Rounding times
6 years ago
Québec

In the frame count spreadsheet, I used a m:ss.000 format (Google Docs doesn't support a m:ss.0000 format) to display the different times.

This leads to a rounding to the nearest 0.001 of all the times on the spreadsheet. However, we typically round all the times down - in any game that times its runs down to the second, a 12:59.6 would always be considered as a 12:59 rather than a 13:00, even though it's more accurate to say 13:00.

For example, Kosmic's time is a 4:56.4617576. It was automatically rounded to .462, but technically, it's faster than .462, so we would typically round it down to .461.

On one hand, rounding down would be more in line with what everyone has been doing with timing speedruns. On the other hand, when getting down to thousandths of a second, rounding down is kind of a gimmick, and it would be closer to the true time if we round normally.

I'd like to have opinions on this, and if there is a consensus to start rounding down at the 0.001 level, I will update the spreadsheet accordingly. In that case, statistically, 50% of all runs timed to the millisecond would find themselves "gaining" 0.001 seconds due to the new rounding method.

Edited by the author 6 years ago
Quivico, KingOfJonnyBoy and 4 others like this
United States

yeah i think "stopwatch" time is always truncated, not rounded. But it seems a bit late to change policy now. The game only really needs to be timed down to the hundredth of a second anyhow....

Darpey and KingOfJonnyBoy like this
Oklahoma, USA

Since you can't finish a run on every millisecond anyway, only every frame which is 16/17 milliseconds, it seems more important to keep doing it the way it has been done... For uniformity, otherwise you'll have to retroactively change at least all the times below the threshold for including milliseconds, or end up having some of the duplicate frame finishes off by a millisecond. Just my thoughts. It's simpler just to cut off the remaining decimals rather than round. You're talking about an accuracy issue that is 3% the time of a single frame.

Edited by the author 6 years ago
coolestto and MrTim131 like this
Québec

I like this answer. Since no one can get a 4:56.4614 anyway, there's no point in truncating the times. In this case, accuracy is the way to go.

Game stats
Followers
7,784
Runs
8,790
Players
1,781
Latest news
Requirements for High-Level Any% Runs

Any% (NTSC) runs below 4:57.000 must now fulfill additional requirements in order to be verified.

  • The run's full session must be included in the submission description.
  • For emulator runs below 4:57.000, some form of input display must be visible for the duration of the run. A hand-cam or input
3 months ago