To be fair, in that anecdote you are extracting meaningful information just from observing the literal survivors, which sort of points in the opposite direction

To be fair, in that anecdote you are extracting meaningful information just from observing the literal survivors, which sort of points in the opposite direction to the GP’s point. I’d say examining survivors is useful and can provide information, but you need to be careful to correct for survivorship bias. But that’s not to say you cannot gain any information from examining cohorts of survivors. I think most people probably under-correct for survivorship bias so it’s good general advice to highlight it.