Jump to content

Wikipedia:Diffpicking

From Wikipedia, the free encyclopedia

Diffpicking is the careful selection of individual Wikipedia edits in order to construct a desired narrative about Wikipedia or any subject on Wikipedia. Due to the transparent nature of Wikipedia, it is possible to link directly to any change to an article and any comment in a discussion. By curating such links without context, an edit or a comment which was quickly undone, ignored, or inconsequential is indistinguishable from consensus-based decisions and persistent article content.

Transparency

[edit]

One of the defining features of Wikipedia, and one of the primary ways it differs from other sources of information, is its radical transparency. It is transparent both in the editorial sense that all of its policies and sources are publicly visible, but also in the technical sense in that every change anyone has ever made to any page is archived and accessible through that page's "history". In the context of articles, this means if you change a comma to a semicolon, that trivial change is accessible via a dedicated URL indefinitely. This is true for articles as well as discussion pages.

The page that displays one specific change to an article or other page is called a diff (the "difference" between two versions of a page).

Narrative construction

[edit]

Page histories are used for a wide range of purposes. Anyone can look back at different versions of the article from various points in time to understand how a subject (or Wikipedia's coverage of that subject) has evolved. They are also indispensable tools for enforcing our behavioral rules. When a user frequently makes problematic edits (perhaps they are prone to edit warring, copyright violations, or personal attacks), diffs constitute the evidence needed for administrators to take an action, such as a warning, subject-level ban, or loss of editing privileges.

If not used carefully, and with context, diffs make it possible to construct nearly any narrative one wishes about Wikipedia. This is diffpicking (cherry-picking diffs).

In journalism

[edit]

Journalism about Wikipedia is tricky. Our editing policies, decision-making processes, and organizational structure can be complex for any observer. Navigating and understanding article histories and talk pages are especially challenging, and often time-consuming. This is as true for journalists as it is for their readers, who want concise stories without the need for a primer on wikipolicy. For this reason, diffpicking is an easy good-faith mistake and effective tool for bad-faith persuasion.

Here is a silly example:

Because anyone can edit nearly any article on Wikipedia, and anyone can edit nearly any talk page on Wikipedia, people say and do dumb things all the time. While we have policies, procedures, and tools to trivialize their impact, we do not altogether prevent them in order to preserve the open editing model. The downside is that, unless you examine the context, edits which were immediately undone and talk page comments that were inconsequential are indistinguishable from those with lasting impact. Just as anyone can make a bad edit, likewise anyone can propose changes using dedicated consensus-building processes, like proposing to delete the page about cheese, changing the page about cheese to focus on vegan cheese, or floating a conspiracy about powerful cheesemongers. Those, too, can appear consequential to an observer even if they never stood any chance of being enacted.

What matters is what persists in the article, and which perspectives find consensus on the talk page. Consensus is Wikipedia's decision-making model; while you can read about the many ways consensus is determined on this page, when it comes to controversial subjects it is typically an evaluation of a discussion as a whole, and how well various arguments align with Wikipedia's policies and guidelines. Properly understanding where lasting content was added or developed, the conversations which led to it, and which arguments were involved is time-consuming even for experienced Wikipedians, but typically involves examining large numbers of edits across a period of time and several full discussions on the article talk page (not to mention where the article might have been discussed on other talk pages, like noticeboards. If you want help figuring out context, experienced Wikipedians are often willing to help, but it would be wise to find one who is not involved with any active disputes you're covering.

However, a common form of diffpicking is less naïve than rhetorical, beginning with a narrative about Wikipedia or about one of the subjects it covers and selecting diffs that support the narrative, indifferent to context. Here are some questions you can ask to figure out if you're reading this type of manipulative writing.

It's worth remarking that there are many instances of news stories criticizing Wikipedia articles which failed to live up to Wikipedia policies and directly resulted in change. In such cases, while Wikipedians would generally prefer the journalist simply make the change themselves, this is a valuable form of constructive criticism.

Examples

[edit]

For example, in July 2022 screenshots circulated on Twitter depicting edits to the recession article, apparently showing removal of one of the term's definitions. They were accompanied by a narrative that Wikipedia altered a long-standing definition to protect the Biden administration, which had just released data that would qualify as a recession under the removed definition. The definition in question, which is used officially in some countries and as a rule of thumb in others, persisted throughout the dispute in the "definitions" section of the article. What was contested was an addition to the very top of the article characterizing it as being used by economists broadly. Putting aside the matter of who was correct, which has been settled after much discussion in our article, the narrative relies on omitting the first edit that kicked off the dispute. The material in question wasn't long-standing content, but newly added to the very top of the article, without citations, just before the data was to come out. It was then removed as unsourced and repeatedly reinstated. What could've just as easily been a story that "Wikipedia tried to elevate a particular definition of recession to make Biden look bad" was transformed into "Wikipedia tried to remove a particular definition of recession to make Biden look good" by omitting the first part of the dispute.

In 2024, The Free Press ran an article that used the page on U.S. executive branch czars to launch into a broader criticism of Wikipedia. The author claimed that users who accessed the list on July 24 would've seen Kamala Harris as "border czar", but it was removed a day later at 4:02 pm Eastern Time. Besides being demonstrably false to anyone who opened the page as it looked the day before (the last revision before July 25), more relevant for this essay is the edit at 4:02 removed not long-standing content, but material about Harris that had just been added a couple hours prior, once it had become a right-wing talking point to call Harris the "border czar". As elsewhere, the merits of the changes were then debated extensively on the talk page.

See also

[edit]