Encourage Innovation by Not Punishing Honest Failure
Tell people to try, kill it fast, debrief, and punish only laziness
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 74%
Taking a real risk means increasing variance: a chance of a much better outcome and a chance of a much worse one. If a manager demands creativity but fires or docks anyone whose bet fails, employees rationally retreat to boring, safe, incremental work — the mixed signal cancels itself. The fix is to tell people explicitly to try, and if it isn't working, to report it fast rather than throwing good money after bad to force it. Then debrief: if the failure came from a sound intuition that simply didn't pan out, learn and move on; if it came from laziness — sleeping until noon — then punish. If the downside truly can't be tolerated, that's fine too, but then say 'play it safe here' rather than sending contradictory signals.
Origin
Uri Gneezy, who runs incentive experiments for a living, frames this in Mixed Signals; the discussion also cites Rory Sutherland's point that 'nobody ever got fired for hiring Accenture.'
Core principles
- 01Innovation means increasing variance — bigger upside and bigger downside.
- 02Punishing failure while demanding creativity is a self-cancelling mixed signal.
- 03Distinguish failure from a solid intuition versus failure from negligence.
- 04Fast kills plus debriefs turn losses into compounding learning.
How to run it
- 1
Decide whether you actually want risk here
Determine if the domain can tolerate downside; if it genuinely can't, tell people to play it safe rather than demanding creativity.
Pro tip Being explicit that a role is not the place for creativity is a legitimate, honest signal.
Watch out Don't say 'be creative' in a zero-tolerance-for-failure context — that's the core mixed signal.
- 2
Give explicit permission to take real risks
Tell people to try genuinely different approaches, accepting that variance cuts both ways.
- 3
Demand fast failure, not forced success
Ask people to surface a failing bet quickly instead of dragging it out to avoid looking wrong.
Pro tip Fear of punishment is exactly what makes people throw good money after bad.
- 4
Kill it and debrief
When something clearly isn't working, stop it and run a post-mortem on what happened and why.
Pro tip A failed sound intuition is often more instructive than a success.
- 5
Punish only negligence
If the failure traces to laziness or carelessness, penalize it; if it came from a solid good-faith attempt, protect the person.
Pro tip This is the line that keeps risk-taking alive without rewarding sloppiness.
Watch out Punishing good-faith failure trains everyone watching to stop trying.
In the wild
Gneezy notes he can write many boring papers by tweaking one small parameter of someone else's work, or take a real risk on something completely different. The risky path raises variance — it could be far better or far worse. If a boss fires or denies a bonus every time an experiment fails, there's no reason to take the risk. His rule: try it, kill it fast if it's not working, debrief. If it failed because he slept until noon, punish him; if the intuition was solid and it just didn't work, learn and move to the next thing.
→ Sustained willingness to attempt high-variance, potentially breakthrough work instead of safe iteration.
Common mistakes
Saying 'be creative' then punishing failure
Demanding innovation while penalizing the losses it inevitably produces is a self-cancelling mixed signal that drives people to safe, boring work.
Treating all failure the same
Failing from a sound intuition and failing from laziness deserve opposite responses; conflating them either kills risk-taking or rewards negligence.
Is it for you?
Best for
Managers and researchers who genuinely want breakthrough attempts, not safe incremental iteration.
Not ideal for
Roles where the downside is catastrophic and safety must trump creativity, like surgery or aviation.
From the transcript
“If you want creativity, don't punish. If you don't want creativity, which is also fine, tell them that that's not what you're looking for in…”
“Whenever I see that it doesn't work, just kill it, that's fine, and then do this debrief. What happened, why did it happen?”
From the episode
How To Understand Psychological Incentives - Uri Gneezy - #609
Uri Gneezy