Our practices for covering AI leadership decisions have shifted meaningfully, and the specific changes are worth naming.
We started trying harder to separate process from outcome
Early coverage sometimes judged leadership decisions primarily by how they turned out. We now try more deliberately to evaluate decisions based on what was reasonably knowable at the time they were made.
We got more comfortable praising well-reasoned decisions that didn't pan out
We're now more willing to describe a decision as well-reasoned even when its eventual outcome was disappointing, rather than letting hindsight color our assessment of the original reasoning.
We became more skeptical of leadership narratives built entirely around outcomes
We try to be more skeptical of leadership profiles that credit a leader's judgment primarily based on results, without examining the actual decision process behind those results.
Why this evolution matters
Outcome-based leadership evaluation risks rewarding poor process that happened to work and punishing sound process that didn't, and we think correcting for that bias, however imperfectly, matters.