DNS over Wikipedia
github.com
github.com
Uh huh
https://en.m.wikipedia.org/wiki/Wikipedia:Village_pump_(poli...
Perhaps this could be realized by letting logged in users fill out a field for sources of third-party domain records, each of which could be a source of domains across the entire Wikipedia site. Or, more simply, just flag certain domains as such, and silently require users enable something like HN's "showdead" bit.
Otherwise, if you look at the size of that talk page, trying to decide this once and for all brings to mind the idea that "hard cases make bad law".
I thought it might be something like automated edits to Wikipedia talk pages as a database or cache, which would be abusive.
But actually it's using Wikipedia entries as a smart name search. This is read-only and a clever and legit use of the WP. If the browsers allowed the use of spaces in the domain name portion URLs it would be even more useful, but it's probably better that they don't (out of spec for TLDs)
Doesn't this create a situation where a bad actor could change the Wikipedia page for a `semi-popular-brand.com` url listing to something bad? Anyone who used `semi-popular-brand.idk` in that timeframe would land in the bad page. Perhaps I'm misunderstanding.
I think you might be onto something...
What it does not do is prevent anyone from temporarily editing the associated wikipedia page and phishing you into an attacker-controlled website.
The same concern was raised by the user "segfaultbuserr" in that Show HN thread[1].
In 2019 I wrote up this very idea as part of a brain dump of long-stalled thoughts that I had been kicking around but not published anywhere. There are mitigations to the attack you describe, which I included in my original terse writeup about such a "Wikipedia Name System"[2]. I pasted the explanation in a comment in the original Show HN thread. The idea lies in the fact that although wikis can be edited to point to your own fake honeytrap, there are things that can't be faked; it's a public ledger sort of thought exercise in the vein of Bitcoin—but no need for proof of work or what is currently associated with cryptocurrency, etc. As I (re-)explained three years ago:
> Not as trivially compromised as it sounds like it would be; could be faked with (inevitably short-lived) edits, but temporality can't be faked. If a system were rolled out tomorrow, nothing that happens after rollout [...] would alter the fact that for the last N years, Wikipedia has understood that the website for Facebook is facebook.com. Newly created, low-traffic articles and short-lived edits would fail the trust threshold. After rollout, there would be increased attention to make sure that longstanding edits getting in that misrepresent the link between domain and identity [can never reach maturity]. Would-be attackers would be discouraged to the point of not even trying.
(This also provides mitigation for what is (currently) the top comment in this thread—the issue of Wikipedians censoring certain domains, provided that they have a long enough history to meet the trust threshold before they were decided to be too contentious to share information about them.)
1. See <https://news.ycombinator.com/item?id=22791534>
2. Originally published, along with other thoughts from that month, at <https://www.colbyrussell.com/2019/05/15/may-integration.html...>
Also, am I just needlessly splitting hairs?
In DNS, domain names are the keys. In this system, domain names are the values.
Sure it’s a very narrow form of DNS...
Since the linked README doesn't even mention "IP", I suspect the author misunderstands what "DNS" is about, and this is actually a search tool, and the headline is bad.
Still, this is arguably a domain name system, just one that isn't compatible with the DNS as we know it.
I still use this technique "manually" for finding translations when dictionaries, though. Wikipedia has a lot of information which isn't in a typical API-friendly format, but with a bit of regex there's some interesting possibilities.
If this sounds interesting, the Objective-C source is still available here: https://github.com/garrettalbright/wptrans You might also be able to find clones of it in the App Store since I know people were taking my source and republishing it back in the day, some even for a profit (mine was always free).
> Wikipedia's asshole lawyers didn't like that I used a serifed W in the logo
It’s not the lawyers which don’t like it. Trademark exists to protect consumer confusion. If a reasonable person could reasonably (but incorrectly) believe that your app was officially associated with Wikipedia based on only looking at the logo, then you have created consumer confusion. Trademark law disallows you from (using trademarked branding) letting people believe that your app is official.
> I made it clear it wasn't an official Wikipedia product everywhere
Posting “No copyright infringement intended” on a YouTube video does not, contrary to masses of people, turn a copyright infringement into something else.
Likewise, just writing that your app is not official will not make a trademark infringement vanish.
(Of course, if Wikipedia does not have a valid trademark which you infringed, that’s another story. But it’s not what I would guess has happened here.)
Note that it does not use the same colors as the Wikipedia logo or the same puzzle piece/globe iconography. It uses a serifed “W” as well as characters from other writing systems, with a brown and yellow color scheme - alluding to Wikipedia as well as the app’s purpose, but intentionally avoiding copying to a degree that would cause confusion. If, after considering this, you still believe I was intentionally inviting confusion between my app and an official Wikipedia product even before all of the documentation stating that is not the case, I believe that you are either operating on very bad faith, or are one of Wikipedia’s lawyers trying to justify your invoice. “Our firm removed 20 violating apps from the App Store this month!”
ITT: people who didn't actually read what the project does
I'm laughing in pain. I detest google so much for being so bad at its only job - finding stuff - merely to maximize $$$. Google was such a great product and genuinely serving humanity. Now it's only a shell of its former self.
Long story short, about ten years ago, my recommendation of a specific search tool (shareaza) to a friend went south, since it was over the phone. I'd found telling people to use web searches didn't work out well, wikipedia on the other hand, usually had an up to date address -- I no longer recall what trouble it caused, but I spent some time wondering what malware was redirecting to the bogus software site -- there was none, it was the edited entry at wikipedia. (When I tried the bogus software I recall it looked like it was doing something, but in fact it was doing nothing as far as searching p2p networks. Being busy I wasn't interested to see what its objectives were before deleting.)
Frustratingly, beyond not actually offering that (which is entirely reasonable), it does not even seem to be using DNS for the implementation of what it does.
As others have pointed out, Wikipedia is starting to censor results just like Google. Maybe it would be better for this extension to pivot into providing their own service that performs this task.
Wouldn't this hammer wikipedia unnecessarily? DNS are very quick, lightweight requests.
On a separate note, someone could actually make a website do this without any extension. It would need to use wikipedia dumps and has the advantage it can't be suddenly edited by a malicious actor.
Why would we trust "a website" more than random wikipedia edits to provide correct or up-to-date data?
The purpose of this app is to get the current url of those sites, which keeps changing url semi-frequently, so a offline copy of wikipedia would not work.
What would an immune-system for online information look like?
(The CFAA has since expired, and those rude little pedophiles DEFINITELY noticed me.)
Note that "wrong" information is not bad as such, it is just wrong. It can be corrected. It becomes bad when its purpose is to mislead people, in which case its originators will never want to correct it.
In the video, it doesn’t show this. It shows going to the scihub.idk domain. And then a redirect happens. So does this tool just host a local a domain resolver (and HTTP redirect server) for all .idk domains that does a wiki search and then responds with a HTTP redirect?
1. Make a Wikipedia search API request for the .idk domain, using the name as the article name.
2. Retrieve the rendered page contents if found.
3. Find the first Wikipedia infobox table on the page.
4. Extract the first "URL" or "Website" entry in that infobox.
5. Return the entry's value, if it's a link.
All this runs in a nickel.rs server on 127.0.0.1:80, which routes the requests as permanent redirects to the destination. Using dnsmasq,[1] if it's an .idk domain, it routes the request through the above Wikipedia resolver.
1: https://github.com/aaronjanse/dns-over-wikipedia/tree/master...
Specifically, Wikidata has a "official website" property [2] that seems to be used. If there are multiple extensions, like in Sci-Hub's case [3], it could pick one based on user preferences.
[1] https://www.wikidata.org/ [2] https://www.wikidata.org/wiki/Property:P856 [3] https://www.wikidata.org/wiki/Q21980377
Maybe that’s a good thing. But let’s not pretend it’s less censored than google!
This a redirect to the right domain. It is not an IP address lookup. This could have just as well been an “I feel lucky” search box using Wikipedia as the source. Or a Duck !bang.
Is this a function of reality or me getting better at spotting it?
I felt as though in the 90s CNN legitimately was a good news source. Early 21centuey VICE was legitimately good.
Today there is next to nothing given anything other than a heavily biased view.
I think the internet has made things worse, there's a new kind of bias. What do you set the headline to for the breaking alert? Because that's what people will see. Follow that up with a more subtle article and boom you've got a basically iron glad propaganda machine.
If all the news sources are rather centrist (i.e., they all "cooperate"), then no one gains or loses audiences because of bias. This is a good strategy when publishing is expensive. However, as the costs of publishing drop, it becomes profitable to shave off niche (smaller) audiences, with biased publication. This begins the defection phase of the PD game series. As defections accelerate, anyone who doesn't defect eventually suffers by continuing to not defect, leading to a new optimal state of all defections--i.e., publishers have to defect to maintain audience.
I dunno, I just came up with that as I was writing it.
There are also things to be said about the ability now to track engagement, which allowed humans to quantify (i.e., put a cost/value/ROI on) the extent to which bias/outrage drive engagement. But, this function would still play into the greater theoretical framework I proposed.
https://web.stanford.edu/~gentzkow/research/biasmeas.pdf
We estimate a model of newspaper demand that incorporates slant explicitly, estimate the slant that would be chosen if newspapers independently maximized their own profits, and compare these profit-maximizing points with firms’ actual choices. We find that readers have an economically significant preference for like-minded news. Firms respond strongly to consumer preferences, which account for roughly 20 percent of the variation in measured slant in our sample./s
Even the BBC which is state funded. Okay, that might be a bad example as they are a state broadcaster and so propaganda is really their remit.
We could just call them specialized audiences with particular interests, instead of equating the neoliberal centrism of a bunch of collaborating oligarchic publications to lack of bias.
edit: e.g. if all of the publications collaborate to not discuss issues concerning Mexican-Americans, a defector who peels off a large audience by being the only publication that attends to Mexican-American issues would not be an example of "bias" except under extremely normative definitions of "bias."
Nope. They were the original cheerleaders of modern militarism, from the first Persian Gulf War, which they turned into an infotainment spectacle devoid of any serious critical analysis, to their promotion of NAFTA and neo-liberalism. This idea that things were better in the past is a form of rosy retrospection and nostalgia. Things weren’t better, they just hid the bullshit under thicker layers of opaqueness. And I say this as a progressive liberal who has always hated CNN and despises Fox.
The sad thing is that we have real problems, and this nihilistic insistence that everything is fake means that those problems will only get worse.
Whenever I get into a discussion about vaccines I say something along the line of:
Of all the shitty things our governments do in our names, you’re choosing vaccines to fight against?!
Could it instead be that the environment (or, to speak in your words: “the western mainstream hidden forces”) has altered your perception?
You can always exit the Matrix.
I do not sense any specific bias. As usual, "freedom of speech" does not imply a right to be listened to.
Also, I do not buy this dystopia claim. In 99.99% of all topics we know exactly what is true and what is not. The sky is blue, clouds are condensed water, the earth is a sphere, there is racism, there is poverty, wealth is unevenly distributed over the planet, hygiene helps prevent sickness, vaccinations have prevented a lot of suffering, social security is expensive, etc, etc.
The facts are easy. We used to argue about consequences stemming from these facts and what do to about it. These days we tend to confuse our disagreements with "alternate facts".
It's really not that hard to stay informed.
https://en.wikipedia.org/wiki/Opinion_polls_about_9/11_consp...
Now, you can ofcourse dismiss all these opinions as "alternate facts". Probably similarly to how devoted christians dismissed theories about a heliocentric solar system as "alternative facts" only aspoused by lunatics.
We'll follow the scientific method; and that is how we learn and continue to explore and learn more.
Do really not a see a difference between that and convenient political alternate facts?
And the grey sky? Seriously? You're looking at clouds, not the sky. Whatever, why am I even responding...
I understand, the truth is not comfortable. It is sometimes incredibly hurtful to know the other sides version of the truth. But even then truth is worth it.
"All I'm offering is the truth, nothing more."
1. Wikipedia is a private entity.
2. It is very well funded.
3. That extension probably uses less resources than someone loading an entire webpage with all associated media just to look up a url.
I presume this is a fun test project though, so I doubt it would cause much harm, but it could be there.