Wikipedia talk:Article size

Word counting example. Notepad++ vs Toolforge vs online word counters

For Gaza genocide article. This version from August 10, 2025 when this comment was started:

See discussion:

Notepad++ counts 19,292 words. Online word counters say 18,100. Toolforge says 14,146

Article is currently (Aug 10, 2025) way above 15K readable prose words. See: WP:TOOBIG. Greater than 15K: "Almost certainly should be divided or trimmed."

Notepad++ counts 19,292 words (view menu > summary) by pasting in everything above the "See also" section. Counting with a text editor is considered the best method for word count according to this:

But I disagree. I used the first 6 online word counters from this Google search:

All were around 18,100 words for everything above the "See also" section that I pasted in. Same as what I pasted into Notepad++.

I think Notepad++ is counting the reference numbers too. The reference numbers are jammed up against words without spaces. So I think the online word counters consider them to be part of the word they are jammed up against.

Wikipedia:Article size also recommends this:

It counts 14,146 words. But it misses many words, which you can tell by scrolling down the result page.

Wikipedia:Article size#How to find word count does not recommend the gadgets and toolforge because:

"other tools do not count words in image captions, lists, or tables."

The online word counters are giving the most accurate results. Some free working paste-in instant online word counters with easy-to-remember URLs:

--Timeshifter (talk) 20:15, 27 September 2025 (UTC)[reply]

Limits

What are the article size limits based on? Attention span doesn't make sense: the average adult attention span, as of 2023, was just over a minute, and I'm not sure we would want articles to be that short. Also, using an attention span limit presumes most readers read articles top to bottom, which isn't necessarily true. Processing power also doesn't make sense; these limits have been largely the same since 2007, and the average computer has dramatically increased in power since then. So what are these limits based on? Ladtrack (talk) 18:40, 26 November 2025 (UTC)[reply]

  • You have concisely summarized most or all of the reasons that this guideline is so ridiculous. But review the archives to marvel at the resistance to bringing sanity to it. EEng 23:15, 26 November 2025 (UTC)[reply]
    In other words: there is no sanity clause. Hawkeye7 (discuss) 03:12, 27 November 2025 (UTC)[reply]
    Hmm. Looking through it, I understand some of the potential issues people have brought up. In particular, long pages causing issues to dyslexic readers makes sense to me (although why exactly 15,000 words is the specific acceptable point is not something I understand), and I have personal experience with long pages causing lag issues with the visual editor. That said, I do think there is merit to increasing the size; for example, Douglas MacArthur, an FA, is still at nearly 20,000 words despite being sliced and diced into a number of sub-articles, and probably should not be any shorter. Also, dividing very tightly related articles into sub-articles solely for the purpose of reducing text size significantly harms reader experience. One suggestion is to simply multiply the upper bound restrictions by 1.5 (rounding when it's not an even thousand), so the top half of the table would become:
    > 23,000 words | Almost certainly should be divided or trimmed.
    > 14,000 words | Probably should be divided or trimmed, though the scope of a topic can sometimes justify the added reading material.
    > 12,000 words | May need to be divided or trimmed; likelihood goes up with size.
    This would catch certain edge cases like MacArthur, Spanish conquest of Honduras, and the Galileo project and somewhat loosening editing restrictions around other articles that are near the border, while still allowing for reasonable limits to prevent articles from simply ballooning in size. Does this seem reasonable? Ladtrack (talk) 05:42, 28 November 2025 (UTC)[reply]
    I think the current limits are more reasonable than those, which would promote articles simply ballooning in size. Nikkimaria (talk) 22:53, 28 November 2025 (UTC)[reply]
    • I personally don't see much use in the size restrictions, but would encourage tight editing. I propose:
      > 23,000 words | An interested editor could divide or trim this article. Otherwise, leave it alone, it's fine.
      > 14,000 words | One could consider dividing or trimming the article, though the scope of the topic can sometimes justify the added reading materials. If not looking for an editing project, leave it alone, it's fine.
      > 12,000 words | May be divided or trimmed if one is looking for a project. Otherwise, leave it alone, it's fine.
    • Obviously my opinion, but my suggestion on size is generally edit it you have the free time and are looking for a project. Otherwise, just leave the articles alone, size doesn't matter much anymore. Mburrell (talk) 06:52, 28 November 2025 (UTC)[reply]
      That would probably be ideal, but unfortunately, Wikipedia editors rarely work that way in my experience. Ladtrack (talk) 21:43, 28 November 2025 (UTC)[reply]

These limits are inline with academic norms and decades of research on accessibility of article length....that is all documented in previous discussions.Moxy🍁 23:12, 28 November 2025 (UTC)[reply]

The guidelines are not based on academic norms or research. They cannot be. At best, this is an ex post facto justification. I looked through when and how these guidelines formed. These current limits were set in this edit in March 2007, with the reasoning that the previous limit of 50 KB was too small an upper limit. Concerns about academic norms or accessibility are not mentioned in setting this limit. They then sat completely unchanged until fifteen years later, when they were converted into word count based on roughly judging the byte-to-word ratio of only ten articles (!). Needless to say, this also did not involve any research about accessibility or consultation of academic norms.
All academic research guidelines cited in the archive appear to have been found after the fact, and possibly selected specifically to align with the pre-existing Wikipedia guidelines. To see how true this was, I picked three of the four most prestigious academic journals and checked their word counts (the fourth, The New England Journal of Medicine, didn't have hard word limits). Of the other three: Nature has 4,300 words, Science has 3,000 words with an "extended format" of 6,000 words, and The Lancet has 3,500 words, with 4,500 under specific circumstances. Unless we want our articles to target about 4,000 words, academic norms don't seem like the best resource for determining the word count of an encyclopedia entry.
Could you explain what you mean by reader accessibility? There was a mention in the archives of dyslexia concerns, and I would understand concerns in that regard, but hypothetical concerns without any solid backing a good guideline do not make. In terms of reader attention span, I've already provided a study above that says the average attention span (for a young adult, all other groups are lower) is 76 seconds. The average time spent on an article, according to the only study I am aware of that involves this topic, is even lower at 25 seconds, with a 75th percentile of 75 seconds. According to this, our target word count should be somewhere between 100 and 300 words, based on average reading speed. This would result in quite an interesting encyclopedia.
Based on what I have been able to find, our guidelines are wholly unrelated to any and all academic norms and/or research. If you have other information that suggests otherwise, I would be grateful if you would be so kind as to provide it. Meanwhile, I have given specific examples of great articles that either currently run afoul of our guidelines, or have been made materially worse in an attempt to avoid doing so. I'm not asking for a huge expansion. Both of my suggested lower-limit restrictions are still below our current upper limit, they're just expanded to allow for more editor flexibility. The upper-limit one is a catch for unusual situations, which demonstrably do exist, and that our current guideline fails to handle. Ladtrack (talk) 05:39, 29 November 2025 (UTC)[reply]
"Basic information for Authors". 'Oxford University Press. The target length for an article is 8,000–10,000 words, excluding notes. Articles should not exceed 12,000 words including footnotes.Moxy🍁 20:44, 29 November 2025 (UTC)[reply]
This is actually still higher than our current guidelines, which start suggesting splitting at 8,000 words and strongly recommending it at 9,000 words when it is still firmly in the acceptable range according to Oxford, but more to the point, are we specifically targeting Oxford University Press's publishing standards? Because this seems to be well out of keeping with most academic journals as far as I can tell. Another two: Cell has under 7,000 words, and JAMA has 3,000 words. So unless we're specifically aiming to make our articles as long as those published by the University of Oxford (or perhaps other specific outliers), it doesn't seem to be in keeping with academic standards. We're not producing academic papers, so I really don't understand why we would let that dictate our guidelines (if we did, this still seems to me to be an ex post facto justification). Ladtrack (talk) 22:02, 29 November 2025 (UTC)[reply]
This is one of the points that our relationships with academic editors comes into conflict - as we have a norm that many are not use to. My academic mentor User:Rjensen made be able to explain better.Moxy🍁 00:46, 30 November 2025 (UTC)[reply]
To cater for the readers who only read a few hundred words, we provide the lead, a summary of the article. Many only read this, hence the low average time per article. Others home in on the section that contains specific information that they are looking for. They expect the article to be richly detailed. A minority read the whole article from top to bottom. Article size seems to have little or no impact on the readers.
Unfortunately, search engines often direct the reader to the main article on a subject even when a subarticle on the specific topic is available. This came to the fore in a discussion on the article on John von Neumann. While most articles are stewarded by a single project, this one was of major importance to several projects, and while the logical split of the article would have been to create subarticles for the different projects, most wanted the information sought by their readers to be in the main article.
The guidelines were not based on academic research, which was not available when they were written, but the image of what an encyclopaedia should look like, based upon the paper encyclopaedias of the 20th century. In paper encyclopaedias, pages cost money, so there was an incentive to keep the number of articles and their word counts down. But Wikipedia is not paper, so those constraints do not apply, and our objective is to produce a comprehensive encyclopaedia, hence we allow unlimited numbers of articles. Most importantly, it is now apparent, as it was not in 2004, that the readers do not access the articles in the same way that they accessed the old paper encyclopaedias. Hawkeye7 (discuss) 19:46, 29 November 2025 (UTC)[reply]
I can understand your point of view as in your alma mater has a very hight thesis acceptance. "Honours Handbook". UNSW Sites. 2025-11-28. Honours is an extra year of study, usually following immediately on from a Pass degree, that combines aspects of undergraduate study with aspects of postgraduate research. It introduces advanced research training through the completion of a 15,000-20,000 word research thesis Moxy🍁 01:18, 30 November 2025 (UTC)[reply]

Post-expand include size

I'm not sure where to ask this, but I'll put it here. Now that we have reached the year 2026, is there any possibility/desire/appetite for expanding the PEIS limit above 2048 kibibytes? I checked some related articles, but I can't tell when this limit was last adjusted, or if it is still a reasonable limit.

I know many readers use mobile devices, and come from all over the world. Document size (word count) is discussed in terms of minimizing the amount of data (html) sent to an article reader. But, as technology progresses, and hardware gets better all over the world, does the article size limit or the PEIS limit get updated to keep up? Some of the longer articles are among the most complex and also (IMO) the most informative in Wikipedia.

P.S. I see some Phab tickets, so this was discussed as long ago as 2018 (https://phabricator.wikimedia.org/T189108).

Thanks. David10244 (talk) 05:18, 2 January 2026 (UTC)[reply]

It looks like this limit was set in 2006, per https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(technical)/Archive_211#WP:PEIS. David10244 (talk) 05:54, 2 January 2026 (UTC)[reply]
@Tim Starling and PPelberg (WMF): The above is above my pay grade, what do you think? Thanks, Polygnotus (talk) 08:51, 2 January 2026 (UTC)[reply]
It looks like T275319 is a related discussion. It seems like it's not as easy as flipping a number in the server, though, since increasing the page size could have repercussions on other things like server load etc. Jo-Jo Eumerus (talk) 12:21, 2 January 2026 (UTC)[reply]
allikas: http://en.wikipedia.org/wiki/Wikipedia_talk:SIZE