Cohort labs without hero fixes

Every lab includes at least one ambiguous signal that could point to two root causes. Instructors resist the urge to swoop in with the answer because the hesitation is where operational judgment forms. Participants pair so they hear another person's mental model. Disagreement is expected; we grade the quality of evidence, not speed. We also disclose limitations: some production-class problems need vendor support or data we cannot replicate ethically. The lab stops at the boundary and documents what a real escalation would look like.