For example, we might be able to say, given a bunch of events or documents of the 17th century, when most of them happened or were written, but we might have no idea whether something happened in 2000BC or in 2001BC (just random numbers, I have no idea where this distinction might appear).
Hi there!
I'm afraid this really isn't a question which can be answered because it's based on a false premise. There is no watershed before which the date of events could not be ascertained and after which they could be. The date of a particular event in Ancient Rome - say the assassination of Julius Caesar on 15 March 44BC, is well established, but the date of a particular event in, say, the 1940s, might not be. It's not a question of time but of evidence - if a 2000 year old event is well recorded, then it's well recorded. If a ~50 year old event is not, then it's not. If you're instead asking about when events started to become "better" recorded, it's still unanswerable as a "how many grains of sand is a beach" sort of question. While we can say, for instance, that the advent of writing helped us record events better, and electronic recording allows us to record them better still, we still can't put clear "dates" on any of these trends, and of course there continue to be poorly recorded "events" the precise timing of which will never be known.
If your question is meant to be more about timekeeping in general and how different events, dates and calendars can be synced up and placed on a timeline, you may be interested in this thread.
As the other poster said - there's no precise date from which point forward we know all of the dates. One thing that determines whether we know the specific date of an event or not is eclipses. When a source mentions an eclipse, we can track down that event to the minute because eclipses happen on a fixed schedule, and we can then find out the dates of events that we have knowledge of the the timeline in relation to events that happened during an eclipse.