On a website full of terminally online autists, people post pretty much whenever they're awake. So, if you scrape their most recent 750 comments, you can plot when they comment and figure out when they sleep. Then, using this IDF-tier intel, you can safely steal their car while they are blissfully dreaming of bussy.
Since we know @carpathianflorist is a darn Yank, he's in UTC-5 EST for most of the data, more recently UTC-4 EDT. We see a small blip in posting at 6 AM when he wakes up, which continues up through 11 PM local time, until he always goes AFD (Away From Drama) from 2–6 AM or so. We also see a prominent noontime spike for lunch. Unfortunately, the rest of the data lacks sufficient statistical power to determine if he has a favored time for an on-the-clock afternoon shit.
Compare that to @bye who is shifted almost perfectly three hours behind
, putting him in UTC-8 PST / UTC-7 PDT on the West Coast. Looks like a 9 PM dinnertime and typically staying up poasting until 2 AM. Imagine: to get info this good a few decades ago, you'd have to follow someone around all day in a suspiciously well-maintained Crown Vic while wearing dark sunglasses. Now they give it away for free!
@Aevann is in UTC+2 EET
. Unlike our other examples, he's obviously a codecel because he doesn't sleep, and he's most reliably active from 4PM to 3AM local time.
Finally, we can use @zozbot to judge general site commenting activity. Ze replies to one in a thousand comments, giving us a truly random sample. It's fairly similar to Carp's chart, so either the site is being carried by East Coast Ameroids, or everyone here except me is actually Carp.
If you want a chart for your own account, ask and you'll get it for free 
. If you want to find out when your rDrama crush is active so you can harass them most efficiently, 100 DC 
gets you whomever you want, either publicly if requested in this thread or privately in DMs with the utmost confidentiality. (Terms and conditions apply: if the account of interest is private, I can't get the comments, and I will keep your dramacoin as an r-slur surcharge and probably post about it to shame you.)
Jump in the discussion.
No email address required.
how do you get the data? i was just curling https://rdrama.net/@aevann/comments?page=123&sort=new&t=all / https://rdrama.net/search/comments/?sort=new&q=author%3Aaevann&t=all&page=123 (with the authorization token header so got json)
it was slow and the pages were rate limited and you seem to do this faster
Jump in the discussion.
No email address required.
GET https://rdrama.net/@user/comments?page=1&sort=new&t=all → regex submatch
timestamp\('timestamp-[0-9]+','([0-9]+)'\)No auth token, so I'm just regexing the HTML. It's so non-compliant you can't actually query the DOM, so thankfully timestamps are fairly unique in the markup. Best I can tell, the HTTP 429 rate-limiting isn't a problem if you sleep 2sec between requests and 30-60sec after 30 requests (hence 30*25 = 750 being the sample size so I don't have to deal with that). I strung a hundred sloppy lines of golang together to do the whole username ↦ chart pipeline, though my zozbot post a bit back was just babysitting CURL around the rate limit. Probably not fundamentally different than your approach.
Jump in the discussion.
No email address required.
So what you're saying is that with a bit of scripting I can draw a peepee on your chart?
Jump in the discussion.
No email address required.
There are 24 free parameters in the graphing, all purely the y-heights for the bars. Most of the bits of info are lost during histogram binning. You could maybe make a pair of low-res twin peak boobae (proper nominative plural of booba btw since it's obviously in the first declension) if you try really hard. Though
@grizzly's scatterplot setup would be much more peepeeable.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
More options
Context
More options
Context