimes we explain reality away.

“The user didn’t understand the prototype.”

“They aren’t our target customer.”

“They weren’t using it properly.”

Maybe.

Or maybe the model was wrong.

Humans Change When Conditions Change

This is why I like watching what happens before and after an arrest.

Before the arrest, someone might be confident.

Then evidence appears.

Now they’re defensive.

Then they’re angry.

Then they’re bargaining.

Then they’re calm.

Then they’re surprisingly cooperative.

It would be easy to ask which one is the “real” person.

I think that’s the wrong question.

They’re all the person.

The environment changed.

The information changed.

The incentives changed.

The perceived options changed.

So behaviour changed.

That’s an important thing to remember when designing for humans.

We love turning behaviour into attributes.

This user is impatient.

That user is risk-averse.

This person doesn’t understand technology.

That customer values convenience.

Maybe.

But behaviour doesn’t emerge from a person alone.

It emerges from a person inside a situation.

Change the situation and you may change the behaviour.

Which means sometimes the thing we should be designing isn’t the person’s next action.

It’s the conditions surrounding that action.

The Plan Lasts Until Reality Shows Up

There’s another lesson buried in these videos.

Everyone starts with some version of a plan.

Then reality happens.

An officer walks toward a car expecting a routine traffic stop.

Thirty seconds later, something doesn’t add up.

Five minutes later, another officer is there.

Ten minutes later, the original reason for the stop barely matters.

Nobody could have designed the entire sequence beforehand.

They have procedures.

They have training.

They have constraints.

But within those boundaries, they have to improvise.

That’s an important distinction.

Improvisation isn’t the absence of a system.

It’s what happens when a system encounters a reality it couldn’t completely predict.

We do this badly in design sometimes.

We try to anticipate every path.

Every state.

Every edge case.

Every possible thing a human might do.

Eventually reality invents another one.

Maybe resilient design isn’t about predicting everything.

Maybe it’s about building systems that can remain useful when our predictions fail.

Before We Know the Ending

I don’t think bodycam footage is some magical source of truth about human beings.

It’s too selective for that.

But it has reminded me of something I don’t want to forget as a designer.

Humans don’t experience our designs as journeys on a whiteboard.

They arrive carrying information we don’t have.

They misunderstand things.

We misunderstand them.

Their priorities change.

Their environment changes.

They make decisions that look irrational until you understand what they believed at the moment they made them.

They improvise.

So do we.

And unlike the researcher looking back at the session, the person living through it doesn’t know what’s coming next.

That’s probably why I keep watching.

Not because I’m particularly interested in arrests.

I’m interested in that moment before everything makes sense.

When several people have incomplete models of reality.

When new information is still arriving.

When the plan has stopped working.

And everyone has to decide what to do next.

Because that’s not just what happens during a traffic stop.

That’s what happens when you design for humans.