The following screenshots illustrates the problem. A comment written "about 1 year ago" responds to a comment written "24 days ago". Either the order wrong or more likely (following from the content) the wrong date is displayed.
Bonus bug: It shouldn't be possible to save empty comments (like the one by Damien Maier).
The frontend currently displays the time at which the answer / comment was last edited rather than the time at which the answer / comment was initially created (the answer in question was edited at that time [1]). That behavior is probably a bit confusing and we can modify the behavior slightly to make it easier to understand. I would prefer to still have the time of the last edit displayed somewhere. We could try to make it a complete gitlab copy by displaying the edit time below the content [2].
Alternatively we could just prepend edited to the a minute ago label in the header if it was edited (if edit time != time).
(Can you edit the title / description if that's what you want? "Wrong" seems like a bit exaggerated because it's not completely wrong)
Regarding the bonus bug (probably worth an additional issue): Although simple cases could be filtered out in the backend we would probably need to do some non-trivial processing for cases like $$ or [LEFT-TO-RIGHT OVERRIDE U+202D] or some weird unicode / markdown shenanigans (idk empty 1x1 images). If we want to prevent that from happening accidentally we can add a client side check, but if a user wants to add empty comments / answers there will probably always be a way to do it.
(If you see something like this as an admin you could remove it)
{"oid":503,"longId":"5cff75cabe9eb5000ebace02","text":"I think that CNAME records do have a TTL. The book (Kurose & Ross) does not say otherwise (as far as I know) and I tried out the following: \"dig CNAME www.wikipedia.org\" which gives as result a CNAME record with TTL.","authorId":"haslebas","authorDisplayName":"Sebastian Haslebacher","canEdit":false,"time":"2019-06-11T09:35:06.660955+00:00","edittime":"2020-07-09T11:41:32.518435+00:00"},{"oid":504,"longId":"5cffe919be9eb50010c18167","text":"Huh TIL, makes sense though, as you wouldn't want a CNAME to persist forever / have a client cache it for longer than possibly intended.","authorId":"tgeorg","authorDisplayName":"Georg Teufelberger","canEdit":false,"time":"2019-06-11T17:47:05.765746+00:00","edittime":"2019-06-11T17:47:05.765746+00:00"},
This is probably less confusing than just displaying the edittime. Copying Gitlab always seems to be a good idea, an alternative would be to just append the edittime to the time.