Hacker News new | past | comments | ask | show | jobs | submit login

I love this. The search box had an option to autocomplete terms I had recently searched, I clicked on one (a term I was interested in, obviously) and immediately found an interesting HackerNews post. Nice!

I think HackerNews and other link aggregators (e.g. reddit) have a kind of recency problem, where there is a lot of great content, but people only see the recent stuff. This seems like a great way to uncover some of the latent value of old HackerNews content.

Is there a way to suggest content too? e.g. If you liked that, you'd probably also like X, Y, and Z.




>I think HackerNews and other link aggregators (e.g. reddit) have a kind of recency problem, where there is a lot of great content, but people only see the recent stuff.

With Reddit and HackerNews, I want a relative ranking index. I can search by the top content, but something 5th today could have more votes than the top submission of 2015 because of forum/subreddit growth.

I want them ranked by something like the ratio of views to upvotes or upvotes compared to total upvotes for that day.


> I want them ranked by something like the ratio of views to upvotes or upvotes compared to total upvotes for that day.

Yeah, this is definitely possible with public data. I did something similar on reveddit [1] for removed reddit content. Hovering over the graph shows the item with the highest vote ratio [3], and clicking skips to that point in time. Code is here [2] for anyone interested and I apologize in advance..

[1] https://i.imgur.com/p3Bi5IS.png

[2] https://github.com/reveddit/ragger/blob/master/revddit_aggre...

[3] https://www.reveddit.com/r/worldnews/?showFilters=true


Absolutely. HN user 'EvanMiller had a lot to say about this, 11 years ago. His tl;dr is that the ranking score should be "the Lower bound of Wilson score confidence interval for a Bernoulli parameter."

I believe HN's ranking system is extremely creative and works great for the main page and day-to-day use (my understanding is that it additionally makes use of time-decay terms for comments and stories). Like you said, it's really just historical search (i.e. algolia) that seems broken.

https://www.evanmiller.org/how-not-to-sort-by-average-rating...

https://news.ycombinator.com/item?id=15709405

https://www.evanmiller.org/bayesian-average-ratings.html

https://www.evanmiller.org/ranking-items-with-star-ratings.h...

https://redditblog.com/2009/10/15/reddits-new-comment-sortin...


I want a map-reduce over their content that turns them into a canonical “master guide” of their topics.


Agreed! The problem I see is that the score does not correlate well with the perceived quality of the community. I'm researching that for some time now and am in the process of preparing a blog article with data analysis and solution approaches.

Work in progress:

https://github.com/fdietze/downvote-scoring

https://felix.unote.io/hacker-news-scores




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: