LLOS.ai
How Things Work
● How Things Work · Module 6 · Error

Error

You will delete the wrong file and send the half-written email — not because you’re careless, but because you’re human. Kind design forgives you.

Signals, mapping, feedback, models, constraints — five modules of stopping mistakes. But no design catches them all. A tired, distracted, ordinary human will still slip through every net you build. This last module is about that moment: what a design does after the error, when the damage is one tap away.

The humane answer has three parts: understand the error, catch it, and above all forgive it. A slip (right goal, wrong action) is not a mistake (wrong goal), and they want different fixes. A sensibility check notices the missing zero before the money moves. And a reversible action makes a slip cost a click, not a catastrophe. You’ll feel undo, a sanity check, and the difference between a slip and a mistake.

one that prevents you.

~8 min3 felt labsslips vs mistakes10-question bankglossary · FAQ
The journey — how things work
signalsaffordance
mappingmatch the world
feedbackdid it work?
modelsthe story
constraintsno wrong moves
errorforgive
Same promise, the last step: when a design confuses you, it’s the design’s fault — not yours.
What you’ll be able to do
01 · Forgive the slip

The undo

The single most humane thing a design can do is let you take it back. Delete the wrong photo, send the wrong message, drag the wrong file — and a good system answers not with “are you sure?” but with “done — and here’s undo if that was a slip.” The mistake still happens; it just stops mattering.

Delete the same photo twice. First where gone means gone, then where a quiet toast offers it straight back. Feel the drop in your stomach vanish.

The undo
Delete the photo. See what each design gives you afterwards.
▶ Start on No undo and feel it vanish for good. Then Undo — one tap brings it home.
Beach trip · 2019
Photo deleted.
recovered?
02 · Catch it before it commits

The sensibility check

Some slips you can’t design away — you can catch them the instant before they matter. A sensibility check is the system glancing at what you did and quietly asking, “that doesn’t look right — sure?” The classic: a bank noticing a transfer is 100× your usual, and pausing on the extra zero your eyes slid right past.

Wire some money. Fat-finger an extra zero, then hit transfer — first on a system that just does it, then on one that catches the nonsense.

The sensibility check
Your usual transfer is $100. Add a zero (or two), then hit Transfer.
▶ Start on No check and watch it sail through. Then Sanity check — it catches the slip.
To: A. Landlord
$100
watching

caught?
03 · Two kinds of wrong

Slip, or mistake?

Not all errors are the same, and the difference decides the fix. A slip is when your goal is right but the action goes wrong — autopilot, a wrong button, a look-alike jar. A mistake is when the goal itself is wrong — you planned badly, or believed something false. Slips want better feedback and constraints; mistakes want better information and models.

Sort each of these four. Right goal, wrong action — or wrong goal all along?

1 of 40 sorted right
The craft of forgiveness

People will slip. Catch them.

Error feels like the user’s department until you accept the truth: everyone slips, all the time, and the only question is whether the design was ready for it.

Design for the mistake, not the ideal user.

Start by naming the error. A slip is a right goal gone wrong in the doing — the wrong button, the look-alike jar, the deep habit that hijacks your hands. A mistake is a wrong goal from the start — a bad plan, or a false belief. They need opposite fixes: slips want clearer feedback and constraints so the action can’t stray; mistakes want better information and models so the plan is sound. Blur the two and you fix neither.

Then build in layers, because no single guard is perfect. The Swiss-cheese model says every defence has holes; you stay safe only when the holes don’t line up. So stack them: a constraint to prevent, a sensibility check to catch, a confirmation for the truly dangerous, and — the last, kindest layer — an undo for everything else, so a slip that gets through costs a click, not a catastrophe.

And retire the phrase “user error.” When a tired person makes the same slip a thousand others made, that’s not a thousand careless people — it’s one careless design. Treat every ‘error’ as a brief: what could we change so this could not have happened, or could be undone in a breath? A design that forgives is a design that has finally understood who it is for.

Cast your vote

Which design forgives?

Five face-offs. Same slip, two designs — one leaves you stranded, one hands you a way back. Pick the kinder, then see why.

Round 1 of 5Score 0
Seen in the wild

Forgiveness, built in

The kindest features are the ones that quietly assume you’ll slip. Two you lean on without a thought.

Undo Send
A five-second grace

Gmail holds your email for a few seconds and offers ‘Undo.’ It doesn’t stop the send — it makes it reversible, turning the gut-drop of a wrong recipient into a shrug and a click.

Ctrl + Z
The universal net

One key combination undoes almost anything, almost anywhere. It’s so ingrained you reach for it in the physical world by reflex — the clearest proof that reversibility is what people quietly rely on.

The loop closes

Six modules, one promise, held from the first page to the last: when a design confuses you, or trips you, or lets you fail — it’s the design’s fault, not yours. Signals, mapping, feedback, models, constraints, and now grace for the error that slips through anyway. That’s the whole craft of how things work: quietly taking the blame, so the person never has to.

Take it further

The real references

These ideas aren’t ours — they’re the field’s, named and tested over decades. Five worth your time, in the words of the people who wrote them. (Opens in a new tab.)

The vocabulary

The Error glossary

Six words that turn “the user messed up” into a precise, fixable diagnosis. Tap a card to flip it.

The nuance

Questions you actually ask

Busting the myths

What people get wrong

Myth
“User error is the user’s fault.”
If a thousand people make the same slip, that’s not a thousand careless humans — it’s one careless design. ‘User error’ is almost always a design that left the error possible and gave no way back.
Myth
“A confirmation dialog fixes mistakes.”
It doesn’t touch mistakes at all — if your goal is wrong, you’ll confirm it confidently. And people habituate to dialogs and click through. Undo forgives a slip far better than ‘are you sure?’ ever will.
Myth
“Just train people to be careful.”
Vigilance runs out. Under load, tired and rushed, everyone reverts to autopilot — that’s human. Design so the careless moment is caught or reversible, and you protect the person training never could.
Myth
“A good error message explains the code.”
‘Error 0x80070005’ helps no one. A good message says what happened, why, and what to do next, in plain words — and never blames. It’s a helping hand, not a slammed door.
Train the eye

Read the error

Ten questions in seven formats, tagged by thinking level — each with layered hints: a nudge, the reasoning, then a deeper connection. A wrong answer opens a door, never a dead end. Served from the question bank, never hardcoded.

Loading the question bank…
Question 1 of 10
Multiple choice
Keep going

Where this connects

▶ Play the labs