Sorry, but several things you mentioned shot up very large, very loud, very gawdy red warning flags for me.

• He had a dollar estimate. This is something that has been done before, and likely will be done again, too.
• "would normally take much longer to accomplish" - Again, a process that is going to be done again, most likely.
• It "could be a 'throw-away' script" - how many times have we heard that, only to see them still in use much beyond their expected lifetime?
• You mentioned freezes and cuts over the past year, and having to severely multi-project.

Given the above, I don't see this "throw-away" script getting tossed out anytime soon. From the latter of the points above, I would guess it may be sometime before you can clear enough from your plate to rework it, and by the time you get to the point of suggesting it be reworked, I expect you may very well hear, "well, it's worked okay so far-why put the time/cost into redoing something that already works?"

As to being called upon to develop solutions/tools "in three minutes or we're all dead" (to quote Adm. Kirk in STII:TWOK), I think we all get called upon to do that, whether we like that kind of pressure or not. I applaude that you were able to maintain good techniques while under this kind of time pressure.

As to how I cope, I considered two examples:

• Recently, because of unplanned equipment changes, I had to rework a system for managing a service on a machine to handle the service being split between multiple machines, and managed to do so in about 14 hours (after being awake all day)... although the only bug I missed took 2 days to catch because of circumstances necessary for it to happen.
• A few years ago, I had to develop an alternative system for managing users in a service, which I managed in about 10-11 hours at the time (I was still very much a novice then).

After both, I was mentally exhausted and slightly out of it, and so tried to catch up on sleep and recharge what mind I had left.

As to the question, "is it reasonable to turn away such requests," I believe it is reasonable to turn it away or possibly suggest it be held, if possible, when the goal is too unreasonable, or when the project has too little solidity to have any direction, or when to drop what you are in to jump into that would be of more detrimental than the outcome is worth.

(And yes, I follow Mr. Scott's example about over-estimating the time required for something, although maybe not to his factor of 4. Still would like the neural interface Mr. Barkley had in one of the ST:TNG episodes, though.)

Just the $0.02 of an egg (which, when you factor in inflation, may not be worth that much). I look forward to the other comments on this thread.