Wikipedia talk:Systemic bias

'Content reflects the bias in a source'

'Wikipedia content reflects the biases in the sources it uses. The neutral point of view (NPOV) policy requires articles to fairly and proportionately represent the views published in reliable sources. It does not permit editors to "correct" or remove biases they see in sources, or to allow their own beliefs and opinions to "get between" the sources and the content. Editors should put their own opinions aside and "stay out of the way" by neutrally documenting what a source says, including its opinions and biases. That means that when editors edit neutrally, Wikipedia content will reflect the biases found in reliable sources.'

I don't think that this is an adequate interpretation of NPOV. Whereas we are obliged to proportionately represent all views found in reliable sources, each separate reliable source does not have that same objective, so we can't just mimic them in making the same claims equally unconditionally and in the exact same way; unlike them, we need to take into account the existence of other views in other reliable sources as well. And we also need to write a single coherent text, not a collection of quotes.

To take the simplest example, when we have reliable source A saying 'Rap music is the worst music genre ever' and reliable source B saying 'Rap music is the best music genre ever', we aren't obliged to actually claim, in wiki voice, 'Rap music is the worst music genre ever' in one paragraph and 'Rap music is the best music genre ever' in the next one. That would be schizophrenic - we are writing a single text, whereas the sources we use are several. Since we know the views to be controversial, we have to state them only as the views of the sources we cite, not slavishly 'reflect' them. Similarly, while a relatively uncontroversial claim ('polar bears are carnivores') can be stated in a wiki voice, one known or likely to be controversial ('polar bears are evil') must be presented only as the view of the source we are citing, rather than faithfully 'reflect its bias'.

Furthermore, a reliable source may provide a useful and probably uncontroversial factual claim, but that doesn't mean that we have to render it together with the clear expressions of bias and more controversial positions with which it is connected. If a British reliable source on the Napoleonic wars says 'glorious Wellington defeated evil Bonaparte at Waterloo', it can be useful to cite it for the fact that the defeat happened at Waterloo, but we know that there are French reliable sources which would disagree with the 'glorious' and 'evil' part (and would instead insist that 'evil Wellington defeated glorious Bonaparte'). So what we should write is just 'Wellington defeated Bonaparte at Waterloo', not parrot the 'glorious' and 'evil' parts (as for whether they were glorious or evil, that can be discussed with proportionate representations in separate sections on their legacies). Indeed, when dealing with such obvious expressions of attitude, we are on the safe side omitting them in most cases.

However, there are also more subtle contentious issues - say, the word usages found in a source may reflect whether the author recognises a certain title, government, territorial possession as legitimate or not, and their position may be controversial or, indeed, a minority one. In such cases, we need to exercise editorial judgement and discriminate, not blindly copy every aspect of the text. Anonymous44 (talk) 09:42, 3 May 2026 (UTC)

sparsely attended discussions

I keep noticing a pattern of what appears to be systemic bias related to disambiguation-related (or really navigation-related) requested move discussions. It's not consistently observable in the outcomes, rather it is apparent in how these discussions are relatively sparsely attended, so the levels of consensus can vary, as it often depends on a relatively small amount of us with an interest in disambiguation.

Here's some examples that happen to involve gaming and recentism, topics already listed in the essay:

• Talk:Luigi + Talk:Luigi (character)
• Talk:Deadlock (computer science)#Requested move 13 August 2024 + Talk:Deadlock#post-move
• Talk:DLC (disambiguation)#Requested move 22 December 2025
• Talk:Turian#Requested move 1 April 2026
• Talk:Asari (Mass Effect)#Requested move 1 April 2026

These various discussions are technically advertized via the article alerts bot mechanisms, but it still seems like we're not getting editor attendance anywhere close to the level that might be commensurate to the readership. --Joy (talk) 20:48, 10 May 2026 (UTC)

As in a systemic bias towards treating video games as primary topics? Perhaps it could be encouraged to put a note on Wikipedia talk:WikiProject Disambiguation as well as the other WikiProjects, but I'm not sure that will cause a significant difference. CMD (talk) 22:44, 10 May 2026 (UTC)

Like I said, I don't think that is the outcome - just that we tend to focus on these sorts of things to an outsized extent.

For example, in the above list, the outcomes varied: we did eventually manage to 'dethrone' the recent game character from a common foreign name, and avoid pointing a TLA to a recent gaming meaning, but at the same time we removed a primary topic from a scholarly computing topic mostly in favor of a recent game, and set aside historical people in favor of a recent game topic.

It's the inconsistency that is problematic - the encyclopedia probably needs to be more coherent in presenting reality, and our biases make that harder.

I agree that it's hard to see how to actually influence this. Posting big "dude, chill." markers on everything related to recent topics would probably not have the intended effect :) --Joy (talk) 08:16, 11 May 2026 (UTC)