Monday, my employee could not make it to the office due to a fever.
Tuesday, my employee showed up for work at 9am, but the power went out at noon, and the whole office was given the rest of the day off.
Wednesday, as my employee was driving to work, he got in a motorcycle accident, and did not come into the office.
Thursday, my employee worked a full 8 hour day, but did not `git commit` anything, did not email me about his status, and did not, apparently, get anything done.
Friday, my employee was lost in a flood. His manager called me to explain that, while she has no idea where my employee is right now, she's going out into the flood, personally, to search for him.
I guess they didn't look at redundancy or reliability in their cost equation...So we marched down the road of the Windows platform. We don't have any Unix; we don't have an AS/400; we don't have any mainframes — we don't have anything outside of Windows. There has been tremendous cost savings.
...everything's Windows. Every technician that works on a server works on a Windows server. Every technician working on a desktop works on a Windows desktop. That's quite a bit easier than other flavors of desktops or OS/2 or whatever else is out there.
No amount of careful planning will ever replace dumb luck.