A bit less cynical take on this is that reliably assessing if you are in outage situation is hard. First of all not so easy to define. Secondly, hard to model i

A bit less cynical take on this is that reliably assessing if you are in outage situation is hard. First of all not so easy to define. Secondly, hard to model in the monitoring. In classical attitude to monitoring you can always err on the side of bit too many alerts (oncalls may take a few false positives). Having said that, AWS has surely the brain power and capacity to do that right. Maybe they just lack motivation.