Week 1 Session: Week 1: Fact, Inference, Unknown

Building a timeline from the Engagement Signal Pilot's dated documents, and learning to say which of three kinds of claim each entry actually is.

22 February 2027

Before the session#

This course’s central skill is reading a paper trail without letting it read you first. Before you can ask whether a decision was pre-made, you need to be able to tell, entry by entry, what a document actually proves versus what you are filling in yourself.

Three kinds of claim, used all semester:

• Fact — a dated document states this directly, and you can name which one.
• Inference — a reasonable reading of the documents, but one more step removed from what any single document states outright. Someone could disagree with it without being unreasonable.
• Unknown — the public record does not settle this either way. Saying so honestly is not a failure to find the answer; it is itself a finding.

Read the case file below — ten dated entries from the Engagement Signal Pilot’s first six weeks, Slop University’s fictional staff engagement-scoring rollout, which the whole course follows. For each entry, write down which of the three kinds you think it is, and one sentence why.

Case file: the Engagement Signal Pilot, January–March#

“Sign-off memo” in the source column is the document’s own title, not a claim about what authority it actually proves — Week 9 comes back to exactly that gap.

This is deliberately not a real institution’s leaked records — it is teaching fiction, invented for this course and reused across the semester so the case can build rather than restart each week.

In the session#

In pairs: compare your ten classifications before either of you sees the answer key. Where you disagree, that disagreement is the exercise — argue it from the document, not from what seems plausible. Then build a single shared timeline document from all ten entries, in order, each with its agreed kind and a one-line justification. Keep this; you will use the same format for a real decision in the Decision Autopsy assessment in Week 6, and the decision-autopsy template uses this exact structure.

What good analysis of one entry looks like#

Take the “between the two drafts, someone decided the item should read as settled” entry, dated only to the 1–3 March gap between the drafts either side of it:

Kind: inference, not fact. No document names who made this call or states a reason. The inference is supported by the two drafts existing at all — something changed the item’s framing between 1 and 3 March. But a different, equally reasonable inference is available: the underlying question may have been genuinely settled in an earlier, undocumented conversation, and Draft A simply reflects that a decision had already been reasonably reached, not that it was concealed. The public record supports “something changed” far more strongly than it supports “why it changed” — and does not by itself support either “the process worked as it should” or “the process pre-empted the room.”

That last sentence is the standard the rest of the course holds you to: naming what the evidence does and does not support, including the possibility that it supports no strong conclusion at all.

A little further reading#

Any research-methods or library “evaluating sources” guide covers the primary/secondary source distinction this exercise builds on — your own university library almost certainly has one online. Read the part that distinguishes a primary document from a claim inferred from it, and bring one question to class: where does an inference drawn from several primary documents together sit, if it isn’t quite either?

After the session#

Keep your shared timeline document. Week 2 adds a second document type — an invitation list — and asks who a decision like this one actually included.