3 ms·
Have you tested the behaviour of your algorithm on pagination and dynamics of link position over a day? I assume the good thing about the snowball effect is th
by pyentropy 5y ago
Have you tested the behaviour of your algorithm on pagination and dynamics of link position over a day?
I assume the good thing about the snowball effect is that "making it to the frontpage" gives you guaranteed and pronounced exposure over a day or two - even though it's not maximizing the number of good links shown.
- manx 5y agoOur simple simulation had a 1-second time resolution, so it evaluated the formula for all link positions over the submission's lifetime. But there was no pagination. We're working on a more accurate simulation that will have many features of real Hacker News, like pagination, new-pages and so on.
- manx 5y agoI'm thinking again about your question. I think you're talking about the time span a submission is visible on the frontpage, right? This can be controlled by the age exponent in the formula and (in our proposal) by the upvote-threshold between new-page and front-page. If only a few very high quality posts make it to the frontpage (because of an upvote threshold), they will have more exposure.
- Wowfunhappy 5y agoRight, and that "pronounced exposure" is actually super important! It takes time for interesting discussions to form in the comments, as new people read the story and jump in to share their thoughts. That doesn’t happen when stories disappear too quickly.