On this page
The law
People start using software straight away instead of reading instructions, even when a little learning first would save them time overall.
The paradox has two halves. People are motivated to get their task done now, so they skip learning. And once they find a way that works, they keep using it, even when a faster way exists.
Where it comes from
John Carroll and Mary Beth Rosson, researchers at IBM, described the paradox in a chapter of the collection Interfacing Thought, published by MIT Press in 1987. They had observed people learning office software. Instead of working through manuals and tutorials, learners jumped into real tasks, skipped explanations and improvised, often making errors that the instructions would have prevented.
Carroll and Rosson called this a production bias: people care about producing results, not about learning the system. They also described an assimilation bias, the habit of applying what people already know even when it does not fit. Both make sense for the individual in the moment and are costly over time. Their answer was to design for this behaviour, not to fight it.
What it means for an interface
- Design for people who never read the manual. The main paths must be usable without instructions, because most people will not read them.
- Teach in context. Short hints at the moment they apply work better than an upfront tour people dismiss.
- Make help findable when people get stuck. People turn to help after a problem, so help should answer the question they have right then.
- Surface faster methods gently. People settle on their first working method. Show shortcuts where they are used, such as next to the menu item they replace.
Examples
Onboarding tours. A five-screen tour before people can use an app is usually skipped. A product that lets people start immediately and points out one feature when it becomes relevant fits the paradox.
Keyboard shortcuts. Menus that show the shortcut beside each command teach the faster method at the moment people use the slower one.
Empty states. An empty project screen that explains what to do first, with one clear action, teaches through doing rather than through a manual.
Where it is misapplied
- Removing help entirely. People skip documentation until they need it, and then they need it badly. Keep it, and make it searchable.
- Blaming people. "Users don't read" is not a reason to accept errors. It is a constraint the design has to meet.
- Interrupting with tips. Too many in-context hints become the tour people were avoiding. Show one at a time, when it applies.
- Hiding power features. Not everyone will find advanced features alone. Make them discoverable, without forcing them on everyone.
Checklist
- The main tasks can be completed without reading any instructions.
- Guidance appears in context, at the moment it applies, one hint at a time.
- Any onboarding tour can be skipped and revisited later.
- Help is searchable and written around tasks people try to do.
- Faster methods, such as shortcuts, are shown next to the slower method they replace.
Related
- Jakob's Law: people apply what they learned elsewhere instead of reading yours.
- Tesler's Law: complexity the product absorbs is complexity no one has to learn.
- Zeigarnik Effect: people are driven to finish the task in front of them.
- Patterns: Empty State.