It does not cancels out. It is not just simplification, you have to grossly overestimate traffic deaths now and underestimate accidental deaths in the past. (Basically regardless of when and where you place "the past".)
> Actually even now these early years are most risky ones, just the scale is different.
Babies mortality used to be huge too. And maternal mortality used to be huge by our standards too.
Actually even now these early years are most risky ones, just the scale is different. Now I have witnesses bias here also - in my 40ies I can expect to live 50 years more, very different from a newborn with immediate terminal condition. So lifespan does have different meaning for us.