Hacker News Reader

Comment by TZubiri | Hacker News Reader

But if every clock was like that, then 12:59:30 would be the new 13:00:00

But not every clock would be like that - only those clocks which don't show the seconds precision would use this rounding.

The consequence of that would be that statements like "fireworks start at 12 AM" would mean two different points in time depending on how much precision your clocks have.

That's true if you catch the exact moment the clock changes. If you don't, the only thing you know with a truncating clock is that the fireworks started 0 to 59 seconds ago. With a rounding clock, you know the starting point is within [-30, 30] from now. So on average, you're closer to the starting point when seeing the clock show 12AM.

A good reason for truncating is that we have a strong bias against being late, but not really against being early.

> If you don't, the only thing you know with a truncating clock is that the fireworks started 0 to 59 seconds ago.

More importantly, I know whether the designated point in time passed or not. If you have a submission deadline set at 14:00, the most important thing you care about is whether you made it or not.

But what bothers me more is that this is inviting ambiguity into time definition. What 14:00 means (in addition to all the timezone complexity) now depends on the type of the clock you use. That's just bad.

> A good reason for truncating is that we have a strong bias against being late, but not really against being early.

TBH I don't understand what you're saying here.

So if my watch shows seconds, I'd be late at 12:59:31?

If someone without seconds on their clock starts the meeting, yes.

This is a weirdly satisfying thought, if entirely unrealistic in a corporate setting: what if meetings in general were more of a consensus thing, and less of a rigid time slot thing.

No, this is the point of the article. The person without the seconds could not tell if you were late or not when his watch says 1:00. A display of 1:00 means the actual time is between 12:59:30 and 1:00:30, so there's a 50% chance the person arrived early. Only the person with the second hand could tell for sure.

It's a bad point. If the clock says 1:00 and you aren't there you're late.

Your boss won't like hearing "well technically I wasn't late because you see your clock has the seconds rounded to minutes so you can't actually know if I'm late".

The best possible outcome you could get from that is that the meeting starts at 12:59:30 now.

I'm simply communicating the point of the article. Take it up with the author if you don't like it or don't agree. And he'll tell you that if sub-minute precision is necessary (i.e. someone would be reprimanded for being under 30 seconds late to a meeting) the simple solution is to include a second hand on those clocks, and there won't be any confusion. (Of course, what's the likelihood of a basic office wall clock being accurate to within 30 seconds anyway, but that's a whole other rabbit hole)

How does 12:59:30 floor to 13:00:00? Wouldn't that be the result of ceil?

Only until the next article saying that "all clocks are 0.5 seconds early" and we then switch to randomized rounding.

I still think that flooring would be better; however, if you did insist to do this rounding instead then you could use a different convention for numbering seconds with e.g. -30 to +30 instead of 0 to 60. However, I think that this is not worth it, and that the existing use of flooring is much better, although if you want such precision with timing then you really should display the seconds, rather than using a clock that does not display seconds, anyways.

Reply on news.ycombinator.com