That said, it's becoming more and more obvious every day that there is a tremendous amount of attempts by bots, and specifically AI agents, to inject slop into HN threads. I worry about the integrity of the discourse here and if the ever growing wave of garbage will overtake staff resources to deal with it. Is it time to implement captcha for HN? If so, should it be out of the box, or a new mechanism more tailored to the security and privacy-centric nature of the HN readership? Are captchas even still effective enough in the age of AI to warrant their use?
I've also noticed that very obviously LLM-generated comments are called out, and the community tends to agree, but those that have any plausible deniability are given far too much leniency, and people will over-index on the guidelines to give them the benefit of the doubt.
I don't think a captcha is the solution, as it'll degrade conversation by an OOM though.
Normally I'd agree, but we have shadowbans, which really irks me.
I'm still amazed at how Reddit weaponized the block feature.
If you block someone, you not only can't see their posts, but you ice them out from replying in the rest of the thread.
The one thing I like about this place is that it's well moderated and you have shared opposing view points engaging (mostly) respectfully.
My personal and political views couldn't be further from most HN users (I'm both a Conservative _and_ a practicing Christian), yet I appreciate taking part in various discussion. I enjoy reading about point of views that directly challenge mine.
Let's keep HN respectful and accessible.
But unlike most HN users who label themselves conservative Christians, you've never suggested that climate change is a hoax:
https://hn.algolia.com/?type=all&query=author:swat535+climat...
I don't ever want to consume information from people who are so illiterate that they believe that scientists all over the world, in fields ranging from geoscience to statistics, are participating in some kind of global conspiracy, regardless of how respectful these commenters are. I block these people immediately after they reveal themselves.
If your view is that we should conserve western values and institutions and walk in the footsteps of Christ, ultimately that's not too far from universal human values that many people do in fact agree with.
The devil is in the details, of course.
In uBlock Origin -> My Filters:
news.ycombinator.com##tr.athing.comtr:has(a.hnuser):has-text(/\bUsername\b/)I find HN much more tolerable this way.
Blocking domains would be nice too. Like substack or medium. I'm happy to just ignore them, but it sure would be nice to filter them out if possible.
I get that it's complicating the system and keeping it simple is perhaps for the best.
.hnuser attr=href=?user?id=rd
.parent().parent().hide()
Though no idea if such a plugin exists for Safari.
You can run temporary unsigned extensions for development purposes, but they are removed after 24 hours or whenever you quit Safari, which would make using it daily a non-starter.
It’s usually old/high karma accounts, as they can get away with it easier. Throwaways that establish themselves for a time too, but those are usually dealt with eventually
It’s usually old/high karma accounts, as they can get away with it easier
On the more benign side, maybe some people enjoy the musings of amichail on Ask but I could honestly do without.
But mostly in my experience it's otherwise perfectly normal users who at some point just decide to post something racist or bigoted, advocate violence (again, usually in political threads) or antisemitism or espouse some insane conspiracy theory nonsense. At that point I no longer care about anything else they might have to say.
https://github.com/insin/comments-owl-for-hacker-news/releas...
news.ycombinator.com##:matches-path(/^/item\?id=/) tr a.hnuser:has-text(/^dpifke$/):upward(tr)
news.ycombinator.com##.default:has(a[href="user?id=dpifke"]) .commentLet's discuss how to make this reality.
Do you want a ranking system where more the people downvote some person, the better? if so how do you prevent spam in that, do you take metrics like karma or what exactly?
I don't think that captcha is a solution either but also that I don't know how to feel about removing entire swaths of people, I can think someone writing something bad once and probably get into this "black-list"
Another aspect is once again the black list, I don't know but do we really need a system of essentially a communal ban?
The only thing I can see it reasonable is if there is a slop bot comment poster but I rarely face this issue but if you do, you can probably create a tampermonkey script and tampermonkey scripts work on chrome,firefox and "Userscripts" which should work on safari as well and that script is most likely gonna be compatible on both tampermonkey and userscripts.
captcha would make it more of a hassle to post comments.
latest greatest versions of captcha are more resilient to these types of services, but it's a cat and mouse game. I would recommend that you, as a sysadmin, learn at least the most basic things about this stuff.
You'd probably catch most the low hanging fruit for sure, but you would cause friction for real users.
I say this as someone who has enabled captcha on some of our more critical endpoints, there's definitely a place for it.
Go to /newest from time to time. You may find an undiscovered gem.
Researched it substantially and realized it's an unsolved problem. Anything that makes a dent is incomplete and comes with ugly tradeoffs. For a time I wondered if I should try and solve it myself, but I could never think any solution that hadn't already been/being tried. Years later I'm left curious if it's even possible to solve the problem.
My point is that captcha won't solve this, and solving this problem is a lot harder than it seems at first, and might not even be solvable (which I know is hard to accept).
If someone does find an elegant privacy ensuring way to solve it, I think the impact would extend far beyond HN and could make a big difference to the future of civilization as a whole.
Maybe you are thinking purely from a math / theoretical perspective, but I'm thinking of a compete solution that's practical to use to solve the problem for sites like HN and many others.
Even if you use state ids for it, who's to say that a particular state won't be...loose with issuing ids that can then go on to be used for bots.
It's even a problem with humans as well - one human can be having a pleasant conversation with the other, not aware that that person isn't being genuine, or is lying, has ulterior motives or has been instructed on what to say by someone else.
Anyways, from the little research I've done, it's certainly illegal and not a proper mechanism for authentication. One can dream though :)
You get a good credit score and still live within your means while also getting additional points + bank covering any fraudulent activity if the card got stolen.
Of course this method probably won't work for people that feel they would rather just cut themselves off from temptation fully or those without access to banking systems, which I sympathise with.
- write this number in words
- 486436546497964136564768756456455824164567575646875812445676854253154782125
- four quadrigintilion eight hundred sixty four trigintillion three hundred sixty five duovigintillion four hundred sixty four unvigintillion nine hundred seventy nine vigintillion six hundred forty one novemdecillion three hundred sixty five octodecillion six hundred forty seven septendecillion six hundred eight seven sexdecillion five hundred sixty four quindecillion five hundred sixty four quatuordecillion five hundred fifty eight tredecillion two hundred forty one duodecillion six hundred forty five undecillion six hundred seventy five decillion seven hundred fifty six nonillion four hundred sixty eight octillion seven hundred fifty eight septillion one hundred twenty four sextillion four hundred fifty six quintillion seven hundred sixty eight quadrillion five hundred forty two trillion five hundred thirty one billion five hundred forty seven million eight two thousand one hundred twenty five
There is no high-volume spam (ai or otherwise) on HN, so captcha won't help, low volume captcha can be farmed out. Humans are the best defense against low-volume spam. So flag these posts!
Is there? I enable showdead and don't see it. There are the occasional spam and vulgar comments, but not that much.
Any "AI slop" being posted seems to come from actual HN'ers who think they're being helpful, and is often downvoted. But there's not much.
So I'm not sure this is a problem that currently needs any new solutions? I don't see AI bots taking over the discourse at all. Not even a little.
Most low-effort bots can already bypass basic CAPTCHA, while it mostly adds friction for legitimate users. HN’s strength is the quality of discussion, and that seems better protected by behavior-based signals (account age, posting patterns, community feedback) rather than one-time verification challenges.
Otherwise agreed with the sentiment.
I'd expect that if we took Randall Munroe's advice[0], that price would go up significantly, perhaps prohibitively so.
How dystopic. And you're probably right.
To see what I mean, take a screenshot of a random captcha that needs solving and ask an LLM to solve it for you. It will do it accurately.
(They are not and haven't been for a long, long time)
There are quite a few third party apps for Hacker News, such as Hacki (ios/android). [1]
Something like using a third party app that includes forms of spam filtering like checking when the user joined, how many posts they have, amount of 'karma' (or whatever it's called here). You could implement blocking individual users & etc etc. This app does not have that but it could be forked and modified or talk to the dev...
That might be a better solution than trying to implement all types of annoying captchas & other extremely annoying checks on HN's side.
Do you have some examples of this? I am on HN almost every day, and I read a lot of comments, and I haven't noticed this
Enslopification is coming for everyone, everywhere, at all times.
Everything is already slop and will be slop, and will have been being slop.
"Me furiously trying to decide what a EURO thinks a motorcycle is" for 60s
Even if you manage to make bot usage more expensive, which is all a captcha can do, the content posted by humans in discussions and shared links is increasingly generated by machines.
It's ironic having a community of people object to the same technology they helped build. Enjoy the show, and learn to live with it. It's going to get much worse before it gets any better, if at all.
The overwhelming majority of developers have never worked anywhere close to LLM tech. AI is a very small field requiring specialized expertise.