Being a tool of the Man

I've been working exclusively at places with .edu in their addresses for the last seven years or so; The .edus have their own hierarchies, codes, and bureacracies, but as you know, they are an order of magnitude smaller than those in the Real World. Now that I'm working for the third largest employer in DC, it's taken a bit of adjustment.

The usual trappings of tooldom are banal but ignorable. The office takes up a full city block, so the walk to the water cooler or the underpowered microwave is a long one, an expanse of speckled grey carpet. The tie is irrelevant. I have so many other pieces of individuality that matter (biking to work, running my own IT system outside of theirs, rearranging the furniture) that the problem of wearing nice clothes seems an aside. Yeah, I'd rather not wear it, and I have to change in the bathroom downstairs every morning, but I never understood why the tie became the symbol of oppression when there are so many other dumb, silly customs that are truly onerous.

We're working on a report on migration (i.e., immigration and emigration), and it's an entirely different world from the academic. My boss stresses the fact that economists are storytellers; avid readers of this site will recognize this as a word choice away from my own insistence that all academic papers are a persuasive essay.

The difference is in the story that one chooses to tell. One of those standard jokes about the Scientific Method:

And ya know, it's kinda true for the theoretician. It is not unheard of for a researcher to write down a model hoping to derive a result, and then finding that the result needs a half-dozen caveats to work, or that the exact opposite is actually true. This is an honest means of deriving good theory, in which reality advises the math and then both advise the results. The same is true for a qualitative survey or any other sort of study whose primary intent is not apophenia-crushing.

But the revised scientific method will not work for a two-volume report covering a dozen aspects of the migration process. If the report is to work out to a coherent story, then the broad outline of the story needs to be written before the research is anywhere near completed. So, what happens when the data disagree with the story?

Strategy one: prevent this outcome by devising a hypothesis so bland that it can't possibly be false. This already probably explains the format for most of these big overview reports, which are a great resource for tables and figures, but which manage to say approximately nothing. The base hypothesis of our report is that cooperation among countries instead of snippy bickering will be beneficial to all involved. If you're somehow not convinced, you can read the supporting evidence in volume II. [OK, there's a little bit more to it...]

Strategy two: maintain enough flexibility. Before writing this, I had often complained that my Boss's story kept changing. He'd ask me to write up a model to show that the black market for labor can be dried up by various changes one week, but the next he's all about the demography of an aging Europe-but then the next he's lost interest in that and wants me to look at remittance rates. From the perspective of the story here, his vacillation is trying to manage the fundamental conflict between telling a coherent story before the data is gathered and the story reality wants to tell. Although I'm often frustrated, I guess he's doing as well as one could.

Strategy three: spin! I've been doing a good amount of this, since my model produces a lot of null results. I usually put a header in front in the way of: the model shows that all the people who think an incremental change will lead to the sky falling are wrong. Instead, nothing ever happens.

While we're on the subject of spin, it's worth distinguishing between the sort of report here and reports from people like FAIR (Federation for American Immigration Reform). FAIR will never publish a report that supports any loosening of immigration laws of any sort. Of course, this is not their official policy, but I've read much of their stuff for the last few years, and that's certainly what turns up. There are other comparable fanatic organizations in any field and on both sides, and their story is rigidly fixed and will not change to accommodate data. What does such a group do when data appears which directly opposes their position? Strategies one and two (be bland or flexible) are out, which leaves strategy three (spin!), and perhaps another:

Strategy four will be used by both the bland megastudy and the fanatics. After all, one run of my model will give me about 15MB of data, which could potentially say something about any of the above topics; but in the interest of telling a coherent story, I'll be supressing 14.98MB of that data. The same holds of the surveys and regressions and other parts of the study: they're all flexible enough to tell a story in any field, but will be used for only one. In that respect, this suppression is sort of like not talking about the paint job on the car that hit you--some detail is not really apropos to the story.

Suppression doesn't get insidious until one finds data that isn't just not-apropos, but actually contradicts the story. Hopefully, strategies one and two kick in, and a good researcher finds some sort of compromise between the data and the story; for the extreme organizations for whom strategies one and two aren't available, their only resort is to lie by omission, which we see more than often enough in such reports.

Gosh, this was going to be a me day, but it's turned into an abstract essay on the conflict between coherent storytelling and data gathering. Um, I love the view (see Figure One) and spend a lot of time staring out the window. I'm having trouble with life in a paperless office, because it just looks really empty. Turns out that the solution to the problem is a plant: one plant makes the whole space look suddenly occupied. I've taken the advice of Mr. KO of somewhere in Japan and have not putting a dozen cards and pictures on the wall to compete for my attention, and instead just have three widely-spaced prints that I got from the Corcoran. And a tea set. I spend most of my time staring out the window and typing on the usual little laptop, which makes me more than content.

From time to time I have to turn around and deal with people and officeness. The secretary is brilliant, and knows exactly how to deal with the bureacracy, for which I am eternally grateful. E.g., she knew exactly what to say on the forms to get the funding for me to buy my badass server. I can not express the joy of having a good bureaucrat handling it all for me.

As for the badass server, this is the conflict between academic computing and business computing. Academic computing realizes that academics are whiny bitches and if they want to run a 1993 copy of WordStar, there's no way you're gonna talk them out of it. Academic computing is therefore built around the fact that the IT department is infrastructure, like the janitors, and if you don't pay them enough, then it'll suck to work there, but they're fundamentally there to make things comfortable for the others.

[As only sort of a digression, here is a commentary by a Microsoft programmer, who wrote key portions of Windows and has been nonsarcastically described as the smartest programmer on Earth: "I don't like installing software on my computer. The more programs you install, the more likely there's going to be a conflict somewhere."]

The business computing approach is to minimize effort on the part of the IT staff. They have their hands full worrying about 100% stability at all costs, and don't want to waste their time with handling conflicts between applications. Standardization saves the IT staff time at the cost of taking up time for the people using the PCs. From a Battlin' Busines Units perspective, this makes complete sense, maximizing the IT staff's time budget.

Oh, I've fallen into abstract conflicts again. But we found a way to get me a badass server which does not fit the specs of what they can buy from Dell. It's gonna run Ubuntu, since I've been frustrated with Mandrake and have been wanting to switch to the Debian side for a while. Finally, it's gonna be here at [name of university], because the network at my place of work is too locked down to host it. I.e., the business computing setup at work is such an impediment to scientific computing that I need to locate my own system in another state to get work done. I'm delighted that my boss-man and the bureaucracy-savvy secretary fully understand this, even though they're not scientific computing people themselves. In short, they've been very helpful about getting me those things that I need to get work done (window, badass server, tea), and so I am a happy tool.