How do you balance robust enterprise IT services with computer science as a research area?
This post has been floating around in my head for a while - I even had this started but lost the draft in a tragic overnight OIT loaner laptop reboot. I can't actually answer this question, and trying to has left this draft in my queue for way too long. So this is really some thoughts and more questions.
In a research organization that does CS research, you'll still have an IT department to keep the lights on, so to speak. They run the network, set up new machines, and all of the other typical things. So you have a series of conflicts at lots of different levels. The CS researchers know how to administrate their own machines and also don't like a lot of imposed security things even if they are needed to keep the organization as a whole safe and, well, manageable. Uniformity is much easier to manage.
IT is inherently conservative. The culture of 'no' and all that. For any new thing, there's a matter of deciding it's worth attention, then gathering requirements, then laying out a project plan, doing the project, testing, etc. Everything must have a return on investment. Have to baseline this, compare to peers, see what Gartner/Forrester/Burton Group thinks, talk to 3 different vendors. Then maybe in a couple of years a decision happens. It's easier for the CS people do things themselves, but they're supposed to be working on other things.
The CS folks are more inclined to just build it themselves, but once they've built it, they're not necessarily about the mundane maintenance. According to some of the things I've heard in some of the discussions of software tools for scientists, they often develop these elaborate tools without considering usability and then stop developing them at the prototype stage when the money runs out.
some of the issues:
- proof of concept vs. robust enterprise
- functional vs. standards compliant
- something cool and then move on vs. ongoing support and development
I give up trying to perfect this post, maybe I'll add more later.