4 ms·

The time and votes don't explain this, and "more comments = bad" doesn't either. Because [1] has 27 points with 8 comments from 3 hours ago, whereas [2] has 28

The time and votes don't explain this, and "more comments = bad" doesn't either. Because [1] has 27 points with 8 comments from 3 hours ago, whereas [2] has 28 points with 7 comments from 1 hour ago, and yet [1] is ranked #8 and [2] is ranked #13. I don't see how any of those explain this. AFAIK this story never went above ~#10 either, so it's not like it's been hanging around forever. It really looks like something/someone has been either artificially boosting other stories or artificially penalizing this one to prevent it from going near the top. [1] https://news.ycombinator.com/item?id=25494109 https://news.ycombinator.com/item?id=25494109 [2] https://news.ycombinator.com/item?id=25493495 https://news.ycombinator.com/item?id=25493495

jagged-chisel 6y ago
I didn't say "more comments = bad", I said comment traffic affects the ranking. Actually, a busy comment section improves the ranking of a story (again IIRC.) I think time is the most aggressive part of the formula - it would take a fantastic number of votes and very busy comments to keep a story on the front page for a full 24hrs.
- mehrdadn 6y ago
  I mean either way I still can't use that to explain this. Can you? Again, we have 8-hour-old stories with fewer votes and comments ranking higher right now.
  - jagged-chisel 6y ago
    I can only say these things are variables. I have no idea what the weights are. For all I know (I'm sure they tune the thing without publishing details) it now involves some kind of sentiment analysis. But rather that attempting to reverse engineer the process, I recommend getting in touch with HN and expressing your concern - if enough folks feel strongly about this story, it'll be buttressed a bit.
    - mehrdadn 6y ago
      What I'm saying is this is beyond a "weighting" thing. You don't need to know the precise algorithm, but you can make some reasonable inferences (e.g. more votes probably doesn't penalize). If you just watch this story on the front page, it's acting as if it's hard-capped at #10, and where it's happy to stay for a while (though occasionally going to #11 and then coming back). It really looks like someone has interfered with it rather than it being an algorithm thing, unless the algorithm imposes deliberate hard caps on some stories. And given that it seems to have had some manual interference, I'd rather they address this publicly; everyone is entitled to know. (Update: And just as I finish writing this comment, it's at #13, whereas the following 3-hour-old story with 19 points and 1 comments is somehow at #9 now: https://news.ycombinator.com/item?id=25493577 https://news.ycombinator.com/item?id=25493577)
dang 6y ago
Correct, time and votes aren't sufficient to explain story rank. This is in the FAQ: Q: Why is A ranked below B even though A has more points and is newer? A: You can't derive rank from votes and time alone. See "How are stories ranked?" above. From "How are stories ranked?": Other factors affecting rank include user flags, anti-abuse software, software which demotes overheated discussions, account or site weighting, and moderator action. https://news.ycombinator.com/newsfaq.html https://news.ycombinator.com/newsfaq.html.
- mehrdadn 6y ago
  Right, thanks, that part I had figured out. The detailed weightings don't concern me, but could you let us know if this involved moderator action (vs. automatic), and whether the ranking was hard-capped?

jagged-chisel 6y ago

I didn't say "more comments = bad", I said comment traffic affects the ranking. Actually, a busy comment section improves the ranking of a story (again IIRC.) I think time is the most aggressive part of the formula - it would take a fantastic number of votes and very busy comments to keep a story on the front page for a full 24hrs.

mehrdadn 6y ago
I mean either way I still can't use that to explain this. Can you? Again, we have 8-hour-old stories with fewer votes and comments ranking higher right now.
- jagged-chisel 6y ago
  I can only say these things are variables. I have no idea what the weights are. For all I know (I'm sure they tune the thing without publishing details) it now involves some kind of sentiment analysis. But rather that attempting to reverse engineer the process, I recommend getting in touch with HN and expressing your concern - if enough folks feel strongly about this story, it'll be buttressed a bit.
  - mehrdadn 6y ago
    What I'm saying is this is beyond a "weighting" thing. You don't need to know the precise algorithm, but you can make some reasonable inferences (e.g. more votes probably doesn't penalize). If you just watch this story on the front page, it's acting as if it's hard-capped at #10, and where it's happy to stay for a while (though occasionally going to #11 and then coming back). It really looks like someone has interfered with it rather than it being an algorithm thing, unless the algorithm imposes deliberate hard caps on some stories. And given that it seems to have had some manual interference, I'd rather they address this publicly; everyone is entitled to know. (Update: And just as I finish writing this comment, it's at #13, whereas the following 3-hour-old story with 19 points and 1 comments is somehow at #9 now: https://news.ycombinator.com/item?id=25493577 https://news.ycombinator.com/item?id=25493577)

dang 6y ago

Correct, time and votes aren't sufficient to explain story rank. This is in the FAQ: Q: Why is A ranked below B even though A has more points and is newer? A: You can't derive rank from votes and time alone. See "How are stories ranked?" above. From "How are stories ranked?": Other factors affecting rank include user flags, anti-abuse software, software which demotes overheated discussions, account or site weighting, and moderator action. https://news.ycombinator.com/newsfaq.html https://news.ycombinator.com/newsfaq.html.

mehrdadn 6y ago
Right, thanks, that part I had figured out. The detailed weightings don't concern me, but could you let us know if this involved moderator action (vs. automatic), and whether the ranking was hard-capped?