DNS over Wikipedia | Hacker News Reader

DNS over Wikipedia

github.com

290 pointspyinstallwoes82 comments

Interestingly, Wikipedia editors are aware that Wikipedia articles are used to find the current URL of sites that are forced to change URLs frequently due to legal or moral issues, and they face the same dilemma registrars and service providers face.[0] So although it seems somewhat more resilient than search engine companies to demands from copyright holders, it's not uncensored, something to keep in mind if you're infrastructurizing it for that purpose.

[0] https://en.wikipedia.org/wiki/Talk:8chan/Archive_2#Inclusion...

Yeah for example 8chan and kiwifarms are usually censored. I'm not that mad about it, some censorship is always necessary (you don't want links to child porn), but it's weird that Wikipedia pretends there is no censorship. And it's kind of arbitrary.

Why is stormfront - an openly nazi forum (a really old one at that) - allowed, but kiwifarms - an anti-trans doxxing forum - isn't? It's both bad

Do they pretend there's no censorship? I don't see that. They block spam and I'm certain no one objects to that, so the bare fact that they exclude some information clearly does not constitute the status "censored"

I'd imagine the reason kiwifarms gets different treatment is because the site is a lot worse than the descriptor "anti-trans doxxing forum" might make you believe — it's a website designed specifically to facilitate long term stalking and harassment campaigns. Trans people are their flavour of the month right now but a few years ago it was anyone disabled.

https://en.m.wikipedia.org/wiki/Wikipedia:What_Wikipedia_is_...

https://foundation.m.wikimedia.org/wiki/Resolution:Controver...

If there is some official policy which links are allowed and which are not, I'll shut up.

Why are some links allowed and some not, what is the policy, if there is some.

I see

> Wikimedia projects are not censored. Some kinds of content, particularly that of a sexual, violent or religious nature, may be offensive to some viewers; and some viewers may feel such content is disrespectful or inappropriate for themselves, their families or their students, while others may find it acceptable.

which seems to me against link censorship.

> Do they pretend there's no censorship?

Yes, they do. Censorship of official links is against explicit Wikipedia policy[1], but it doesn't matter because every policy can be overridden by consensus. In practice this means that a handful of professional activists can (and do) censor it as they see fit, since they can determine for themselves whether such a "consensus" exists.

[1] https://en.wikipedia.org/wiki/Wikipedia:External_links#Offic...

The Mediawiki environment became increasingly hostile to "the external world" though.

I am making a research project on grammatical gender in French, that I host on Wikiversity (there is a dedicated research space there). Lately I get an increasingly large number of rejection of saving my contributions, because some sites are considered "unreliable sources". But in my project, I am looking to document what people use in practice in their written exchanges. That they express lies or try to spread disinformation is irrelevant from the linguistic perspective I’m conducting this project. But due to this software enforced policy, I get prevented from documenting my sources from time to time.

If they're spreading disinformation, obviously they also aren't accurately representing their own speech patterns. That's just common sense.

> If they're spreading disinformation, obviously they also aren't accurately representing their own speech patterns.

But GP is not documenting the 'true' speech patterns of the people spreading the disinformation, but rather the speech patterns they use when they are spreading disinformation (which, as you pointed out, might be different from their normal speech pattern). So the sources are still good enough for that.

Is it really true that Wikipedia doesn’t have a formal, credible, method of determining whether a “consensus” exists?

It’s true that there is nothing which should work in theory, and yet mostly does in practice.

> an anti-trans doxxing forum

That is reductive. Kiwifarms is a shit hole for sure, but it has more than just anti-trans doxxing. It's like classifying 4Chan based on /b/

It’s unclear how you meant this, but it reads as

> Sure, the Klan has some bad hombres, but we also run Checkers Tuesdays, maths tutoring for underprivileged youths Friday mornings, and Bible study Sundays.

That is, at some point, the bad overwhelms any good.

And sometimes the good is used to spread the bad, or the whole. Cults sometimes recruit like this, abusive people can be nice at first, narcissistic traited particularly, violent fringe groups offer camaraderie, yakuza doing charity and relief supplies. So I agree 100% that it really matters where the good is coming from.

>That is reductive

I agree, Kiwifarms is much more shitty than that. To quote from Wikipedia "It now hosts threads targeting many individuals, including minorities, women, LGBT people, neurodivergent people, people considered by Kiwi Farms users to be mentally ill or sexually deviant, feminists, journalists, Internet celebrities, and video game or comics hobbyists."

Do you see the irony in quoting Wikipedia's view of a site to show that Wikipedia's stance on not linking that site is OK, in a discussion about how Wikipedia is censored? Have you actually looked for yourself instead of just trusting the propaganda?

I'm a bit annoyed by it. Censoring "lawful but awful" speech is the thin edge of the wedge. An existing precedent of censoring legal websites reduces my confidence that Wikipedia will be able to stand up to censorship pressure (including from its own editors) in the future.

So called "lawful and fine" speech don't need free speech protections, nor any protections for that matter. It's precisely the so called "lawful but awful" speech that do.

The category of speech most in need of free speech protection is "unlawful but fine".

Let me put it this way: Nobody is going to censor fine speech, FSVO fine.

I think the argument in this case is that it may cross the line into unlawful behavior. Kiwifarms has been linked to suicides, and encouraging suicide is a crime. 8chan has similarly been linked to violent crimes.

There are cases where speech is illegal, even in the USA, which probably has the strictest standards for protecting speech in the world.

If you think a website is doing something illegal, you can report it to the police or the FBI, depending on what type of crime it is. Kiwifarms has a US corporate entity controlled by a US citizen, it isn't like this is some tor darknet market hosted in Moldova or something.

Generally though sites aren't responsible for their users' speech, so if someone does cross the line, that would be on the user, not the site. As long as the site responds to any lawful subpoenas, they would stay in the clear.

So if a person advocates (for example) murder on an American site, this is fine until the police say it isn't? That is not a standard that 99% of the internet follows, and for good reasons.

The US legal system is 100% wholly incapable of keeping up with the pace of internet content for this sort of thing, so embracing the spirit of the laws on speech and applying them within user-content-based-sites is an appropriate minimum.

Even Musk who wanted to turn Twitter into a site dedicated to free speech specifically said he wanted to focus primarily on moderating content based on US laws (something that he has apparently walked back since then since Twitter still aggressively moderates legal content).

That's the whole point of Section 230. Service providers generally have immunity with respect to third-party content posted by their users. If a user posts something, it's their speech, and the user is therefore held responsible for it, not the website. Section 230 is what makes an internet of user-generated content possible.

https://www.eff.org/issues/cda230

I find it very disturbing that the Wikipedia thread posted in the post above discusses the topic like it's about CP. It's clearly not. It's completely about the political implications of 8ch, especially in the aftermath of the connection between the Christchurch shooting by Brenton Tarrant.

I have browsed 8ch extensively in the past, and continue to browse 4chan. You'll be exposed to disgusting imagery from time to time, no doubt, but the idea that 8ch is censored because of illegal and disgusting imagery is so incredibly disingenuous, this is clearly about political censorship of right wing extremism.

If I had a bit less faith in humanity I would even go as far as to suggest that the Wikipedia thread is crafted to be about CP and not politics for the sake of justifying censorship and rewriting history. 8ch was not controversial because of CP, it was controversial because of extremist politics, and attempting to rewrite history like this is just so typical of Wikipedians these days.

This is an extremely dishonest and factually untrue comment.

No, it is not.

the guy who made it said that he found it reprehensible but wouldn't remove it, so yes promoted pedohilia by turning a blind eye in the name of "free speech"

And the reason why they're even having the discussion is not because the site might contain CP. The reason they're having the discussion is because the board was host to a political discussion board which hosted right wing extremists, and various kind of censors in the west don't think these people should be able to act freely on the internet.

CSAM is literally what got it delisted from Google search results in summer 2015

And this is why the website even has a Wikipedia article? No, ofcourse not. The site is known because of it's political board: /pol/.

Its Wikipedia article actually exists due to to both gamergate and CSAM as evidenced by the first actual commits. https://en.wikipedia.org/w/index.php?title=8chan&diff=634741...

>8ch was not controversial because of CP

Huh?

>Google appears to have taken an unprecedented step in filtering its search results by banning an entire domain—and adding a warning about __"suspected child abuse content"__ to a search for the domain itself.

Emphasis mine.

https://arstechnica.com/tech-policy/2015/08/8chan-hosted-con...

It is controversial because of the history of CSAM and the history of extremism.

Stormfront pushes white supremacism in a generic way, Kiwifarms targets individuals by name (and address and date of birth and…).

The issue with most search engines is that on top of not showing the proper link, they push the malicious ones on top of the search page. Wikipedia can simply not link.

Cool idea but it's unfortunate the Chrome extension can read and change data on all sites you visit (as well as the intrinsic phishing risk for a DNS/search substitute). There's a history of such extensions ending up in the hands of bad actors.

Ironic that open-source extensions become obscenely dangerous once you install them from the Chrome store, a platform that's meant to improve security. The store should let you install directly from a repo and that should really be the default distribution mechanism, just as Obsidian does with plugins for example. The store could still monitor said repo for suspicious changes, cautiously roll out new updates, etc.

Not to claim open source is perfect (as recent events illustrate), but transparent code is far safer than opening the kimono to a black box.

It sucks that Chrome's permission model requires "read and change all data on all sites" for so many plugins. I don't mean to trivialize the difficulty in compartmentalizing permissions, but I wish we'd see some progress on it.

It’d be nice if chrome users could manually override the plugin permissions (downward only) in the browser gui and see if an extension still works as needed. Even better if those permissions would sync between browsers.

you can restrict plugins to only run on sites you actually enable it on.

I had to look up “opening the kimono” and it’s kind of an interesting tangent.

https://www.npr.org/sections/codeswitch/2014/11/02/360479744...

Seems my initial vibe on reading that phrase was exactly correct, thanks. Hope to never read or hear that expression again.

Wait, even after Chrome's extension security got revamped into something so strict that it functionally killed adblockers, extensions still have to grab that vague-ass permission for decent functionality? Seriously?

Another interesting application for wikipedia is translation of technical terms, although one has to be careful with that (I wouldn't trust a automated translation).

Translation services often suck at translating technical terms, but if you find your languages wikipedia page on it and switch the language there is a good chance the title gives you a nice pointer. This doesn't work always, but if you are proficient enough in both languages you won't have a hard time figuring out if it is correct.

An issue with this is when articles exist in another language, but you can't link it. Nowadays it isn't as easy as just filling in that O is the Dutch article about what is called Eaux in French, you need to visit another Wikimedia project called wikidata and enter a translation there. But if the target name, O, is already in use, you need to first find a new data object name for O and migrate the wikidata object to the new name and then create a new one for the translation of Eaux. Or maybe the problem was that the wikidata ID for the Dutch article was already linked to something else (so O points to wikidata ID 123 and Eaux points to wikidata ID 321). Either way, you need some migration which first orphans one of the two sides iirc.

It has been a few years so I forgot the details, but this is where I ran into some wall. I think you need special permissions for it and there's a gigantic backlog of such requests, essentially it never happens and so articles remain untranslated and unlinked.

For some languages like German, there's another translation site with human-vetted translations: https://dict.cc. It has support for language pairs beyond DE<>EN but those are pretty limited and iirc the suggestions queue always full. Not that DE<>EN is flawless but it's good and virtually always more helpful than DeepL or Gtrans

Somewhat related, but Wikipedia actually has a "special page" for searching external links, which I've had a lot of fun with:

https://en.wikipedia.org/wiki/Special:LinkSearch

You can e.g. search for all external links to HN, to pg's website, or to any Twitter user's tweets:

https://en.wikipedia.org/wiki/Special:LinkSearch?target=news...

https://en.wikipedia.org/wiki/Special:LinkSearch?target=paul...

https://en.wikipedia.org/wiki/Special:LinkSearch?target=twit...

You can also use it to get a sense of the adoption rates of new gTLDs:

https://en.wikipedia.org/wiki/Special:LinkSearch?target=*.ng...

https://en.wikipedia.org/wiki/Special:LinkSearch?target=*.de...

https://en.wikipedia.org/wiki/Special:LinkSearch?target=*.pr...

https://en.wikipedia.org/wiki/Special:LinkSearch?target=*.cl...

(I only wish there's a way to optionally filter out links originating from Talk pages, since those sometimes introduce a lot of noise, depending on your purpose)

All the other "special pages" are listed at: https://en.wikipedia.org/wiki/Help:Special_page

It's possible to use the external link search from the API, and that does have a namespace parameter. See for example https://en.wikipedia.org/w/api.php?action=query&format=json&...

I like the main idea behind it. Instead of arbitrary people registering and selling domain names in a first-wins manner, there is a collaborative/consensus process behind it, to decide which word should point to which website.

I think that in the long term, such a collaborative process to establish a <search-term> => <urls>[] mapping, would be more useful than search engines or domain names.

Nah, this would just mean that a mob can censor/cancel people they don't like, won't it? The majority doesn't care, but a small, loud minority does enough to break stuff like this.

Would be interesting to combine this with a web-of-trust.

None of my one-hop trusted people are part of the cancelmob. If I found that somebody two hops away from me did something absurd it's quite easy to do something about it -- apply negative trust to whatever path endorsed that nonsense.

Unfortunately people these days seem allergic to running anything that isn't a web browser or served by an app store, and the entities that control those two channels are extremely unexcited about decentralization.

I've been thinking about that for years.

I believe there are sooo many problems that would be solved by that kind of system

And although it might not be technically practical in many cases, I think in this case it would

Why not get the URL from Wikidata json instead of parsing Wikipedia pages? Some of these pages can be quite long, and you’re throwing away most of the data.

Wikidata can also list a choice of URLs if there are mirrors, or URLs that, on the Wikipedia page, are censored.

It is not a DNS in the sense of "resolver" but rather a DNS in the sense of "name validator" or "name search".

Cool idea, in any case!

Neat, but from the title I'd have expected somebody running an actual DNS resolver somewhere deep in the Wikipedia: or Talk: namespace (over which one could then of course run IP over DNS in networks that only zero-rate Wikipedia) :)

This is better because it's "serverless" DNS over https

The README says:

> Instead of googling for the site, I google for the site's Wikipedia article

This seems like a bit of unfortunate wording – there does not seem to be a Google search involved, just a search directly on Wikipedia (the English version).

Much like Kleenex or xerox, https://en.m.wikipedia.org/wiki/Generic_trademark

Well yes, but not here. While "googling" is generic, it's still a specific narrow term and it does not mean "search".

Yeah, I might use “googling” to mean generic web search in everyday language, but I wouldn’t use it to refer to a site’s internal search function, and I wouldn’t use it at all in a technical description of some software.

you might not, but clearly the author does.

Right. If that wasn't the case, this comment chain calling that word choice by the author as unfortunate wouldn't exist.

I'm happy for them.

I understood that statement as an example of author's workflow before the existance of this addon, as in instead of googling for the site directly, he would rather google for the Wikipedia's article of the site and find the correct URL there.

>If you Google "Piratebay", the first search result is a fake "thepirate-bay.org" (with a dash) but the Wikipedia article lists the right one.

Note that Wikipedia editors also remove links to sites they don't like.

DNS over Wikipedia - https://news.ycombinator.com/item?id=33830759 - Dec 2022 (159 comments)

Show HN: DNS over Wikipedia - https://news.ycombinator.com/item?id=22790425 - April 2020 (104 comments)

There's also:

Wikipedia over DNS (2008) - https://news.ycombinator.com/item?id=39628304 - March 2024 (13 comments)

Show HN: Wikipedia over DNS - https://news.ycombinator.com/item?id=22808121 - April 2020 (47 comments)

Wikipedia over DNS - https://news.ycombinator.com/item?id=2428198 - April 2011 (16 comments)

Wikipedia over DNS - https://news.ycombinator.com/item?id=1233499 - April 2010 (10 comments)

Wikipedia over DNS - https://news.ycombinator.com/item?id=382642 - Dec 2008 (1 comment)

... but I don't know how related those are

I wonder what it would take to get .idk root tld, and make it work without extensions.

$185,000 and a lot of paperwork.

https://newgtlds.icann.org/en/about/program

Previously on HN: Wikipedia over DNS - https://news.ycombinator.com/item?id=22808121

How sure are the devs that .idk won't become an actual gTLD?

They aren't and if happens will just use to another TLD. At least this is what OpenNIC did when .free was proposed to ICANN.

I wonder if one can take the English Wikipedia ZIM archive and process it to an efficiently-stored offline map of names to URLs, hopefully even with fuzzy matching.

Can this be installed for Iceraven? I really want it on mobile

wikipedia for blockchain gambling resolution

Begs the Q: Why is someone using Google when DDG and Startpage exist?

Can't someone !w piratebay and click the link?

Google has a near monopoly on web search, and they are very aggressively doing everything they can to keep it that way, both by tightly integrating it in their products (browser, browser engine, and OSs, mainly) and funding competitors (Safari and Firefox main revenue source are contracts with Google, under the condition of making Google the default search engine).

Sure, you're free to use DDG and find a workflow, but people are less likely to remember it, especially as they overwhelmingly use Google.

These days, the question rather is, why anyone on this site is not using Kagi? I understand perhaps the general populace simply doesn't know but here?

https://pluralistic.net/2024/04/04/teach-me-how-to-shruggie/

> I hadn't even heard of Kagi, but the Nielsen Haydens are among the most effective researchers I know – both in their professional editorial lives and in their many obsessive hobbies. If it was good enough for them…

> I tried it. It was magic.

...

> That was before I started playing with Kagi's lenses and other bells and whistles, which elevated the search experience from "magic" to sorcerous.

> The catch is that Kagi costs money – after 100 queries, they want you to cough up $10/month ($14 for a couple or $20 for a family with up to six accounts, and some kid-specific features)

> I immediately bought a family plan. I've been using it for a month. I've basically stopped using Google search altogether.

I will happily switch from Google when a search engine provides really good local results ^. It seems DDG and Startpage are mainly tailored for US. ^ Random suburb in random state in AU

Both work fine here in the Netherlands. Have you used the region toggle both provide to select your region?

Reply on news.ycombinator.com