An interesting thought is how when devices couldn't auto-update, they had to work out the gate. I imagine this encouraged companies to do much better testing to reach a gold-plate before deploying.
Assuming these are network devices - it can be harder to certify future working if the network services they rely on become unavailable or when the failure only occurs at scale.
Case in point when 700,000 Netgear routers pinged the University of Wisconsin–Madison NTP server (harcoded IP address) every second.
Less critical, but video games are the same way. Companies will press and ship discs with known broken games and then issue a patch of dozens of GBs day one. The whole point of having a disc is when all the servers are off line, and the store have shut down, the game is still playable.
This is true, but mostly only relevant to the expected user functionality. User acceptance testing in waterfall development doesn't often identify security vulnerabilities.
