points by dang 6 years ago

This had a major discussion 3 weeks ago: https://news.ycombinator.com/item?id=22855357, as well as 6 months earlier: https://news.ycombinator.com/item?id=21001969.

It's true that this is a new announcement, but that's a distinction without a difference for HN. Since this issue shows up frequently, I'm going to post a detailed explanation of how and why we moderate these, with the intention of linking back to it in the future, and maybe also from the FAQ.

When new releases of a project get submitted, the HN discussion is invariably about the project in general. That's totally fine—there's nothing wrong with it. If we're lucky, there may also be some discussion of the new features—nothing wrong with that either. The problem is when the same discussions keep getting repeated. HN is for curiosity [1], curiosity withers under repetition [2], and frontpage space is the scarcest resource we have [3].

On HN, we handle this by treating submissions of the same-ish story as dupes for a year or so, then allowing reruns after that. What decides whether a follow-up counts as same-ish? Whether or not it contains significant new information (SNI), meaning whether or not there is enough new information to support an interestingly different discussion. On HN, what counts as interesting, basically, are diffs [4].

That is the general rule. Now let's apply it to the special case of project release news. What are the things the thread will discuss? The project in general, for sure, and maybe, if we're lucky, the features in this version. Because those things don't change as a particular release goes through its life-cycle of status updates, status updates don't count as SNI.

If we allowed this, threads like "Foo 1.3.2 Beta 1" would be on the front page all the time, then "Foo 1.3.2 RC 1 is out" a couple months later, and so on. People come to HN to escape repetition—except of their favorites, of course, which they love to see repeated. But everyone's favorites are different, so allowing this in general would guarantee a cranky community. Since most projects aren't your favorite, it would guarantee a cranky you as well, even if you don't feel that way at the moment because you do like Foo.

Of course, there is ceremonial value when a release becomes official. Ideally the discussion could wait (on HN there is no harm in waiting) until the release was ready, and then we could all have a big thread about the release, plus rejoicing. But that won't work. For popular projects, submitters submit the same release every time it gets a new label, and commenters who like the project (or don't like it—that also works!) want to have a thread each time. Since the official release comes at the end of that sequence, it's also the most anticlimactic moment for thread purposes; that's when the odds are highest that the release will already have been discussed. At least for popular projects, the hivemind's preference is clear: it would rather have the pleasure of an early peek than the joy of a feast at the end.

There's one more important aspect: in most cases, there isn't SNI when a project goes from "Foo 1.3.2" to "Foo 1.3.3" either. If it sounds like I already made this point, look again—I haven't, and believe me this is the sort of distinction that users care passionately about, especially with their favorite projects. This part doesn't apply to Inkscape 1.0, but it does to the majority of incremental-release submissions that HN sees, and we see a ton.

The reason that "Foo 1.3.3 is out" doesn't usually count as SNI is not that it isn't significant new information for the users—and especially the developers—of Foo. That's one reason I'm using a silly acronym: SNI! — to convey that it's a specialized use of those words. When we say things like "this is not significant new information, so we're treating this post as a dupe", or even the gentlest, most watered-down and tiptoey version of that language, there are always people who feel aggrieved on the project's behalf, as if we're putting it down or belittling the hard work of its devs. This explanation is for those readers.

Obviously Foo 1.3.3 being out is new, and significant too. But most of the time—looking through the lens that we need to use for HN moderation—the diff between Foo 1.3.3 and the last time Foo came out is not enough to support an interestingly different discussion. Especially for popular projects, commenters will flood into such a thread not to discuss Foo's 1.3.3-ness, but rather their favorite pros and cons, likes and dislikes, about Foo in general. There is nothing wrong with that—it's great! The only problem is repeating it too often. At a certain point "wow! cheesecake again!" becomes "i don't feel so great right now".

The above is mostly about popular projects. For obscure projects, the situation is usually different. Often they're lucky to get any attention at all, for 1.3.3 or 2.15.7 or anything else. Sometimes, in fact, 2.15.7 gets posted when the project itself has never been discussed on HN before. In such cases, we sometimes use a reverse tactic: we'll swap out the incremental release announcement with the project's home page and invite the community to discuss the project in general—which is what the thread would do anyway. Those are fun moments because they happen so rarely. HN gets the juiciest of all diffs to sink its teeth into—the diff between nothing and something—and no one is unhappy with the mods.

Previous versions of this explanation (is that ironic?) are at https://news.ycombinator.com/item?id=22857809 and https://news.ycombinator.com/item?id=19103247.

[1] https://news.ycombinator.com/newsguidelines.html

[2] https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

[3] https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

[4] https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...