I've noticed that now some timestamps have a full stop at the end. Like some of the submitted articles on the front page now say "2 hours ago.". What does that mean?
This is one of those things that pops into your head and takes 2 minutes to deploy, so I just did it. It was late last night and I forgot to not turn it on for everyone. Then I thought it would be fun to see what sort of discussion I'd wake up to.
Not sure whether to keep it. Advantage: it's a concise way of displaying some surprisingly useful information—useful to mods, at least, but I think maybe also to readers.
Disadvantage: it's obscure. The inconsistency will drive some people nuts. We'll have "Ask HN: Why do some times on HN have a full stop at the end?" threads for the rest of our lives. (No one reads the FAQ. The FAQ, you say? Why yes, at the bottom of every HN page: https://news.ycombinator.com/newsfaq.html...)
It also leaks when a user posted something later and deleted it. We could fix that, but as so often in software, it would make the thing way more complicated. So I'll probably just drop it.
The reason I noticed this feature is that it adds a pause when I'm reading through the HN front page with my screen reader. I don't really care either way, but it did make for a jarring change to the rhythm I'm used to when listening to that page.
I find these kinds of terse shorthand features valuable: they allow for a richer experience for users who care enough about HN to make the effort to understand the features that lie beyond the basic stuff, and for those who don't, they're basically invisible.
I think perhaps something to consider would be a `title` attribute on a span surrounding the full stop which says something like "most recent comment by this user", so that someone who mouses over the dot can get more information about what it means (thus hopefully avoiding the Ask HN threads)
This makes me think you have a lot of moderation tools like this. Is there any way you could discuss the kinds of tools moderators have that most users don't see? It would be interesting to read about, though I guess it also would increase the chance of gaming the system...
Yes, the moderation tools have developed a lot over the years. I have a Chrome extension written in Arc (it has gone through a number of Lispy incarnations over the years, but we implemented an Arc-to-JS transpiler to get the last version and it works nicely). I rely on it heavily to navigate around the site. Eventually I'd like to open-source it and/or make it an option for any user to wants to load that JS to do so.
I use both, but at the time I first wrote this software it was easiest to make it be a Chrome extension. This was long before FF adopted Chrome's extensions API. It's on my list to see if I can get it running in FF but, like most things on that list, it never seems to get too close to the top.
I don't prefer this feature to remain for end users (those who are not mods). If someone wants to know the most recent post by a user, they can visit their profile and find it. I do not see value in making that easier, and for a reason I cannot explain well, it also triggers a sense of exposing someone's habits, behavior, etc.
That's a subtle point but I think you may be making the right distinction here. There's a difference in both style and ethos between the regular view of the site and the kinds of things that moderators need to keep track of.
For the front page, it doesn’t seem to add much insight to know that someone’s post is their most recent HN content. But in a thread of active discussion, in which a someone’s top-level comment gets highly upvoted and kicks off its own fascinating tangent...sure, it’d be neat to know (vaguely) at a glance if the commenter has or hasn’t yet been on HN since their thread went off
It triggers mild OCD reactions for me, since there is inconsistency between the formatting of posts, a bit like missing apostrophes do, and as noted in a sibling comment affects how screen readers cover it.
I think this is right, where "most recent post" is chosen from both comments and submissions. Looking at top of the https://news.ycombinator.com/newest page illustrates this pattern. Almost every timestamp there ends with a dot, except for the cases where the submitter commented on their own submission, in which case that comment has the dot. For example: https://news.ycombinator.com/item?id=25198319.
Stakes are high! At the end of the trail of clues, the Council of Immortal Wizards will ask you to join their ranks, at which time they will confer immortality upon you. There are only 10E100-1 steps to go!
Haha, this was great. I arrived at an article about “John Frum” which did sound like a secret, and was quite the rabbit hole [1].
It got me worried about history. The whole story sounds like 90s zine fiction, which I think it is - there is no record I could find beyond the original story from 1995 on Harper’s Magazine [2], yet it is now in Wikipedia, stated as a fact.
The wikipedia article includes references from a couple of decades earlier, at least; e.g.
> Nat. Geographic: May 1974. "Tanna (Island, New Hebrides, South Pacific Ocean) Awaits the Coming of John Frum (cargo cults of Melanesia since about 1940)".
> It got me worried about history. The whole story sounds like 90s zine fiction, which I think it is - there is no record I could find beyond the original story from 1995 on Harper’s Magazine [2], yet it is now in Wikipedia, stated as a fact.
Did you look at the references section in the article?
Off topic: is it new for HN to elide urls this way? I’ve never noticed it before but it really annoys me. I like that HN is just text, and preprocessing it this way adds no usability. It makes it harder to copy a URL, if anything, and it certainly isn’t prettier.
Personal preference I suppose. There are a lot of urls where the text of the url (or large parts of it) conveys no meaning. In that case it breaks up the comment without adding value. Sometimes you see people use a footnote format here because even the short url is in the way.
What's the name of the effect that when you click on something random you still end up on something related to you and you think "it's not that random". Because I got that https://en.wikipedia.org/wiki/Microbridge
That article "Microbridge" should really be deleted because it's not notable, but it looks like it was proposed in 2011 and failed so it isn't eligible for WP:PROD and needs to go through the full WP:AFD process.
This reminds me of a time I wrote a Facebook post which tagged about a half-dozen couples I know. I consciously wrote all the het pairs as girl, boy e.g. "Sarah Smith, Dave Smith" not "Dave Smith, Sarah Smith" because on the one hand that's how I most often think of them, and I noticed our culture usually puts the man first so why not be contrary?
Anyway, Facebook re-ordered them. And so I followed up with a post asking why that happens. It turns out some of my friends saw other orders, some got the order I intended. Intriguing.
Fortunately one of the scientists thought of a perfectly rational and testable explanation, she tested it, and demonstrated it's apparently right.
Facebook has some internal "How much you interact with this person" metric and it's using that for sorting. It has no way to know your relationship to Sarah and Dave in the real world, it's just putting Dave first (for example) because the two of you exchanged a dozen terrible fish puns last week, a measurable Facebook interaction, whereas that walk out by the lake with Sarah when her mother died isn't on Facebook.
Makes sense as FB hyper optimizes for engagement. That said I wonder how the data is structured in their DB, how messy it is, how complicated a post can I make to foil their parsing.
There was a rumor back in college when we cared about social clout that the handful of people you see on the Friends list on your profile are the people who most recently/frequently check your profile (i.e., viewing someone's profile is already an interaction metric for FB). This trained me to not visit other people's profiles. To this day (when that list is just "most recent friends added"), I visit profiles very rarely and only if there is really something I want to check.
It's not so baseless either. Of course this might be some bias or another working but there was a time I fell behind certain commitments to a Dev Group I was helping organize and poof, the friends list on my profile started to show my co-organizers. Felt like they were checking on me, on why I'm behind my stuff, why I was taking time to echo replies. I was just busy; sorry it happens. :)
I'd seen it claimed repeatedly that when others view your profile it doesn't affect People You May Know. I don't believe the claims. Despite being in the same location three days a week for several weeks with my fellow classmates, it was only after I gave a presentation to the class and my (unusual, non-English, impossible to spell phonetically) last name appeared on screen that I started seeing my classmates in People You May Know. It was only female classmates and IIRC we didn't even have mutual friends. Like you, I've been selective about going to people's profiles since then.
When we used to use Facebook in high school there were tonnes of fake apps that I assume just existed to suck up data that would claim to be able to show you who viewed your profile the most too. There was something floating around that you were meant to be able to put in the DevTools console for the same thing - this was probably before they had the big warning there and presumably also sucked up data or worse.
I imagine that algorithm isn't really trying to rank people you know vs. other people you know, but rather trying to list people you know ahead of acquaintances and friends-of-friends that you've happened to "interact" with once (in the sense that you just happened to be in the same comment thread at the same time, even if you weren't speaking to one-another.)
It is in fact traditional formal style, at least in the U.S., when writing the name of a married couple with a shared last name, to put the _woman's_ name first: "Jane and John Doe". In contrast, when using the titles it is the man's title first: "Mr. and Mrs. John Doe".
Not to detract from your main point, which is very sound, but on this:
> whereas that walk out by the lake with Sarah
If you both have the Facebook app with location tracking on, Facebook does know about this, which is why you sometimes get friend suggestions for people after you spent time near them (eg you were at the same party or they visited your house to do cleaning or plumbing), even if you have no mutual friends.
Absence of the period (aka full stop) after the timestamp appears to indicate that a commenter had previously (perhaps recently) posted a comment which was subsequently flagged. Or that the commenter is new (see below).
Then I reviewed some of the posters' recent commenting history. Those that had no period after the time-stamp had recent grayed-out comments. I did not find any commenters with a period who had recent grayed out posts.
For instance, here's a page from esteemed commenter buran77's profile: https://news.ycombinator.com/threads?id=buran77 . You'll notice that not only does buran77 have grayed out posts without the period-after-timestamp feature but so does grayed-out commenter elmo2you.
The absence of the period also appears to apply to recently created accounts. See green commenter jn6118 in this thread.
This appears to be a subtle indicator designed to be used in HN's comment moderation. If so, what can we glean from this quirk of HN's comment moderation procedures or policies?
This is one of those things that pops into your head and takes 2 minutes to deploy, so I just did it. It was late last night and I forgot to not turn it on for everyone. Then I thought it would be fun to see what sort of discussion I'd wake up to.
Not sure whether to keep it. Advantage: it's a concise way of displaying some surprisingly useful information—useful to mods, at least, but I think maybe also to readers.
Disadvantage: it's obscure. The inconsistency will drive some people nuts. We'll have "Ask HN: Why do some times on HN have a full stop at the end?" threads for the rest of our lives. (No one reads the FAQ. The FAQ, you say? Why yes, at the bottom of every HN page: https://news.ycombinator.com/newsfaq.html...)
It also leaks when a user posted something later and deleted it. We could fix that, but as so often in software, it would make the thing way more complicated. So I'll probably just drop it.
Edit: dropping it.