29 July 2018

So there was this article...

It was mostly someone with a disability talking about why it's hard to get the existing system of (alleged) supports for someone classified as they are classified to impose "allowed to use a laptop" on the professoriate.

So, let's see.

We've got a system that functions to turn conflict into a fight for control, which in effect means the existing organizational system exists to maintain control.

We've got at least three judgement-driven taxonomies.  (The profs, if they agree (snicker-snort), the support system, and the person, all have a different judgement driven taxonomy about what counts as "disabled".)

The entire notion of "disabled" rests on the presumption of a prescriptive norm which all meet.

There's a lot of wrong involved.

You can have success, or you can have control.  It's an exclusive "or".  This is a robust result from operations research back in the 1940s, which is probably much of why everyone hates operations research.  As an effective way to prevent change, this kind of control fight is quite excellent, and any hierarchical system is run by people determined to prevent change.  (If you're at the top of the heap, change makes you worse off.)

Judgement-driven taxonomies are inevitably materially false.  Arguing about them turns into exercises in social power. (For a semi-safely-historical example, look at the Cope-and-Marsh bone wars of the late 19th century.) 

Prescriptive norms are a tool of authoritarian control; "you must do this".  The norm -- presuming it doesn't come with a methods discussion, careful peer review, and error bars -- doesn't represent the population; it represents what the person in authority wants.  (Even with the error bars, it represents a specific population at a specific point in time.)

(No, I don't think a kid trying to get an education should start off knowing this stuff; that doesn't mean their sympathetic outrage results in an effective choice of tactics.   Especially given an environment where the older, wiser, and institutionally funded people have made such terrible errors and created a system to prevent effective change or more useful results.)

What to do instead?

Make sure the system has materially defined success criteria which are the thing used to generate feedback.  This is NOT "graduates X percentage of entering students"; it's something closer to "nothing is ever more than Y difficult to do for anybody".  (There are several devils in those details, but evaluating teaching is not about "is it possible to learn this?", it's about "did we make this more difficult than the material requires?")

Crush all use of judgement-driven taxonomy; taxonomy, outside comprehensive and testable statistical approaches, is a tool for concretizing bias in ways that make it harder to argue with.  (This is a system constraint; no judgement-driven taxonomies.)  For electronic devices in class, this goes two ways; of course you can have them, of course you're not going to use them in ways that make anyone else's learning more difficult.  ("difficult" needs to be measured, not asserted.)

If you want to constrain infrastructure to wider utility and to do it effectively, you need to specify a broad range of actions in specific material terms, you need to include actions the owners of the infrastructure want (e.g., deliveries), and you need to make it widely known that being out of compliance makes it unlawful to charge for services on those premises.  So, for example, the courier companies and the post office can specifically and without penalties refuse deliveries to commercial entities if the force required to move a set mass of package up the ramp is too high.  A legislator interested in doing a good job puts this in the commercial building code, makes it retroactive, and sets the amount of force based on a broad study of mobility aids.  ("special" is not only bad politics, it's materially wrong; the practical difference between someone with a mobility aid, someone with a baby, and someone with a lot of packages is small.)

No comments: