Nothing to prepare — the personas and their partial information sheets are
handed out in the room. Resist the urge to Google what a real used-car lot
looks like; the metaphor works better if you stay suspicious on priors
alone.
Given a lot of fictional personas, each with a few visible attributes and
several hidden ones, choose who you’d pair with, then compare notes with
someone who saw a different subset of the same lot. The reveal at the end
shows why the average “car on the lot” looked worse than the average car in
the population — the mechanism, not just the moral.
If you found yourself reasoning “the fact that they’re still available is
itself information,” you’ve derived adverse selection from the exercise
rather than being told about it — which is the point.