Jump to content

Wikipedia talk:Wikipedia Signpost/2021-08-29/In the media

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Discuss this story

  • @JimmyWales must remember a different internet than I do. The internet has always been a great source of disinformation. That's what makes it such a wonderful marketplace for the good and bad. If it weren't for crap, you couldn't recognize quality. Not Wilkins (talk) 14:29, 31 August 2021 (UTC)NotWilkins[reply]
    • @Not Wilkins: Almost everybody chooses to remember the past wearing rose-colored glasses. Jimmy certainly did in that story. Or perhaps he did it simply as a rhetorical device - he can say it, but nobody was really expected to believe it. In either case, it brings up the question of how the disinformation now differs from the disinformation then. I'd say that back then it was more disorganized or chaotic. Now it looks more organized, or at least concentrated into a few groups. Another way to look at it is that now the disinformation is more politicized and commercialized. It always was - just more so now. Smallbones(smalltalk) 16:08, 31 August 2021 (UTC)[reply]
  • Some lovely articles here. I enjoyed "Why Basketball 3×3 Star Stefanie Dolson, Others Fact Check Their Wiki Pages", "There are 11,656 athletes at the Olympics. Guy Fraser wanted them all on Wikipedia" and "7 Notable African Women Activists Who Deserve Wikipedia Pages". All have issues (I'm reminded of a Signpost report I can't find summarising research that found that the media, wrongly, presents us as static, uniform-quality and mostly complete) but they report on Wikipedia well enough. The crypto nonsense is less lovely; apparently some bros who want to invent the biggest scam since pyramid schemes say that Wikipedia lacks "an advanced consensus mechanism" (we have many of these) and "a scalable and fast blockchain with smart contracts that implement any operating logic" (well... I suppose this we do lack). I see no reason to trust that they can introduce a system for paying editors that won't be gamed to oblivion and the ungrateful bastards don't get that our "uneven coverage" is the result of millions of hours of hard work that they simply will never get people to donate. — Bilorv (talk) 20:57, 1 September 2021 (UTC)[reply]

break

[edit]
  • TRIPODI: From above: "Dr. Tropodi, a researcher at the University of North Carolina, reports that biographies of women are more likely to be nominated for deletion than similar male biographies." - not true at all. Tripodi's paper neither tested this proposition, not made any such conclusion. What Tripodi's paper was actually about was well reported in the last issue, with an excellent analysis of the statistical flaws in her conclusions. Johnbod (talk) 14:52, 31 August 2021 (UTC)[reply]
    • @HaeB:'s review of the paper in the last issue, was indeed very good and I'll ask him to let me know if I've made a mistake here. Probably the most confusing part of Tripodi's paper was that it wasn't about discriminatory deletions of women's articles, it was about discriminatory nominations for deletion of women's biographies. She tried to show that there was a surplus of non-justified deletion nominations for women. *Not getting deleted* here is considered evidence of bias! HaeB pointed that out very well. But how to determine that there was a surplus of nominations for deletion? Just compare these nominations to similar nominations. She couldn't compare it to something totally dissimilar, e.g. AfD nominations of geography articles. So she picked something similar, men's bios. Perhaps I misstated when I wrote "similar male biographies" instead of "similar AfDs of male biographies". As pointed out very well by HaeB, it turns out the males bios really aren't that similar, e.g. different ages of the bio articles, and different ages of the subjects. Again HaeB addressed that very well. I'm sorry if leaving out 2 short words caused any confusion. BTW, I'm sure there must be a joke in here about Seemingly unrelated regressions vs. Totally unrelated regressions, but I'll let others take that risk.

Smallbones(smalltalk) 16:46, 31 August 2021 (UTC)[reply]

      • This is what Tripodi says: "I tested the following hypotheses:
H1. The proportion of biographies about cis-gender women (she/her/hers) nominated for deletion each month will be greater than the proportion of available biographies about cis-gender women (she/her/hers) on Wikipedia during the same time period.
H2. Articles about cis-gender women (she/her/hers) are more likely to be misclassified as non-notable (i.e. “kept”) than articles about cis-gender men (he/him/his).

- I'm not seeing anything about "biographies of women are more likely to be nominated for deletion than similar male biographies". As I suspect Tripodi knows, there is other recent research more relevant to the the male/female ratios, which I won't attempt to summarize here. Johnbod (talk) 16:53, 31 August 2021 (UTC)[reply]

H1 is clearly a typo in the original. It literally says H1: A > A. Not a promising hypothesis. The 2nd "cis-gender women (she/her/hers)" should be changed to "cis-gender men (he/him/his)" or H2:A>B. This just says that "biographies of women are more likely to be nominated for deletion than ,,, male biographies". ("similar" not included here)

"H2. Articles about cis-gender women (she/her/hers) are more likely to be misclassified as non-notable (i.e. “kept”) than articles about cis-gender men (he/him/his)" can clearly be restated as:

H2. Articles about cis-gender men (he/him/his) are less likely to be misclassified as non-notable (i.e. “kept”) than articles about cis-gender women (she/her/hers)", i.e. similarly nominated men's bios are less likely to be kept than women's nominations. She's trying to compare similar things, AfD nominations of men to AfD nominations of women.

Hope that helps. Smallbones(smalltalk) 17:52, 31 August 2021 (UTC)[reply]

  • Actually H1 could be correct as written. It is comparing the proportions of women's bios at AfD to their prevalence in all bios. In other words, if 15% of all biographies are about women, then an equitable proportion would be 15% of the biographies nominated for deletion are about women. ☆ Bri (talk) 18:23, 31 August 2021 (UTC)[reply]
    • Possible with an emphasis on "available" but seeing as she's talking about proportions i,e #W/(#W + #M) it comes down to the same thing - comparing the number of women's articles nominated to number of men's articles nominated. Smallbones(smalltalk) 18:53, 31 August 2021 (UTC)[reply]
  • Your first change is also completely wrong. The whole point of her research was that she was comparing, in proportion terms, nominated female biographies against "available" (ie all existing) female biographies, not against male ones. A rather dubious excercise, as pointed out last issue, but that is what she did! From her final section:

"My dataset revealed that the proportion of women nominated for deletion each month (out of all biographies nominated for deletion) was greater than the proportion of available biographies about women on English-language Wikipedia more generally." Johnbod (talk) 02:40, 1 September 2021 (UTC)[reply]

Somebody is missing something here

[edit]

@Johnbod: It could be me - but as I understand what you're saying is that Tripodi did not compare AfDs for men and women. Is that what you are saying? It is quite obvious that she did. Quoting from her paper:

  • "this article demonstrates that biographies about women who meet Wikipedia’s criteria for inclusion are more frequently considered non-notable and nominated for deletion compared to men’s biographies." (abstract)
  • "Specifically, my data indicate that biographies about women who meet Wikipedia’s criteria for inclusion are more likely to be considered non-notable than men’s." (intro)
  • "I analyzed nominations by month for the entire year of 2017, 2018, and 2019 and the first two months of 2020 (totaling 22,174 biographical entries around a she or he gender binary)." (i.e. dataset examined includes bios classified as male or female. (Data and methods)
  • "If no gender bias exists, the percentage of miscategorized biographies should not vary by gender." (Data and methods)
  • Figure 1. is a direct comparisons of men and women bios. "Men" and "Women" are labeled. (Findings)
  • "My data indicate that women’s biographies are more frequently miscategorized as non-notable than men’s (see Figure 2)." (Findings)

So men's and women's nomination are directly compared. What am I missing about your complaint? If she didn't mean these statement, according to your view, what do you think she meant? If I can figure out what you mean and consider it to be correct, then I will issue a correction. Until then. I have to ask that you not try to correct Signpost articles. That is a matter for staff of The Signpost and ultimately the editor-in-chief, me. Signpost articles are signed by the contributors, changing them is equivalent to changing somebody's comment on a talk page. Smallbones(smalltalk) 14:24, 1 September 2021 (UTC)[reply]

And I have to ask that you don't attempt to just rewrite what you think a (presumably peer-reviewed) paper ought to have said, when it clearly doesn't, as you did above (previous section). The analysis in your last issue was spot on; it's pity you didn't leave it at that. Many people including me have corrected various things in Signpost articles in the past, & I don't accept we can't. It's a bit rich saying that when you had just claimed the Tripodi paper must have meant something different to what it actually says! Johnbod (talk) 03:48, 2 September 2021 (UTC)[reply]
"Dr. Tripodi, a researcher at the University of North Carolina, reports that biographies of women are more likely to be nominated for deletion than similar male biographies," appears to be correct, as illustrated by the bullet points above. The analysis in Recent research presented in our last issue is indeed very impressive and readers and anybody who has any questions about the paper should consult it. I believe the statement above is entirely consistent with what the original paper says, as well as what HaeB's analysis says. No correction is needed.
As far as readers - or anybody else who is not on the staff of The Signpost - making post-publication "corrections" to the content of an article, that is inconsistent with the project's rules and 16 years of The Signpost's practice. We are an independent newspaper that presents the news truthfully to the best of our ability as well as our contributor's opinions. We do not necessarily represent - or claim to represent - the views of the WMF, its affiliates, ArbCom, admins, or even non-staffers who are part of our very diverse community. I believe that our readers would not want it otherwise. That means that The Signpost has the final say on our content. Please do not ever change our published content in opposition to the views of our staff. Of course the overall rules of enWiki apply here. The applicable rules are those that apply to WikiProjects and talk page content, e.g. do not change the content of another user's signed content. You may make a request at WP:MfD or even ArbCom if you disagree. Smallbones(smalltalk) 13:31, 2 September 2021 (UTC)[reply]

Wikipedia's "language problem"

[edit]
  • The "one Wikipedia" proposal is naive and vague, and can be dangerous if actually put into action. Forcing all Wikipedia language versions to present one monopolized version of the truth, which will inevitably be Anglophone-centered, would be a disaster that would completely infuriate the non-English Wikipedia communities, who tend to see the English Wikipedia as one of the most problematic language editions with its incomprehensible bureaucracy and often belligerent users. They would perceive this as intrusive digital colonialism and an outright invasion (see also Wikipedia:Ignore Meta). Also, imagine the level of cross-wiki vandalism and edit wars that would exponentially increase as a result.
While written with good intentions, the "language problem" article is clearly written by someone who does not know the details of why it is necessary to have separate projects with separate policies and content. Notability, copyright (especially regarding local fair-use laws), real-life inherent cultural differences, and many other aspects of wiki projects all need to be different, and none of these topics are addressed by the article.
Having original content in different language versions has benefits far outweighing the downsides. At most, we can use WikiLambda or something similar to import basic statistical facts into Wikipedia articles in different language versions, but this should not be extended to complex prose prone to POV problems. — Sagotreespirit (talk) 05:43, 10 September 2021 (UTC)[reply]