This post is going to be somewhat of an open response to Ingbert's comment to my post on the IT@Illinois launch, and it will segue a bit into talking a bit about some aspects of a concept I've been working on with an amazing group of IT people from around campus.
The first point that I need to clarify is that I believe the current IT organization on campus came out of necessity but due to a different specific necessity than what Ingbert thought I meant. To me the necessity arose when central IT was unable to provide the IT services needed by the colleges, units, and departments. When central IT was unable to meet those needs at all or adequately, those organizations needed to do some of IT on their own to meet their needs. This still happens today, and you don't have to look far to find it: e-mail. There are hundreds of mail servers around campus, and it's really simple to see why that happened. The features of the e-mail service provided by central IT fell farther and farther behind as people sent more e-mail and with more attachments. Quotas are laughable in comparison to what it costs to provide gigabytes of space. It has also only been a few years that a good anti-spam feature was available to all campus accounts. Add a less than desirable interface on top of the technical features, and it becomes easy to see why there are hundreds of mail servers on campus tucked into departments, colleges, and even the offices of individual people.
The issue though is two-fold. Yes, part of the reason for this wide array of mail services is due to a questionable product being offered by central IT. There is however a second component, and that is governance. Let's imagine for a minute that the mail service provided by central IT was more on par with something like Gmail, Yahoo or an enterprise-class Exchange service. It is still not going to meet every single need of every college, unit, department, faculty, staff, and student around campus. Without a governance model, there are two problems. First is that if anyone wants to do so, they can still just set up their own mail server to do what they want without considering the real costs and benefits of doing it alone. Second, and a point that I think most people miss when they think of governance, is that there is currently no mechanism for the user population to drive the features that are in the service provided by central IT. In a good governance model, if that centrally-provided service is not meeting the evolving needs of the customers, there is a documented way for them to get their needs evaluated and potentially adopted into the system.
Let's think for a minute of your FLOSS example. With cutting IT costs on campus being one of the goals of IT@Illinois, FLOSS needs to receive more attention. If there is no governance, the concept will get adopted by the organizations that see value in it and ignored by the ones who currently prefer COTS solutions. With governance, if FLOSS becomes the encouraged model, then part of the process for having an IT investment approved could include whether FLOSS solutions have been given adequate consideration and value. Without that though, what is otherwise a great idea with potentially lots of benefit to the University will only see adoption by those with big picture views.
I will finish by saying that having a central IT organization does not mean that everything is dictated down to specific solutions that everyone on campus must use. There should be governance for changes to go up the chain instead of just down as they currently do. Because of the size of our organization and the complexity of our needs, I do not think we will end up with one course management software, one office productivity suite, or one web content management system (although I think there are some things where we might end up with one solution). That does not mean though that all those things need to have their entire technology stack run at the college or department level. Using your example of Moodle, does Education and LIS need to have the physical hardware that is running Moodle? Do they need to administer the entire machine? Do they need to maintain and administer the database behind it? Does even every department need to have code-level access? There are economies of scale that we are missing because we currently force every college and department to manage the entire stack for the IT they need. I know that at GSLIS, there are programmers working on Moodle who also have to worry about maintaining the LAMP stack behind it. What if they could just focus on Moodle and making it do the things that GSLIS needs without having to worry about the new Linux patch that just came out because that part of the stack is being provided by another part of the organization? The concept of centralized IT invokes a lot of responses and assumptions, and I think we need to think about those assumptions and what the real possibilities are if we start trying to imagine something completely new.