Simon Baker has written a superb post, capturing the elusive nature of empowerment.
He mentions “an environment that tolerates mistakes to cultivate learning”. For me, this is an essential part of innovation; of the Lean metaphor that suggests a Software Development team is more like a Product Development team (designing new cars) than a Production Line (churning out the same thing repeatedly).
In a Production Line, we try to minimise variance (think Jidoka and Continuous Integration and Deployment).
In a Product Development team, we try to maximise learning opportunities, which means experimenting (think risk, innovation and doing things no one’s done before).
“There is no such thing as a failed experiment, only experiments with unexpected outcomes.”
— Richard Buckminster Fuller
Are mistakes in Software Development really mistakes? Or just experiments from which we learn something new? And if they keep happening… what didn’t we learn?
Liz Keogh, lunivore 
If you think about it, the only reason why it’s worth suggesting that minimising variation is valuable is because there is actually a lot of variation.
Whether it’s a production line or product design, everyone is still experimenting and learning. For a production line, the experimentation is focused on how the job is done; for product design, the experimentation is focused on exploring and re-defining boundaries.