Hacker News Reader

Comment by dang | Hacker News Reader

> No discussion of this sort shall be allowed on hn

Here is the largest HN thread from the last 3 months (and the 3rd largest in all of the last year):

The young, inexperienced engineers aiding DOGE - https://news.ycombinator.com/item?id=42910910 - Feb 2025 (2954 comments)

Other recent large discussions include:

Elon Musk's Demolition Crew - https://news.ycombinator.com/item?id=42968430 - Feb 2025 (304 comments)

DOGE employees ordered to stop using Slack - https://news.ycombinator.com/item?id=42951458 - Feb 2025 (355 comments)

Onlookers freak out as 25-year-old set loose on Treasury computer system - https://news.ycombinator.com/item?id=42936421 - Feb 2025 (130 comments)

In fact, this is probably the most-discussed topic of the last month on HN; I can't think of another, offhand, that would come close. (Edit: probably the Tiktok ban business?)

View on HN

@DanG, you replied to my previous comment about this being a conversation that needs to happen.

Thanks for being even handed and giving folks forbearance.

Don’t tell me with a straight face when you have access to the data more than anyone that these posts aren’t getting flagged en mass. We are all watching it happen.

Which part of dang's post are you asserting is inaccurate?

The part where he tries to point to a handful of exceptions to refute the idea that discussions about Musk are getting flagged en mass. It’s really not a difficult bit of text to parse.

Edit: I have to reply to the comment below by editing this comment because I’ve now been banned from making new comments but I couldn’t help but notice you managed to take the original quote and literally remove half of it which would have made your argument look very silly. To be fair Dang did the exact same thing.

Second edit: oh and would you look at that… it’s flagged. Exactly as predicted.

But the claim was "No discussion of this sort shall be allowed on hn", and his data shows that significant discussion is happening. Mass flagging doesn't change this.

That is despite violating HN's "intellectual curiosity" rule.

You know perfectly well that the original top comment was expressive and hyperbolic in nature, stop trying to analyze it as if it were a dry quantitative claim. The point, which is extremely obvious, is that many posts are being systematically flagged for reasons of favoritism.

They're clearly HN relevant, given that they relate to a tech titan effecting drastic changes in government structure, with significant security and other, by technical rather than legal means. Systematic flagging of posts about the activites of the Biden administration would have been equally questionable.

Please don't cross into personal attack.

I haven't. It's a critique of the argument.

"You know perfectly well" is a way of accusing another person of being disingenuous and this is a step into personal attack.

We don't know what others know—that would require being a mind-reader—and it's against the site guidelines to be aggressive about it.

https://news.ycombinator.com/newsguidelines.html

We have a good idea because other people have comment histories and have demonstrated their own linguistic analytical skills on many occasions.

I don't believe that's reliable, but more importantly, you seem to be denying the emotional thrust of what you wrote. "You know perfectly well [etc.]" is an aggressive trope to begin with, and you followed it with other similar thrusts ("Stop trying to", "which is extremely obvious").

Your comment would have been just fine if it started at the "Many posts are being [etc.]"

Weird you are able to characterize 'you know perfectly well' as an aggressive trope without having any special insight into my mental state, but when I characterize something someone else wrote, you want to be all philosophical and say that we can't make any inferences about other people.

My original comment expressed skepticism about the idea that an expressive remark should be interpreted as a formal assertion of fact. I think your efforts at semantic policing are misplaced.

I see how it came across that way, but I was talking about the effects that language like that tends to have, whether it was your intent or not.

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

> Mass flagging doesn't change this.

It literally does. The majority of posters won’t ever see flagged posts.

It's not violating HN's "intellectual curiosity" rule as many people are genuinely curious about this.

Some are (also) dead.

Dead means, to my understanding, killed by moderation.

I don’t know why it’s supposed to be some secret but there is absolutely an effort here from multiple parties to minimise conversation on the topic.

We can debate the reasons and logic as to why that might be but to pretend it isn’t happening is just silly.

You can see an alternate ranking of front-page stories that ignores flags here:

https://news.ycombinator.com/active

As I write, 4 out of the top 10 stories on "active" appear related to Musk, Trump, or both (numbers 4, 5, 8, 10):

1. Meta torrented & seeded 81.7 TB dataset containing copyrighted data

2. Apple Ordered by UK to Create Global iCloud Encryption Backdoor

3. Asahi Linux lead developer Hector Martin resigns from Linux kernel

4. A German court rules: X must provide researchers access to data

5. Elon Musk's Demolition Crew

6. The origins of 60-Hz as a power frequency

7. Stop Using Zip Codes for Geospatial Analysis

8. Announcing the data.gov archive

9. U.K. orders Apple to let it spy on users' encrypted accounts

10. Feds Halt the National Electric Vehicle Charging Program

If you prefer what you see on "active", you can bookmark that as your standard entry point to HN. I flag stories when they appear to be getting repetitive. I haven't actually done that with any of these stories, but I have previously done it with submissions about COVID, OpenAI, and Matt Mullenweg when they come in waves.

If you go on the further pages of /active you'll find a lot of flagged posts. An epidemic of flagging by the looks of it.

I don't think anyone is seriously suggesting there aren't flags, and a lot of them. The argument is that this is pretty much how HN works and is supposed to, for most repetitive topics. You can certainly make the case it should work some different way, either in general or for this topic specifically. But it's not some nefarious plot.

I'm not interested in minimizing conversation on the topic! I just want the conversation to be fresh in each major thread. That requires enough new information to support a substantively different discussion. I didn't see that in the OP and I don't see that in most of the submissions that get posted about this topic.

In case anyone is interested: the way we approach situations like this comes out of what happened with the Snowden avalanche of 2013. It seems obvious now, but at the time we weren't clear about the difference between articles containing significant new information [1] and copycat / follow-up posts [2]. Tons of repetitive stories were on the frontpage, and there was a backlash from users complaining about it. The current principles around to handle this came from reflecting on how we could have handled that situation differently.

What we need here, in order to keep having major threads about this topic, is a story containing significant new information—something that moves the needle on the discussion. If one of those is getting flagged, I'd like to know about it!

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

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

Definitely not—nearly all [dead] posts are killed by software, other than the [flagged][dead] ones which are killed by user flags.

> nearly all [dead] posts are killed by software

What do you mean by that?

Broken link detectors?

I don't get it.

HN's software contains spam filters and similar things that automatically kill certain posts. Those get marked as [dead] and removed from public view. [dead] posts are visible only to users who have 'showdead' set to 'yes' in their profile.

Other posts that have been heavily flagged by users get marked as [flagged][dead]. Those have been killed by user flags, not software.

Have I answered your question?

Yes you have.

Thank you.

Of course. I've spent most of my time this week posting explanations and answering emails about this, so it's a bit odd to hear that I'm denying it, but I understand how these perceptions arise.

It may seem strange, but both of these things are true:

(1) it's the most-discussed topic that exists on HN right now; and

(2) most of the submissions about it are getting flagged.

If you think about how the internet (and HN) work, it's to be expected: there's a tsunami of submissions; some are getting through and producing condition #1; others are getting flagged and producing condition #2.

This has happened many times on HN over the years, and every time it happens, the community splits between the people who want more and the people who feel like it's too much.

For example, here's a case from 2020 where someone felt that HN was suppressing stories about what in fact was the most-discussed topic on the site: https://news.ycombinator.com/item?id=23624962. This dynamic has been around a long time.

Thank you, Dan! I really appreciate all the work you do making HN what it is

How about you just show who is flagging instead of talking around people? It's trivial to show data about moderation here, but you never do. Why?

Voting and flagging are intimate data—I wouldn't dream of disclosing that about anyone. I think most HN users would take that as a betrayal of the basic community contract.

Also, I don't see what good would come of it. It would lead to much more conflict, and would change the character of the site. People would no longer feel free to express their preferences—actually they would no longer be free to express them. This is not a way to fix the problems we face as a community.

Maybe the number of flags a given post has could be exposed? Or other, non PII information about flags, like number of flags from old accounts or accounts with high karma, or showing that a post has the flag shield enabled? Just spitballing here but theres a lack of transparency into the process that throws some people into conspiracy theory land so just imo but it kinda seems like there's got to be some changes that could be done to assuage the fears of the more rational readers, so that we don't get the same boring accusations of some shadowy cabal burying inconvenient stories on every contentious subject. Maybe just linking to the active thread for an MOT for the flagged&dead posts so people can see the topic is being allowed, just elsewhere? Maybe allowing users to mark a thread a dupe instead of just flagging?

(sorry if this has already been asked & answered. I saw the post about turning flags off but haven't seen anything about exposing when it has been)

I see where you're coming from, but I don't believe that one can assuage people's fears or diminish accusations this way. It's the other way around: it would just provide more material, more degrees of freedom out of which to curve-fit those pre-existing views.

I could be wrong about this—my own fears and preconceptions definitely affect my thinking here. I fear trying things that can't easily be reversed, because I don't want to cause damage. Also, I'm suspicious of the idea of technical measures to solve non-technical problems. The conflicting emotions and perceptions that dominate this question seem deeply non-technical to me.

Imagine if this standard was applied to any other topic.

“We’ve already had four Rust topics this month, better flag any new ones!”

This article is not the same news as the others you’ve listed. It’s disingenuous to claim they’re all the same.

Avoiding repetition is a core principle here. It applies to every topic.

I've written extensively in the last few days explaining the moderation approach to cases like this. Let me dig up some links for you...

Edit: here's one post that covers the basics: https://news.ycombinator.com/item?id=42911011. If you (or anyone) want further explanation, here are links that cover some of the principles we're applying here:

major ongoing topic - https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

significant new information - https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

political overlap - https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

downweighting follow-ups - https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

curiosity vs. repetition - https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

turning off flags - https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

All we're doing is applying those principles the way we usually do. That doesn't mean we make every individual call correctly—we definitely don't. (I wrote about that recently here: https://news.ycombinator.com/item?id=42787306.) But the principles are stable, and as far as I know they're the right ones for HN.

If people want us to change how we moderate, it's not enough to demand that we unflag more stories about a specific $Topic that some users feel strongly about. That argument doesn't work because there is always such a $Topic going on.

Rather, you'd need to argue either that we're not following our principles, or that they're the wrong principles (and in that case propose better ones). I'd personally be very interested in such arguments; like anyone who has worked hard on a problem for years, I'd be both surprised and gratified to learn about a better solution. We just need to be talking about the same problem. In HN's case, the problem is how to optimize an open web forum for intellectual curiosity (https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...).

>> it's not enough to demand that we unflag more stories about a specific $Topic that some users feel strongly about. That argument doesn't work because there is always such a $Topic going on.

Your line of reasoning doesn't make any sense.

Flagging the story hides valuable information, and if the topic is always going on then there is clearly interest in the community.

If my point didn't make sense, it may be because I compressed it too much. Here's some background information followed by a decompressed version of that specific point.

The optimal HN front page is not one consisting of stories that the strongest-feeling subsets of users feel strongest about. That would optimize for indignation and sensationalism—definitely incompatible with HN's mandate.

The tug-of-war between upvotes and flags is core to HN's functioning. Indignant and/or sensational stories are often propelled onto the frontpage by 'hot' upvotes and then demoted by 'cold' flags. For the most part, that's the system working as intended. Not having flags at all, which is (I think?) what you're arguing for, would definitely not be in the interests of the site.

There's unfortunately* still a need for moderation, though, because some stories stay on the front page even though they're not so good (I mean 'good' in the specific sense of 'in keeping with the site mandate), and some stories get flagged off the front page even though they are legit submissions that could produce good-for-HN threads.

Now here's a decompressed version of the point you quoted:

If it were a rare thing for a topic to get flagged even though some users care strongly about it, we could just turn off the flags whenever that happened. However, there are always topics going on that some users care strongly about, which other users want to flag. To unflag all of them would (practically) amount to not having flags at all, and that's not an option. Therefore, there needs to be some way of deciding which ones to unflag vs. to leave flagged; and the argument for this cannot just be "please turn off the flags because I really care about this story"—that would be treating one subset of users more favorably than others. Similarly, it can't be "let's turn off the flags because we (the mods) personally care about the story". There needs to be an argument from principle about why a given story is a good candidate for turning off flags (or not turning off flags). If people want to argue that about a particular story, then, they need to either show how (1) this is desirable given HN's principles as I've explained them; or (2) there could be different principles that are better at realizing the site mandate. Does that make more sense now?

* unfortunately, because it would be so much less work if there weren't!

Thanks, that makes more sense.

While I appreciate all the submissions about running Doom on HDMI dongles, it doesn't affect the health of millions of people. The submissions that are flagged inhibit intelligent discussion about items that affect every person on this forum.

There are endless things that affect every person on this forum that are not a particularly great fit for the forum so that in itself isn't much of a practical criterion.

You're right. But this is the perennial question of what sort of site HN is supposed to be. If the criterion for frontpage threads were importance, then HN would be a current affairs site—a completely different place. That's not the mandate we have (https://news.ycombinator.com/newsguidelines.html).

Optimizing for curiosity [1] means including whimsical things like "Doom on a dongle", "how much do all of the ants in the world weigh", and so on, and less space for a lot of more important things. This follows from the mandate of the site.

Sometimes people feel like we should exclude political stories altogether, but that's not right—what we've learned about optimizing-for-curiosity is that some political discussion needs to be included—just not too much [2].

How much is too much? Well, that's the thing that nobody agrees on. For the people who feel most strongly about a story, it's never too much. That's how "literally the most-discussed topic on the site" turns into the perception "all discussion on this topic is being suppressed" [3].

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

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

[3] https://news.ycombinator.com/item?id=42978572

Reply on news.ycombinator.com