This is a draft of a potential Signpost article, and should not be interpreted as a finished piece. Its content is subject to review by the editorial team and ultimately by JPxG, the editor in chief. Please do not link to this draft as it is unfinished and the URL will change upon publication. If you would like to contribute and are familiar with the requirements of a Signpost article, feel free to be bold in making improvements!
| |||||
The Anti-Defamation League has beef with Wikipedia. Okay, that's not exactly news, but this particular beef is. It has to do with banned editors, the Israel-Palestine conflict, and whose responsibility it is to clean up the mess of others on Wikipedia. Spoiler alert: it's not mine, and it's probably not yours either, and it certainly isn't the Wikimedia Foundation's. Hornet's nest, meet whackin' stick.
Let's step back to 2024. Editors are engaging in all kinds of conduct related to the wider set of Palestine-Israel Articles (PIA). Bans are getting handed out left and right for bad conduct, edit warring, and "disruptive behavior". In the midst of it, the Anti-Defamation League was deemed unreliable as a source on the topic of the Israel-Palestine conflict. Media ran wild with the story and got things wrong to such a degree the Wikimedia Foundation felt it necessary to dispel some misinformation cropping up around the demotion. Fun times to go to a wedding and tell people you edit Wikipedia, I tell ya.
In October 2024, Ashley Rindsberg published what I think is best described as a "bombshell" report on Pirate Wires about a shadowy cabal of 40 editors working to reframe the Wikipedia narrative to be anti-Israel. Fast forward a few months, and ArbCom sanctions two handfuls of editors, including a half-dozen or so who were named in Rindsberg's report. Just one teensy-tiny problem: those editors had collectively made thousands of edits to PIA articles.[a] Who's going to clean those up?
Before we can answer that, we need to ask, "Do they need to be cleaned up?" If you ask the ADL, the answer is, overwhelmingly, "Yes." An October 15 report from the organization found that after nearly two years, edits persisted from banned editors to a high degree, specifically calling out the case of Iskandar323's work on six highly trafficked articles, and even more specifically, calling out Iskandar's removal of a big chunk of content (around 4,000 words) from the Criticism section of the Hamas article detailing many instances of the organization's human rights abuses, hate speech, and other content that one might expect to find in a Criticism section. The ADL also called out Iskandar's removal of descriptions of extreme violence during the October 7 attacks, totaling around 17,000 "characters of text and citations" about atrocities suffered by Israelis in the fighting. The ADL says 91 percent of Iskandar's edits to those six articles persisted, or 362 of the "checkable" edits the ADL identified.
Now, if we stop there and dig no deeper, those claims certainly sound pretty bad. One editor, acting unilaterally, with just a few edits, has dealt a pretty resounding blow to soften the image of Hamas on Wikipedia, burying atrocities associated with the organization deeper into the encyclopedia where they can be hidden from curious eyes. But is that really the case? Well, that might be more difficult to answer.
This is a good moment to take a pause and explain a few things about how the rest of this analysis will work.
First, I am not here to defend Iskandar323's honor or relitigate the ArbCom decision. Frankly, I could not care less about that. I want to write articles about semi-obscure figures from Wisconsin history and journalists. I am not remotely interested in engaging in the political debates surrounding PIA, I am not interested in weighing in on article content on their Talk pages, nor participating in any Wikipedia drama around this topic outside of this little submission. Please do not ask. The central point of this analysis is to examine the specific claims made by the ADL in their October 15 report and analyze the specifics of the edits made. I have data!
Second, I examined these edits through the lenses of "Was this something a reasonable, uninvolved editor might have done?", "Do the content policies and guidelines support the edit made?", and "Should this have been something that was part of a broader discussion before it was done?" Essentially, I tried to grade this as if I were reviewing an article under the Good article criteria.
Third, nothing here should be construed as me saying we shouldn't look into the edits of banned editors. In cases like these, we probably should. But drinking from the fire hose of misplaced commas to find the one nugget that needs a second look sounds like intellectual self-flagellation. There's another way.
Finally, I am not a masochist. This is a limited, non-academic analysis. I started this research after PHP2Shell torched all the fun tools we used to use. I did not examine all 485 Iskandar edits the ADL identified. The edits I did examine led me to believe that maybe the ADL is not acting in good faith, or lacks the technical expertise to click "next edit". You're also not going to find coefficients and regressions and p-values and all of that. This is a limited, subjective analysis with a few basic statistics. I would encourage others to build on this if they want a more scientific approach.
According to the ADL, the "most consequential" of Iskandar323's edits in this grouping was to the article on Hamas, specifically removing around 4,000 words from the article's Criticism section and putting them in Criticism of Hamas. And yeah, if all you do is look at a 50,000 byte (roughly 4,000 readable words) edit that cuts out what can only be described as "unfavorable content" about Hamas, it looks bad. But the reasoning is the key consideration. Iskandar cites WP:TOOBIG and WP:SPLIT as their policy backing for making the edit. Oh, and just to be clear, Iskandar did add the content where they said they would with no readily apparent changes, they didn't just say they were moving it and then delete it.
So, the cited rules. TOOBIG? Check. Prior to Iskandar's cut, the article was over 18,000 words. According to the guideline, more than 15,000 words and the article should "almost certainly should be divided or trimmed." Cutting the content and putting it into the hatnoted article for that section certainly fits.
SPLIT? This is where it gets a little more iffy. Does it present a neutrality issue to remove all this content? Yeah, I could definitely see arguments for that. Reading the content, I can also definitely see arguments that its presence created neutrality issues. The content itself had many things about it that would leave me scratching my head as a GA reviewer. I probably would have stopped and told the submitter to check out MOS:TONE, WP:PERSUASIVE, and MOS:QUOTE and to make a bunch of changes before continuing the review.
So what's the verdict? Well, I'd probably say some discussion would have been in order. A WP:BOLD split probably wasn't the right move here, there probably should have been a Talk page discussion. And there may have been, I am not going back through 35 archives to check. I would hazard a guess, though, that the Talk page discussion ultimately would have resulted in some kind of split for that content, maybe with a more robust summary.
As for the ADL's analysis, I have to wonder if they are actually making these claims in good faith.
The organization created a chart in their report that, rather than use the actual headings from the article, creates groupings that obscure the reality of what's being presented in the article. ADL says this is how the article is organized:
While the actual top-level headings were:
Placing "Military operations" last is just clearly and blatantly incorrect, but more to the point, it makes it sound as though Wikipedia was saying something much different at the time than it actually was. "Violence and terrorism" has a very different connotation than "Military operations", and I am quite sure the ADL knows that. It is much easier to cry foul when you misrepresent reality. If the organization is trying to galvanize Wikipedia editors to look into the edits of the banned, this is a poor way to do it.
ADL's second identified edit was the removal of atrocities committed during the October 7 attacks from the Gaza war article. The ADL says some 17,000 "characters of text and citations" were removed in one fell swoop, which would again be a substantial change. If it had stuck.
But all the ADL had to do was click "next edit" and they would have seen that Iskandar reverted their own removal, intending to write a better summary, which they did, even though the ADL characterized it as a cut of "7,400 characters of what remained".
What rules did Iskandar cite for this change? None. The first edit, Iskandar noted the section was wholly about the specific attack in 2023, rather than the Gaza war as a whole. This requires a little more extrapolating to get at the relevant statutes, and a little more reaching into the land of essay than policy or guideline, but arguments around WP:TOPIC, WP:RELEVANCE, and WP:SUMMARYSTYLE all could be applicable to Iskandar's edit summary.
Digging deeper, a WP:DUEWEIGHT argument could also be made, but perhaps the simplest would be WP:TOOBIG. The Gaza war article was almost 20,000 words when Iskandar made the initial edit, and around the same when they made the summarization, which, when framed as words instead of characters, trades emotional impact for intellectual honesty: from 754 words down to 327 (as measured in a Google Doc, so your mileage may vary a little). A big cut, but without nearly as much drama, and in comparison to the overall size of the article, next to nothing, and absolutely inline with the spirit of TOOBIG and SUMMARYSTYLE. Was it perfectly executed? That's for the Talk page to decide.
The content edited was also fairly graphic, to the point it may have presented issues with neutrality and WP:PERSUASIVE, and raises questions related to WP:GRATUITOUS. Wikipedia may not be censored, but is it necessary to include a sentence which says A large charred mass that when observed by CT scan show the remains of a parent and child who were bound together before being burned alive, or can the information about immolation be conveyed without it?
That, and the summarization of the content as a whole, would have been a great discussion to have on the Talk page, but it's not clear if that happened, and again, I am not searching through dozens of archives to try to find it.
Looking into these edits raises yet more questions about if the ADL was writing their report in good faith, or if they possess the technical skill needed to make the kinds of claims they did. The skill bar is incredibly low to clear. If they need help with diffs, I am sure they could find it if they wanted.
Okay, I am a little masochistic. I did extract all the edits Iskandar made to Gaza war and analyze them individually. Out of 87 edits, here's how I assessed things, again going roughly by what I would try to do if I were reviewing a GA submission:
Of the 68 reasonable changes:
If you're curious for the full breakdown, it's in the collapse below.
Extended content
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
So, of those 87 edits, 78 percent were within guidelines and 22 percent could use another look. But the ADL wasn't just bringing attention to the edits as a whole, they were bringing attention to the fact the edits persisted. In total, they found 91 percent of Iskandar's edits to those six popular articles remained in place. I believe I have already demonstrated their methodology may be a bit flawed, but how much does that persistence figure change if we look only at the potentially problematic edits? That is a much more interesting question, and something that would be much more manageable to look into.
How should we fix this problem? If you were concerned the ADL didn't have any ideas and only wanted to gripe, I have good news and bad news: they definitely have ideas. Those ideas are just kind of antithetical to how things work around here. But there might be ways we can find a middle ground.
There are a handful of recommendations the ADL has for Wikipedia, but I think the two most salient ones to this article are a mandate to verify the identities of users who work in "designated contentious topic areas" and "committing to structured content audit(s)" of articles from t-banned users.
The verification information could be kept just long enough to verify an account belongs to a single real person before being deleted and would be a deterrent for state actors.
Basically, it would require the Foundation to collect exactly the kind of information Asian News International demanded of the Foundation during their lawsuit. Protecting contributors from persecution and prosecution is a critically important function of the Foundation, and one of their best defenses is "We don't collect that information." If the WMF starts verifying identities, that goes out the window, and everyone will be at risk if they ever step foot into these "designated contentious topic areas", even if it's only to add a forgotten period. Who gets to decide what's designated? We had a decade-long fight about how to spell "yog(h)urt". Does that count? Why not? We can be certain that state actors, rather than be deterred, would simply weaponize this verification system to hunt down contributors who do things they don't like. The Wikipedia we know and love would be gone.
The "structured audit" would require the WMF to go through all the banned editors' edits, restoring or rewording content according to where the ADL, INSS, or "independent researchers" have documented bias. In other words, the ADL is demanding the WMF remake the articles to fit its preferred narrative. That's also not going to happen for a variety of reasons. Despite their concerns, fundamentally, this is a content dispute. The WMF can't be making those kinds of editorial decisions or it will lose its protections under Section 230 and similar laws. That's to say nothing of the active revolt of the community.
Does that mean we should turn a blind eye to the edits of these banned editors? Of course not. As the ADL points out, these are some of the most highly trafficked and highly edited articles on Wikipedia. A lot of people have a lot of opinions about what they say. There have been more than 3,000 edits to the Gaza war article since Iskandar last touched it. But auditing every edit would take a huge amount of time and effort. Just going through those 87 took me several hours. Some of that was due to fighting with Wikipedia's table coding, but still. It's a massive effort when that focus gets expanded to thousands of edits by a group of editors.
It's unfortunately not as simple as filtering out edits with "ce" in the summary, either. Even minor wording changes in topics like those in PIA can substantially change the meaning of the text and alter the framing. And I certainly don't expect everyone who looks at my analysis will agree with it 100 percent, but assuming that all of the edits these editors made are problematic clearly isn't viable either. So what can be done?
The ADL learning more about how Wikipedia works and its purpose would also go a long way. Taking the time to learn the content rules, look at how they are applied, reading up on tone and summary style and other rules I've cited here, that would all help a lot. It's frankly absurd to expect 4,000 words to be added back to a 15,000 word article. And please, please, click around just a little in the edit history. It will be better for the ADL's reputation long-term.
We're not here to be a medium to generate sympathy or outrage. We're not here to right great wrongs. We're here to make an encyclopedia. A summary of knowledge, not an exhaustive accounting. The ADL and other organizations must keep that in mind before they start beating down the gates.
The ADL clearly has some kind of system in place for extracting edits and analyzing them to see if they remain. If they take it one step further, and specify which edits remain that may genuinely introduce bias, then bring it to the community, I think we would be far more likely to listen. Making suggestions on those article Talk pages, or even WP:COI edit requests, would probably be the easiest route to go, and I think demonstrating the ADL is trying to make a genuine contribution to Wikipedia instead of trying to control it would go a long way.
The ADL will also have to accept the ruling of the community. One huge reason why editors likely haven't undertaken the analysis the ADL wants them to is because we know it will put a target on our backs (hi, Mom). We will almost certainly be named and shamed if we don't toe the ADL's line, and that could lead to all kinds of harassment and disruption in our lives when most of the people the organization would want reviewing these edits absolutely do not want that. If Jewish editors feel traumatized editing on these topics, how will we feel if the ADL doxxes us because it's mad about our decisions?
The ADL came to the table with a lot of sticks. I've got one, too. It's an olive branch. Their move.
Cite error: There are <ref group=lower-alpha> tags or {{efn}} templates on this page, but the references will not show without a {{reflist|group=lower-alpha}} template or {{notelist}} template (see the help page).
Discuss this story