Search result ranking is an obvious casualty, but it's not the only one. My Firefox start page (as an example) now has a cached but empty version of the HN pages that doesn't really look quite right. Any system that pulls content automatically, like that, or an offline reader, etc. is going to get confused.
It's simply good practice to get the HTTP status codes correct in the ways you outline.
HN was down much longer than I thought it was because the bad pages were stuck in caches. I had to do a force-reload to see that it was back up this morning.
HN was down much LESS time than I thought, because they told browsers that the "Sorry, but we're down" message should stay in cache until 2024! I believed the status report until today, because it didn't occur to me that someone would issue a status report on a temporary site-wide outage and instruct browsers to consider it permanent.
It was also down for much longer than a lot of services (e.g. status checkers) thought, since their only sensible method of working is to inspect the HTTP status. The result being that HN uptime looks better than it really is; of course, this isn't a paid service or something mission-critical, so exaggerating uptime doesn't really make much difference.
same here, the only way i finally got out of bad state was to click on the logo to go to /news
if i navigate to news.ycombinator.com i'm still seeing the "we're down" message in chrome.
It's simply good practice to get the HTTP status codes correct in the ways you outline.