←back to thread

Two HN Announcements

(blog.ycombinator.com)
698 points tilt | 4 comments | 29 Sep 15 18:40 UTC | HN request time: 1.409s | source
Show context
huhtenberg ◴[29 Sep 15 19:33 UTC] No.10298926[source]▶
Dan, speaking of experiments, can I pitch an idea?

    Mix a random story from the New page into the front page on each page load.
This should get more eyeballs on new submissions, with more even exposure across HN populus and ultimately increase the overall appeal of the front page material. Should make the FP a bit more dynamic too.
replies(5): >>10298971 #>>10298973 #>>10298996 #>>10299034 #>>10299038 #
dang ◴[29 Sep 15 19:41 UTC] No.10298996[source]▶
We alpha-tested that and it didn't work. The median item on /newest is far too low-quality (by HN's definition) for randomness to do much good there. In practice, it just mixes junk onto the front page—the worst thing that can happen to the front page.

(I've posted about this a few times; let me see if I can find the links.)

Edit: I found some of them, in addition to the ones scott_s mentioned downthread.

https://hn.algolia.com/?sort=byDate&prefix=false&page=0&date...

replies(4): >>10299044 #>>10299118 #>>10299144 #>>10299167 #
1. scott_s ◴[29 Sep 15 19:46 UTC] No.10299044[source]▶
https://news.ycombinator.com/item?id=9866140 and https://news.ycombinator.com/item?id=8790134
replies(1): >>10299086 #
2. dang ◴[29 Sep 15 19:52 UTC] No.10299086[source]▶
Those links are about an experiment we tried that did work, and which we've expanded. If you get an invitation asking you to repost a story, that's the latest incarnation of this. Where we're stuck is on figuring out a way to let the community manage it that wouldn't just reduce to how upvoting works now. We're working on it though.
replies(1): >>10299193 #
3. scott_s ◴[29 Sep 15 20:04 UTC] No.10299193[source]▶
The latter also mentions the old thing that did not work, where you just pulled a random post for /newest, and put it at the bottom of the front page. I see the current, successful, technique as an iteration of the failed one.
replies(1): >>10299584 #
4. dang ◴[29 Sep 15 20:59 UTC] No.10299584{3}[source]▶
Ah thanks!