Grading in the Age of the Draft

A teacher in Tulsa asked me this in the least rhetorical way possible. She had just read thirty-one essays on the causes of the First World War, and twenty-six of them were fine. Not good — fine. Organised, grammatical, empty. She wanted to know whether her assignment had been measuring understanding for the last nine years, or measuring the ability to produce a certain kind of paragraph.

It was measuring the paragraph. That was true before this technology existed. What changed is that the paragraph became free, and free things stop working as proxies.

Three rewrites

I do not think the answer is detection software, and I do not think it is surrender. It is that the assessment has to sit where the work actually is. Three examples I have run with real classes this year.

1. The literary analysis. Instead of an essay about the novel, ask for an essay plus the two readings the student rejected and why. The rejected readings are where the judgment lives, and a model that has not read the room cannot invent a rejection that sounds like this particular student.
2. The lab write-up. Keep the write-up, add ninety seconds of defence. "Point at the step you would repeat if we gave you one more afternoon." A student who ran the experiment answers immediately and specifically.
3. The history essay. Hand out a generated draft on day one. The assignment becomes: find its three weakest claims, then repair them against the sources. You have just moved the grade from production to evaluation, which is the harder skill and always was.

What this costs

I will not pretend these are free. The third one takes longer to mark than the essay it replaced, because you are reading an argument about an argument. The second requires class time that some schedules do not have. Anyone selling you a costless answer here is selling something.

We did not lose the ability to assess writing. We lost the ability to assess it cheaply, which is not the same loss, although it will feel identical in week three of a semester.

The teachers I have watched do this well share one move: they told the students exactly what they were doing and why. Not as a policy paragraph in a syllabus — as a conversation in week one about what the class is actually for. Roughly half the cheating problem in the rooms I visit is a students-and-teachers-disagree-about-the-point problem wearing a different coat.

The teacher in Tulsa rewrote the First World War essay over a weekend. She reports that marking now takes longer and she can finally tell who understands the material. She considers that a good trade. So do I.