As a follow on to my earlier post on the topic, note that permission to fail isn't permission to be stupid. Nor is it permission to not really try to succeed, nor permission to be defeatist about things. You have to be trying to succeed or your failure is true failure.
And as a general rule we should expect to be succeeding much more often than we're failing. If you're trying something monumentally difficult like discovering a unified field theory then that's probably different, but most of the time we're doing things that have been done many, many times before. We're creating profitable direct mail campaigns or making advertisements that work or figuring out what our customers want and giving it to them. If you're frequently failing at tasks such as these, then maybe something else is wrong and you should look into that possibility.
When I grant permission to fail, it comes with the condition that the James Bond employee is doing her best and is performing up to her ability (or preferably a little beyond it). It comes with the condition that she is making every decision for a reason, and in each case she knows what it is. If you do that, everything's cool. If not, it's not.