,
8 tweets,
3 min read
Read on Twitter
Software people: I'm thinking about implementing dates for my post-scarcity operating system/language/environment. I could easily afford to devote 128 bits to the representation (actually I could allow more, but 128 seems enough). >>>
128 bits gives me slight more than 1 x 10³¹ years at one second resolution, which takes you a very long way towards the life expectancy of the universe. >>>
At millisecond resolution, it still gives me 1 x 10²⁸ years; at μs, 1 x 10²⁵. That's still much longer than the period from the estimated date of the big bang (1.4 x 10⁹ years ago) to the estimated date when all stars will have consumed their fuel (1.2×10¹⁴ years ahead) >>>
So actually I could go down to nanosecond resolution and still cover the entire period of time that people are likely to want to reason over. My question was going to be, is this worth doing? But thinking about it, why not? >>>
So if I use 128 bits, set my epoch 14Bn years before the UNIX epoch, and have micsosecond resolution, does that sound a reasonable use of resource? How would you represent dates, if your hardware was unlimited?
@Science_Hooker, @AstroKatie, what is the range of time which is interesting in astrophysics, and what resolution is interesting to you?
@Science_Hooker @AstroKatie Also, how does the definition of a 'year' change over time? Presumably the period of the orbit of the Earth has changed and will change? Is it interesting to model that, or should I just arbitrarily define a year as 31,557,600 seconds?
@Science_Hooker @AstroKatie @threadreaderapp , be so good as to unroll, if you would be so kind. At your own convenience, of course.
