The Sleeping Geek Kitten - Angers - / Nathonline-Beta
Before any potential future employers find this and take my resume off the pile, let me explain. Knowing how long I can work before getting tired is important to me. If I go too long, then I'm likely to just make things worse. Plus, if I know I only have a set amount of time to get things done, I can focus better - there is no later. The standard 8 or 9 hour day seems to work perfectly well for me.
That's not to say that I'm not willing to put in some extra time around a deadline, of course. I have stayed late a few times during my co-op terms and I've worked into the evening on school stuff more than once (though I've only ever done one all-nighter, and it wasn't even strictly necessary). I just ensure this is very much the exception and not the norm.
So, back to this proof I mentioned. The IGDA has a great article on why perpetual crunch modes just don't work. Most other industries started figuring this out 100 years ago (hence why the 40 hour work week is fairly standard). Research literally proves that after a certain number of hours worked in a single week, output goes down when compared to a more regular work week. Working more just doesn't pay.
High tech can be a brutal field when it comes to overtime expectations. My strategy? Don't make working longer a precedent. If you do, then that amount of work will be expected of you. But, of course, the more you work long hours, the less you'll do over time, making you want to work longer to make up for it, making you accomplish less... and so it goes...