← The five-lens method

Historical parallels

Where have we seen this before?

Why it matters

Earlier decisions can expose mechanisms a new technology makes harder to see. The value comes from identifying the mechanism and comparing the conditions under which it held. An analogy carries less weight where the actors, constraints, or scale are materially different.

Apply the lens

  1. Find an earlier case with a similar mechanism, such as a rule delegating authority or a metric replacing judgment. Include a case that turned out differently.
  2. Compare the actors, evidence, constraints, pace and ability to reverse a decision. Write down where the comparison breaks before drawing lessons.
  3. Translate the parallel into a prediction about the current decision. Look for evidence that would distinguish that prediction from its alternatives.

Questions to ask

  • Which mechanism is shared across the two cases?
  • What makes this case different enough to change the outcome?
  • What evidence would show we chose the wrong parallel?

Worked example

An organization may compare automated purchasing with the approval limits long used for purchasing cards. Both delegate discretion under a threshold. An AI classifier can change rapidly and act at greater volume, so the old monitoring schedule may no longer fit.

One test to run

Review historical purchasing-card exception patterns. Compare them with shadow decisions from the new system and examine which controls still catch equivalent cases.

What to capture

Useful lessons, the limits of each analogy and competing explanations.

Explore the other lenses