Conversation started Sep 28, 2013 at 13:21.
user41796
Sep 28, 2013 1:21 PM
@gnat I'm not convinced that there isn't a bug in the way the hotness formula is calculated right now. Or more accurately, my understanding of the time decay function is incomplete and I don't see how it's supposed to work. It's not working the way I would expect it to work though.
 
gnat
gnat
Sep 28, 2013 1:48 PM
@GlenH7 interesting. I need to dig this a bit deeper. But it somehow decays (in non sticky questions), right? :)
 
user41796
Sep 28, 2013 2:37 PM
@gnat - My observations are based upon the listed collider score and what I calculated for the numerator from Jeff's formula. Somewhere around the 7 hour mark after having been opened, the denominator went from less than 1 to greater than 1. So in the first ~7 hours, the time decay is actually acting as an accelerator on the score. And it only starts to slow the growth after that 7 hour mark. The question was constantly active, so I don't think the "last active" portion was doing much ...
 
user41796
... doing much or doing what it was expected to do from my understanding of the formula. That's why I asked in chat eons ago about how it was supposed to operate. I believe Thomas pinged Anna & Shog with my question, but I haven't heard back on the matter. Keep in mind I wasn't promised an answer on that portion of the formula either, so I didn't expect to hear back.
 
user41796
So I think it's supposed to decay the collider score, and it appears to do so after those first 7 hours. But I couldn't divinate the correct behavior of the formula based upon the scores I saw and how the denominator in the formula was expressed.
 
gnat
gnat
Sep 28, 2013 2:56 PM
@GlenH7 a-ha. I think I am beginning to understand. It looks like sort of undocumented magic is going on first 7 hours. Given how it works (on questions that aren't hit by too many answers), it feels like intentional, sort of additional normalization / push for questions that are "too young" to compete with more mature ones. Hmmm I probably understand the motivation, but can not figure how it is supposed to be better than straight denominator, without these 7 hours of magic
 
gnat
gnat
Sep 28, 2013 3:24 PM
oh I got it I got it. 7-hours magic makes it easier for visitors to arrive while question is still fresh, has little or no answers and votes. Then they can watch its activity "live", how answers and votes appear. This adds suspense, fun, intrigue... and, the last but not the least, motivates them to stick around longer than just needed to read question and answers(s) - and, possibly, look at other questions at the site
now this makes good sense to me. The only thing I really dislike is, it's not documented
Glen and gnat Reverse Engineering, Inc. Bring us your hot crap and we'll tell you what it does
 
 
2 hours later…
gnat
gnat
Sep 28, 2013 5:00 PM
interesting how bug in formula damages this feature, just how it did to everything we looked at before. Questions with artificially high score push valid ones out of collider. As a result, instead of intriguing "live translations", collider visitors watch piling of boring, repetitive, zero score answers, on and on and on, over and over and over again
 
 
Conversation ended Sep 28, 2013 at 17:00.