1. It clearly indicates that automatic systems are failing to detect the outage. 2. It also indicates that no one is aware about the incident to manually signal the outage (or that there is no manual override).
Basically, it makes a difference between "yeah, shit happened, we know (and maybe working on it)" and "hah, they don't even know themselves".
Partially because these large systems have some kind of ongoing issue at any given time, so it's challenging to provide a meaningful live status that isn't a bit misleading and could cause misdirected panic.
Partially because you don't want to give potential attackers (eg ddos) any insight if/how their methods are affecting your systems.
Partially because there are SLAs and reputation at risk, and you don't want to admit to any more downtime than you absolutely have to.
Of course, in the real world, I don't think there's a single IT admin who didn't just start nervously sweating and pulling at their collar after reading the above, imagining it as something their CEO said and encouraging them to target it. Nobody can really do this — and that in turn says something important about how those metrics really look in most systems.