Over beers with Patrick at our favorite 97214 watering hole, I discussed my Lightning Talk thesis that universities should be eating their own dog food, by which I meant:
Why outsource the core basics of course registration, instructor assignment and compensation, grading and crediting, scholarships, student and instructor evaluations, to off-site vendors?
Answer (from Patrick): universities have better things to do than be coding huge database systems -- when would they find the time to do any teaching?
Patrick had lots of experience at Dartmouth, and saw from the inside how letting in-house personnel write a compiler for a specific language ended up causing a PhD thesis to come out with the wrong results. A competing institution charged academic fraud -- a serious matter.
Sobered by Patrick's story, I've come away thinking that at least any large institution should invest in its organizational memory, such that newcomers have some way to figure out how it got to be that way (whatever way that it is).
I call this a Book of Lore.
Any large organization, say a hospital, or a system of hospitals, should have its Books of Lore, giving a sense of the heritage and history. What decision-makers reached what decisions? What were the debates?
Based on my own experience, Books of Lore are hard to come by because personnel are uber-cautious with their positions and assessments, don't always want to get out in front, especially with an unpopular point of view.
A Book of Lore, to be successful, need not be about fingering villains and/or uncovering scandals. Moral judgments may be left to the various readers. Universities may stand a better chance of telling their own stories simply because the liberal arts encourage that kind of openness.
A Book of Lore is more just a blow-by-blow, and will be all the more credible and informative if actual emails are cited and included. Failures may be admitted, wrong turns acknowleged, along with lessons learned.
Sometimes it'll be a later analyst, with the benefit of hindsight, who suggests what some of the big successes were. To write history, you need that rear view mirror. Books of Lore may have multiple authors.
Developing a Book of Lore is not always an easy task, as large organizations tend to be compartmentalized. There's an assumption of monitored emails, an archive of some kind (which is what biases the conversations to stay safely neutral), but when push comes to shove, it's often hard to establish that anyone is actually charged with doing the data mining.
A full blown Book of Lore may actually be a multi-media production, complete with interviews, executive speeches, excerpts from board meetings.
Again, colleges and universities may have an easier time of it, given the contentious nature of executive performance -- not that universities can't be contentious!
Putting a company's history on-line may help people steer the company, but at what cost in terms of reputation? A good storyteller doens't leave folks afraid to communciate on the record.
The role of "company historian" goes unfilled.
Even universities, known for their history teaching, will lack an organizational memory in this respect.
Patrick and I did end up agreeing that some larger and/or more gung ho colleges and universities might well choose to role their own, and make the core shared, as in open source. Other universities might use the system with customizations. We've seen elements of this picture emerging already.
Each department may have a strong identity, and want to express itself in terms of the tools it most favors. Some departments may enjoy XML, whereas others may want to stick with JSON. Avoiding a "one size fits all" solution is part of the promise of open source.