> I actually have a hard time imagining a scenario where a fuzzy date is more helpful than the exact one.
It can be, in niche scenarios. I have an application that pulls information in from another database, reformats it, and displays it for the user. The update is expensive and time-consuming, so it has to be triggered manually by the user when it is needed. In this case, the exact time of the last update isn’t particularly useful; the only thing that matters is how old the data is, which is to say how likely it is to be out of sync with the source of truth. In some cases, information needs to be as up-to-the-minute as possible, so if the info is just a few hours old it needs to be refreshed. Other times, information that is 4 or 5 days out of date or longer is totally fine, so there is no need to trigger an update that could take as long as 10 minutes.
For this particular purpose, fuzzing the date is actually more useful that providing the time of the last update, since order-of-magnitude precision is good enough, and it saves the user from having to compare the update time to the current time and do a calculation. Admittedly though, this is a rare case, and the vast majority of the time a relative fuzzy date is less helpful.