DEV Community

Cover image for Detecting if a User Is Actually Present on Mobile (Without Using Camera or Spying)
iqbal singh
iqbal singh

Posted on

Detecting if a User Is Actually Present on Mobile (Without Using Camera or Spying)

One of the first problems we hit while building a reward-based mobile app was unexpected.

Not performance.
Not scaling.
Not payments.

Presence.

The Internet Counts Activity — Not Attention

Most systems assume something very simple:

If the app is open → the user is there.

But mobile behavior doesn’t work like that.

People open apps and:

put the phone on the table

switch to another device

mute videos

let content run in background

auto-play overnight

From a server perspective, this still looks like engagement.

From reality… the user left 20 minutes ago.

If rewards exist, this becomes a serious problem.

You are no longer rewarding a human.
You are rewarding idle electricity.

Why Traditional Signals Fail

At first glance, it sounds easy:
“Just detect interaction.”

But every simple signal breaks.

Screen ON → phone could be untouched
Video playing → not watched
App foreground → user not looking
Touch → automation possible

Even worse — modern mobile OS privacy rules remove many detection options.
No camera usage.
No intrusive monitoring.
No behavioral spying.

So the question becomes harder:

How do you verify presence without violating privacy?

Presence Is a Pattern, Not a Button

We realized attention isn’t a single action.

It’s a rhythm.

Real users behave inconsistently in very consistent ways.

They:

react inside natural timing windows

pause irregularly

interact slightly differently each time

get interrupted

return at human speeds

Automation behaves differently:

perfectly timed

repetitive

uninterrupted

statistically smooth

So instead of detecting one event…
we started validating sequences.

Not “did you tap?”
But “does your behavior look alive?”

The Shift in Thinking

We stopped asking:

Did the user interact?

and started asking:

Could this realistically be a person over time?

This transforms verification from an event problem into a probability problem.

Not certainty.
Confidence.

A presence score instead of a trigger.

Why This Matters

Reward systems fail when effort can be simulated.

If idle phones earn the same as active humans,
the system collapses into farming.

So before building any economy, we had to answer a simple question:

Can software recognize presence without surveillance?

We don’t fully know yet.

But we are experimenting with something we call:

Proof-of-Watch — measuring participation instead of playback.

Because attention might be the only scarce resource left online.

We’re testing these ideas in a project called WORK Network.

Feedback and Suggestion welcome

GooglePlay
https://play.google.com/store/apps/details?id=com.mobileweb3.workapp

AppStore
https://apps.apple.com/us/app/work-app/id6758291781

Top comments (0)