How should we define Too Big to Fail?

I'd like to propose some structure to the policy debate around the Too Big to Fail concept.