order = log(max(abs(s), 1), 10)
sign = 1 if s > 0 else -1 if s < 0 else 0
seconds = epoch_seconds(date) - 1134028003
return round(sign * order + seconds / 45000, 7)
This time and log weighting is why Reddit content went from essays and explainers (just clicked an arbitrary early day) to images and one-liners as the site grew. It’s not just the lowest common denominator of pop culture, it’s that a cat gif or button-pushing headline take seconds to consume. A single “heh” and upvote in the first minute of a story is worth more than dozens of upvotes on a thoughtful essay that takes ten minutes to read. The Wilson score interval used for “best” sorting of comments is why the top comment is often a long, thoughtful response rather than memes that happened to get pasted quickly.
i remember reading on reddit a long time ago, a suggestion of ranking posts based on the level of discussion in the comments, rather than the number of upvotes. it wouldn’t be a perfect solution, but it would definitely be a start
Unfortunately, there’s a failure case (occasionally observed on lobste.rs, which biases story ranking by comment activity) where absolute horseshit articles will get lots of comments debunking them, and end up ranked highly.
I couldn’t say whether it’s enough of a problem to sink the scheme. It’s certainly detectable (high rank with low article score is indicative), so I guess you could depress the rank in that case.
yeah, some manner of combination of score and level of discussion. possibly also taking into account the average length of the comments? serious askreddit threads and subreddits like askscience, askhistorians, etc. would all benefit, effectively hiding subreddits with low discussion levels.
i mean, i’m neither a programmer nor a web developer, so all this might not be as feasible as i think it is, but honestly i think anything would be better than the clusterfuck of a ranking algorithm they have right now.
I find that posting stuff on Reddit is pretty unpredictable. You never really know when people will like it, or hate it.
But this was a super interesting post. I knew that time had some influence on a post’s ranking, but I didn’t know by how much.