Show HN: Explore Wikipedia edits made by institutions, companies and governments | Hacker News Reader

Show HN: Explore Wikipedia edits made by institutions, companies and governments

ailef.tech

246 pointsailef59 comments

Hi HN!

I have built a tool that allows users to search and explore edits to the English Wikipedia made from IP addresses associated with public organizations. I've written a brief article explaining how it works and you can see the tool in action here: https://wikiwho.ailef.tech

This was just a fun side project for me to build, and so I decided to share it with everybody else. I'll gladly receive any feedback and answer your questions.

Have a nice day!

Do you know about https://www.wikiwho.net/ ?

I was not aware of that, really interesting project. I actually may need to change the name then...

The text is almost unreadably small in firefox on a 1366x768 monitor. Screenshot: https://0x0.st/i1K2.png

Thanks for the feedback. I believe it was because I specified the base font size on the body element using the 'vw' unit. I've switched back to 'px' and it should be alright now.

I have this problem as well, on Chrome, when the window width is > about 1200px.

Uhm zoom out? Ctrl -

Looks great. This could be interesting feature for media monitoring companies.

This is really neat. Great work!

Yeah, at one place I worked at, an employee was banned because she kept trying to remove certain information about the organization from Wikipedia. I was a little amused, but she very nearly got our whole range banned.

I find coordinated efforts to edit Wikipedia -- by whoever and whatever "good" their reason to be -- fairly creepy. It's like a lot of little Ministry of Truths running about the place trying to massage reality.

It's like a lot of little Ministry of Truths running about the place trying to massage reality.

You just described the entirety of the advertising, public relations and marketing industries. It's essentially commercialized brainwashing. Humans, as social animals, have been at it at least since recounting great stories of hunting as pre-sapien hominids and probably earlier as pre-hominids.

The first sentence is also the thesis of Adam Curtis' documentary The Century of the Self.

underrated (timeless) documentary. https://www.youtube.com/watch?v=DnPmg0R1M04

https://www.craigmurray.org.uk/archives/2020/04/information-...

‘Long term readers will recall the Philip Cross Affair. A Wikipedia Editor named “Philip Cross” was relentlessly conducting a propaganda operation’ ..

‘The incredible thing about “Philip Cross” was that he never took a single day off. From 29 August 2013 to 14 May 2018 “Philip Cross” edited Wikipedia every day, including Christmas days, for 1,721 days.’

> Being the BBC it downplays the affair in a number of ways – crucially, it gives several examples to show that Philip Cross’s edits are harmless, and not a single example of his thousands of vicious edits, such as his editing my Wikipedia entry to call my wife a stripper.

I don't think of Craig Murray as the peak of unbiased writing.

Everybody has biases. Murray wears his on his sleeve.

More interesting: do you have concrete and verifiable information that his reporting of the Philip Cross Affair, that Murray' conjectures about its background, are wrong? Can you put forward an alternative explanation?

"he wears his on his sleeve" - his biases you mean. You've downright stated he's openly biased. So we agree.

But that does not mean he's automatically wrong, nor that he's worthy of dismissal overall. Just that I'd take more care to filter what he says than I would with, say, the beeb . Not that the beeb are beyond criticism BTW.

And I'm not disagreeing with him that this PC guy is odd and worthy of investigation.

It's just that overall fact that he indeed is openly very opinionated. That's fine, so am I, just caveat emptor.

But if this was a coordinated misinformation campaign, why not just use a VPN or dynamic ip?

The equivalent feed for the UK Parliament has been running since 2014 at https://www.twitter.com/parliamentedits

This is really cool!

But the approach seems a little flawed, as edits are attributed based on IP address. If I understand correctly that means it:

- doesn't capture edits made using Wikipedia accounts

- doesn't distinguish personal activity of individuals from those of the organization

- relies on a couple of third-party IP databases which aren't authoritative and may be out of date (one of them hasn't been updated in 6 years)

- can be circumvented by organizations using "unlisted" IP's or a VPN

Of course, the landing page is fairly transparent about these limitations. It doesn't mean the results are uninteresting, but when I looked at the most active month for the Government of Canada, it was clearly just an individual that happened to work there who's interested in TV and music (particularly The Robonic Stooges and Madonna).

https://wikiwho.ailef.tech/organization/46a20a0820d609f90314...

You are totally right.

I was thinking the IP address issue could be solved by adding a feature to crowdsource information about IP addresses from users, or maybe using APIs like ipinfo.io to create an updated database. (Didn't have time/money to implement it and wanted to avoid procrastinating the launch again and again...)

But unfortunately one just needs to create an account on Wikipedia (or use a VPN) to circumvent this kind of approach. This is really the main flaw, and I don't think there's a workaround to it.

I believe this flawed approach is the only possible approach.

As far as the individual interested in music the edits that are just individual edits on government time or IP address are a different problem than the other three. I guess could have some sort of analyzer to rank how likely the edits are to be of organizational interest and not just individual interest. But that would probably also require enough work to need to be monetizable for someone to put in that work.

Granted. I wonder if communities who maintain certain pages become aware over time of accounts known or strongly suspected to be associated with interested institutions. Per Wikipedia policy anyone paid to make contributions must disclose the fact (https://en.wikipedia.org/wiki/Wikipedia:FAQ/Organizations) but I'm not sure how vigilantly that's enforced. It would be interesting if Wikimedia released a transparency report about this, or even if organizations themselves did so (i.e. "here's a dump of all our social media activity last year").

This page might be interesting: https://en.wikipedia.org/wiki/Conflict-of-interest_editing_o...

It lists several cases that have been discovered of accounts editing Wikipedia for PR purposes or similar.

12 years ago there used to be a fantastic tool called WikiScanner that did exactly this. It made headlines for revealing edits by intelligence agencies and corporate character assassination campaigns around the world.

One that stood out for me was a series of bizarre edits from CIA IP ranges to Mahmoud Ahmedinejads article

http://news.bbc.co.uk/2/hi/technology/6947532.stm

The site went offline unceremoniously in 2008

Yes, I mention this in the article as well. I came up with this idea and then discovered it had already been done before, but since it has been offline for such a very long time I thought it could be cool to do it again 12 years later!

At a previous enterprise job, I actually emailed ethics@ when I caught a coworker editing a Wikipedia page in violation of the Wikipedia rules and company policy. It became clear during my subsequent interactions with the general counsel that they didn't exactly get many emails to that alias, nor did they really know how to go about handling a relatively minor violation.

Institutions, companies and governments can't bother to edit from a VPN, or from a tethered cell phone?

You're assuming that edits made from within an organization are sanctioned and/or part of some propaganda or cover-up plot. I'm sure that many of the edits are just random people making the edits while at work. I have certainly done that when I saw some errors or vandalism while browsing at work.

> Institutions, companies and governments can't bother to edit from a VPN, or from a tethered cell phone?

IP addresses are masked if using a registered account. It would be interesting to cross-reference those except Wikipedia don't gave this out.

I do know I opted-out of Wikipedia after being censured for posting the birth date of a deceased individual. In violating of privacy policy or some such WP:bla bla bla. The terms of which are impossible for the average individual to decode. A fortnite later the information was inserted by some senior administrator. I guess all Wikipedians are equal, except some are more equal than others.

I stopped editing when it became clear that all of my edits were immediately reverted. They were all simple grammar, spelling, or tense modifications. I assume the original authors had scripts in place to make sure there were no changes.

Someone should make a wikibot to make such small innocuous edits, in order to detect such rogue editors who consider themselves “owners” of certain pages (against policy, I might add). It would perhaps also be possible to find at least some of these by detecting these changes in the editing history in cases such as yours.

The trick is to post the proposed edit on the talk page first, then make it after a few days if nobody objects. If your edit is reverted, you can point out that the user didn't raise any concerns on talk. Reverts are very much supposed to be a last resort measure, the primary way of shaping a wiki article is via discussion/deliberation on the talk page. Also very handy for edits that you know to be controversial.

Wow, that's a thing. Would you care to link your user page.

Please let me know as well, either here or via DM. I can definitely look into this.

This was about 8 years ago, and all my edits were anonymous. They actually relate to this article because it was from an IP registered to a previous employer. Only a handful of people used that particular address, so when I looked at the history they were all mine. So I dont have anything to send, but If I had to guess I'd say it was less than 10 edits spread over 2 years, and all of the edits were less than 30 characters. I later learned about the talk page and the internal politics, and I decided it wasnt for me.

" I guess all Wikipedians are equal, except some are more equal than others."

Where did they say all are equal?

Unfortunately a lot of VPNs are blocked. I have tried to make several innocuous edits (grammar/spelling, adding references) since moving to China but it's almost impossible since we can't get to Wiki without a VPN and VPN users can't edit.

Does wikipedia explicitly block editing on VPNs entirely? Or is it that you're not using an account and the IP is blocked from editing?

Using an account is not enough to bypass a blanket IP address range block on Wikipedia. You have to meet certain conditions and submit a formal request for an exemption: https://en.wikipedia.org/wiki/Wikipedia:IP_block_exemption#R... (And I should point out that using a commercial VPN — which is a form of anonymizing proxy — seems to disqualify you for an exemption; I’m not really sure if an exemption could be granted if you have to use a VPN, but it doesn’t seem to be the case, at least on paper.)

I’ve noticed the same when trying to edit from a commercial VPN service.

This is very welcome, because it's a side of WP content that isn't very transparent without a lot of time-consuming scrounging. Hopefully others can build on this foundation.

Is it just me or the font on WikiWho really small? Also it doesn't seem to scale correctly with page zoom for some reason.

Thanks for this feedback, I'll try to fix this as soon as possible.

Not just you! Some text doesn't scale at all.

Hilarious: https://wikiwho.ailef.tech/diff/115cb29701ac58cd3cdf515ed043...

Airforce on the Pope: https://wikiwho.ailef.tech/diff/88dfea55a8f8f112b404c62da094...

Someone in the Navy taking offense at the content of the 'creation science' article: https://wikiwho.ailef.tech/diff/06d1afb0c1ff80f387c99d007059...

Great work, even with the caveats others have mentioned! But to maybe add yet another caveat: do you or anyone here know of any way that a malicious actor could expose a complete fake IP address, say in order to discredit an organization, other than working though an infected machine within the organization? Pardon my ignorance, and my intention is not to have such techniques be made common knowledge (if they exist), but if this is possible then it is an important aspect too.

Wow, this is really interesting. Nice work!

Just a random example I found clicking around, an edit from the U.S. Department of State, removing a paragraph of criticism from the article "2003 invasion of Iraq". Looking at the history from Wikipedia, it was added back a few minutes later.

https://wikiwho.ailef.tech/diff/3ccc3672d891b62a895d4821c4ab...

I find the title and article a bit misleading. Like "companies" dont edit wikipedia, usually it's a random dude forgetting to log with his phone to correct spelling, a gross mistake when he was reading his CEOs life or vandalism.

I'm not sure why this project stopped at the obvious: it should go further and find "negative" edits that are beyond spelling, fact checking or general updates.

Well, I happen to disagree. It is a well documented fact that companies often try to manipulate Wikipedia (either directly or by paying somebody else to do it). Granted, majority of the edits are non-malicious in nature but finding controversial edits automatically is really hard, and the goal of the tool would actually be to aid people in finding them.

It would be interesting to investigate WP pages that are "ideologically locked" for whatever reason. Maybe even come up with a metric, etc.

Some of this would be rank politics, of course, and some PR edits. But others might be rather surprising. As I recall, for several years, the main C++ page forbade any substantial criticism of the language.

Wow. This is really, really cool. Having this level transparency on who is spreading narratives or propaganda in what way is very interesting. In my few brief searches on some history-related pages you can see hoardes of edits from US military bases for instance.

Wow, I thought the post was dead and I went to sleep with it having only ~5 votes and I find out today it exploded!

Didn't think it was possible several hours after it was posted. I'm reading the comments and trying to reply now.

You do realize that if an employer makes an edit during work, the IP will be of the company. Your title makes it seem like it's the official stance of the whole company.

Why is the US Navy so active on such innocuous articles?

Bored sailors?

Posting from inside does not mean, they do so on assignment.

This is awesome!

Nice job!

Try editing the title to Show HN: (more appropriate).

I've added that now.

Reply on news.ycombinator.com