It's easy to run a pilot and check. From previous experiences, it drops a little in the beginning, but training smooths the transition. With increased decision freedom due to removed vendor lock-in, governments can both choose the best tool for a given job (even if it's incompatible with, say, MS SQL Server) or develop the missing functionality (because they have access to source code). And governments can cooperate over these developments across regional or national borders. Reduced licensing cost is just the beginning.
> especially for information workers who are set in their ways
In other words, who can't learn another IDE beyond Visual Studio and other language beyond VB. Oh! Wait! They have to learn a new Visual Studio every couple years. And they had to learn VB.Net (or C#, or managed C++). They already do relearn lots of stuff.
And if you have a developer who is not willing to learn, fire him.
> OpenOffice is still far behind Office in usability for the average information worker
Come on... How many people within Microsoft really stress Words feature-set? 1%? The average "information worker" would be perfectly fine with WordStar and Visicalc. Also, it's not like Office 2003 users weren't completely lost when upgraded to 2007 and more so on 2010. The GDP of a decent-sized country was lost on menu repositioning over Office's history.
> The savings in license costs is probably outweighed by the long-term productivity hit.
Long term productivity hit meaning these workers cannot learn new things?