Data mining and the college campus... or, We Shouldn't Go There

I thought I was pretty good at writing and composition, and banging out a decent argumentative essay. But I think I have gotten lazy in the past ten years.  It is not often one needs to write a good rhetorical essay in the office environment.  At most, I have had to write a product analysis or argument for purchasing software, or an e-mail requesting (nagging?) the recipient for information in order to do my job.  So when I say the English class I am currently taking at JWU is stretching my comfort-zone considerably, I am not exaggerating.  As it is my goal to tie in every class with my work and with the other classes to the maximum possible extent, when I get a class assignment for one, I look to the other, look to the job, and then open up the laptop and start mapping out what I'm going to do. 

So.  Here's my assignment:  write a letter to the editor of a paper or journal of my choice, responding to an Op-ed piece that appeared in the same publication.  I went online, went to the Chronicle for Higher Ed and found this article. Oh! Yay! An article about data mining, that elusive subject so dear to statisticians and researchers everywhere... I can fit my FIT class in this one! And oh, it ties back in to my job as an Insitutional Research Analyst! How great is that? I gave it a good read: In "Mining Student Data Could Save Lives", author Michael Morris argues that violence such as the Virginia Tech shootings could be prevented by passive surveillance of students campus e-mail and web searches.  Well.  My first reaction? What, is this guy crazy? I mean, the beginning of every really cheesy, or really good, sci-fi film USUALLY begins with harmless relinquishing of privacy by the unwitting public, who then become primitive zombies to the machine.  Oh, am I EVER ready to argue this one! When I wrote a brief to Professor Tamborelli, the letter practically jumped out of my head, ready to be written with original refutation of this dangerous idea.  But now, having read the article a second time and the accompanying comments, I realize that a.) I am not the only person who thinks the author did not do the appropriate research, b.) I am not the only person out there who can show evidence that this is a really bad idea and c.) I am essentially not qualified to refute, as what do I know about data mining? I mean, I haven't got there yet in my courses at JWU, and frankly, there is terminology used in statistics that is dauntingly out of reach for me and remains a pretty, shiny object flying undeterred over my head.  In other words, I would love to be up there arguing with the big boys and girls - the psychologists working on college campuses with hard evidence that you simply cannot predict behavior with a computer; the data guy who built complex data mining systems for a corporate entity that still created false-positives and knows that a system that would be required for such a project would be unwieldy.  So on what grounds do I base my argument?

Okay, well, first, as a student, I find it kind of appalling.  Aren't we, as students, supposed to be in this kind of magic protected bubble land where we can go out and research to our hearts' desire, and sort of, to some extent, be free to do so? Even if our research includes scary keywords such as "gun" or "bomb"?

As a citizen, it just goes against everything that the Privacy Act says.  Do I really want my school to know that I accessed a nail fungal photograph in order to show a classmate an example of an environmental hazard in science class? Out of context, what does that nail fungal photograph say once it is data attached to my name in my record? Who is the person who will be analyzing that data and what is their agenda? What if they were foot-averse and deeply offended by the photograph?

As a parent, it concerns me.  My daughter is a freshman at an art school.  They have to look at naked people, both in real life and on the internet.  I shudder to think what such data might "tell" about the mental state of my child, when taken out of context of her school work.

I'm not an expert, but the idea stinks.  The biggest concern is, okay, now you have the data, but what will you do with it? Sure, there are very kind people in data management.  And there are devious, political, angry, deluded, power-hungry and downright nefarious individuals who could be handling this data as well.  Could it prevent a tragedy? I am not so sure.  Let's say I had a scanner that could read everything my fellow students in my FIT class were doing on their computers over the course of the month.  I compile this data in to a database and rely then on key words (I am guessing) that might indicate a behavior change or "red flag" that might indicate this student needed help.  However, as a "blind" administrator, I might aggregate this data and flag a record for a student who searched on a string that raised an alert, coupled with a blog post that seemed, let's say, dark.  How could the combination of these two items, or three, or however many data markers I as an administrator might use, tell a complete story about an action that has a very low percentage of happening?  And even if you had an explicitly expressed thought or ideation of violence, that does not mean that violence is imminent.  So as an administrator, what action could I even take?

How to get all of these thoughts in to way, way fewer words, and all in the form of a letter to the editor, before Sunday night? We shall see... meantime, it is a subject that will continue to come up as our technology proves more capable of allowing for it, and as Universities strive to cover risk.