I think you're both right to some extent. Both approaches have difficulties, just different ones. The open source approach can have lower up-front costs (but i

I think you're both right to some extent. Both approaches have difficulties, just different ones. The open source approach can have lower up-front costs (but isn't guaranteed to have...), but it requires talent that's harder to find and more expensive as a result. There's also the problem of user resistance, as you point out. We have enough trouble just getting users to use FF/Chrome instead of IE; Linux desktop systems wouldn't go over very well and would require a ton of effort. That's not even touching the problem of a lot of infrastructure being irreversibly tied to software that expects to run in a Windows environment. But the Microsoft stack can also incur a lot of up-front expenses for larger organizations, and license management isn't very easy either. And, it has its own issues and glitches. Just to pull a number out of my butt, I'd guess a city like Detroit would be looking at millions in Microsoft licensing costs alone if they were to upgrade their entire infrastructure. And XP->7 isn't usually too bad, but there are some recurring problems we've encountered (Quickbooks), and XP->8 doesn't work at all.