I've seen this too. One company I worked for had a product that was deployed too soon. We as a dev team decided it was time to roll it out when it wasn't. Our fault. This created a crisis mode. It was a rush - live editing code on production servers to save an engineer rollout (nature of the product was one mistake locked us out of the box), seeing small fixes improve conversion by noticable amounts (or reduce the call queue significantly), etc. Bad engineering practice, but it made me understand the cowboy appeal, and I don't think I want to do it again.

However, this resulted in a bit of crisis addiction, followed by burnout. However, the CEO learned that crisis mode was a death march motivator. Suddenly it turned into crisis-this and crisis-that. Every small feature was a "drop everything we'll lose customers" crisis. It started wearing pretty quick. At some point he dropped the crisis act and just stated flatly that "your terms of employment state that you will work as needed, i need you now" the day before I had scheduled time off, and the "crisis" was getting a new feature complete for shipping in a year. Needless to say I got a new job.

Basically, I agree with clavalle - crisis mode if not used exceedingly sparingly can lead to a new set of expectations that will chew up a dev team.