It was pretty easy for orgs with affected systems to produce concrete examples for themselves. Maybe to everyone else it seemed vague, but for the people who had to deal with it, it was taken pretty seriously from top to bottom.
It was thankless work that is still glossed over and waved away today, but it was all a very big deal throughout the late 90s.
All I can say is that at my level in an org, if I want to say “instead of developing X this quarter, I want to test the effects of 2038 on Y,” that’s a far easier conversation if I can say something like “in similar embedded devices they crashed and wouldn’t even respond to OTA updates” vs “something bad could happen. Not sure.”
The latter is just a ripe plumb, left to rot in the backlog.
That's nonsense. Orgs spent time and resources on it because they grabbed a test server and demonstrated it caused problems. It's not some weird ethereal untestable bug. They set the dev server to a minute before midnight and went "oh shit".
That and quite often the problems started showing up years before 2000 itself. "Hey, the scheduler is giving me a meeting 80 years ago" type weirdness when it crossed the boundry.
It was thankless work that is still glossed over and waved away today, but it was all a very big deal throughout the late 90s.