Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

How a Raccoon Became an Aardvark

newyorker.com

Post reply on HN
Print view View original

71–80 of 96 posts

Re: How a Raccoon Became an Aardvark

#71

Post by stcredzero »

willvarfar wrote: post #55
stcredzero wrote: post #46

Earlier quoted context omitted.

But my team was cleverer: we were editing wikipedia. Speaking as a game programmer: The thing about computer moderated games, is that they foster the idea that, "If the system allows it, it's alright." There are entire national banking systems that have fallen under such attitudes. (Sometimes you just have to get real and handle dice and cards on a table top.)

We did revert the edits after. I agree with your "this is why you can't have nice things" sentiment, but sadly I think the petty and the vandals and the cranks will always do it, and us techies have to design in protection as default :(

"this is why you can't have nice things" sentiment

This is why we can't have nice things. It's also the solution to the Fermi paradox. The aliens have been watching.

Re: How a Raccoon Became an Aardvark

#72

Post by jonnathanson »

arjie wrote: post #66
dtech wrote: post #33

I'm not familiar with the Wikipedia cite policy, but wouldn't this problem be relatively easy to solve by requiring citations from before the fact was added? Granted, there might be some edge cases (like when a Wikipedian was very fast in adding the fact after it became known) but that would seem to stop the feedback loop dead in its tracks.

Wouldn't work. You'd need to find the first occurrence of the fact on Wikipedia. One example exploit (though there are others): 1. On Day Zero, write a falsehood into a Wikipedia article. 2. Wait for the falsehood to become popular through the process described in this HN story. 3. Delete the falsehood while performing other edits. 4. Wait a few days, making other edits. 5. Add the falsehood. Citations will all prece…

And it doesn't even need to be that elaborate. Wikipedia can be exploited recursively, i.e., by making up the same "fact" on two related pages, then linking to each other. If timed correctly, this can work for quite awhile before anyone unravels it.

Another tactic is to cite existing external sources as proof of a new "fact." If an article about, say, bananas has a few citations about the nutritional content of a banana, you can use one of those existing sources to cite the "fact" that bananas are descended from potatoes. Editors and bots usually won't question the authority of an existing source. It can be stretched and extended, like an umbrella, to vouch for new statements and information.

Re: How a Raccoon Became an Aardvark

#74

Post by chinpokomon »

huskyr wrote: post #11
acdha wrote: post #6

Earlier quoted context omitted.

This is too pat a dismissal – it's intellectually akin to the way “all bugs are shallow” aphorism which is pithy, optimistic and wrong. While in theory Wikipedia becomes more correct over time, it's not a given and there are many articles which never receive enough expert attention to catch non-flagrant errors. Wikipedia also has the unique problem that it can become less correct over time, whether due to error or de…

Note that you can permalink to every version of a Wikipedia article using the history tab, e.g. the first version of the Hacker News article [1]. There's also flagged revisions which allows users to 'flag' a revision as the 'accepted' version [2]. [1] https://en.wikipedia.org/w/index.php?title=Hacker_News&oldid... [2] https://en.wikipedia.org/wiki/Wikipedia:Pending_changes

This mechanism actually sounds like it could solve this problem. By making it policy that citations used in periodicals must link to these versioned pages, there is at least a traceable record about when something cited incorrect information.

I could also see this being a service to inform authors when their citations have been invalidated. A truthiness score of an article or blog post could be based on citation accuracy and could be assessed by a third party that specializes in fact checking, but initially it could be done by measuring when a particular like in a Wikipedia article has changed and whether or not that change affects the meaning.

Re: How a Raccoon Became an Aardvark

#75

Post by Pxtl »

Grae wrote: post #45
scott_s wrote: post #28

The article claims this is an example of a falsehood starting on Wikipedia, percolating to external sources, then perpetuating on Wikipedia because it's in those external sources. That's interesting. But what's more interesting is that this may be an entirely different case: a falsehood starts on Wikipedia, percolates to external sources, and then becomes true because it is pervasive . Then it remains on Wikipedia be…

Actually, the final paragraph of the article directly addresses this point: "Taxonomically speaking, this is unfortunate. The coati has no more relation to an aardvark than to any other vertebrate, so the name is misleading. But language, unlike taxonomy, is particularly susceptible to Wikiality. The nickname began because Breves wanted to retroactively prove that he had seen some kind of aardvark at Iguazu Falls. He…

Pedantic clarification: Argentina spells it "Iguazú" not "Iguaçu", since you're talking about the Argentine side of the falls. Spanish vs. Portuguese.

Re: How a Raccoon Became an Aardvark

#76

Post by jamesbrownuhh »

It certainly exposes a flaw in Wikipedia, but perhaps not the expected one. Namely, it's not so much that anyone can edit the encyclopaedia and introduce false information, but that Wikipedia's faith in "Reliable Sources" like newspapers, magazines, and other forms of big media, is so entirely misplaced.

Evidently you can't trust "proper journalism" to actually be true or correct, since many/most articles are written by untrained, unskilled, or uninterested staffers blindly re-writing whatever they can find on Google. And that is perhaps the most concerning aspect of all.

Re: How a Raccoon Became an Aardvark

#79

Post by TheMagicHorsey »

Its not only Wikipedia that is vulnerable to this sort of proliferation of misinformation. In the developer community there are many myths that become canon simply because they sound plausible and are repeated often enough.

I am a user of Go, and when I go to Go meetups I often hear people say, Go is faster than Java, because Go is compiled. This is definitely not the case in all cases ... and may not even be the case for most cases. But it is repeated as truth, because programmers think compiled binaries must be faster than bytecode run through a VM ... it just sounds so plausible. Never mind the real benchmark numbers.

(I still use Go, even if its slower sometimes, because Java can blow me.)

Re: How a Raccoon Became an Aardvark

#80

Post by vanderZwan »

tokai wrote: post #47
vanderZwan wrote: post #30

I wonder, isn't there some kind of way to measure "effective citations"? In biology, if you look at population size, there's the population measured in absolute number of organisms, and the population measured in genetic diversity[0][1]. For citations I can imagine measuring the number of citations from primary sources, and everything else just being citations to citations to... etc, including circular citations. [0]…

There is a scholarly field occupied with the concept of citation and reference. It is called bibliometrics. The tl:dr is that citations are a very complex social phenomenon, which is difficult to describe fully even in the semi structured confinements of scientific literature. Both technical and, even more pressing, theoretical developments are needed before measuring "effective" citations could become anywhere near…

Interesting, thank you!
Post reply on HN