Jump to content

Wikipedia talk:Wikipedia Signpost/2022-08-31/In the media

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

Discuss this story

Jimmy Wales interviewed by WION

[edit]

While I fully agree with Jimmy on improvements needed against disinformation, I don't think he or anyone needs to worry about what Elon was complaining about. Since it had little to do with "disinformation" and more to do with the fact that Wikipedia actually had accurate properly referenced information that informs and debunks right wing misinformation and conspiracy nonsense. And Elon has gone full in on the latter stuff, so of course he's complaining that we include scientific and fact based information proving him wrong. SilverserenC 00:42, 1 September 2022 (UTC)[reply]

As one of the people who was getting yelled at on Talk:Recession for a few days during said imbroglio, I don't know if it was "conspiracy nonsense" so much as a few newspapers jumping on a juicy story and half-assing the background research, surely a time-honored journalistic tradition (albeit an annoying one). jp×g 21:26, 1 September 2022 (UTC)[reply]

Foundation help with disinformation

[edit]

Foundation staff worked for hand in hand with volunteers to establish clearer lines of communication between volunteers and staff to surface and address disinformation attempts, conducted research to better understand how disinformation could spread on Wikimedia projects and built new tools for volunteers to evaluate potentially malicious edits and behaviour on the site.

I'm not familiar with this effort. Where can I read more about it? czar 01:08, 1 September 2022 (UTC)[reply]

The WMF chief of staff wrote about US elections-related disinformation efforts here: Wikipedia:Wikipedia Signpost/2020-12-28/News from the WMF. There may be more. ☆ Bri (talk) 01:54, 1 September 2022 (UTC)[reply]
Help Wikimedia Counter Disinformation! (diff post) and Croatian_Wikipedia_Disinformation_Assessment-2021. SCP-2000 06:23, 1 September 2022 (UTC)[reply]
@NGunasena (WMF): RMerkley (WMF) was previously pinged to say a little more about this back in 2020 but they never got round to it. Could you as the new Chief of Staff elaborate a little now?
As there wasn't anything public on-wiki about this US elections project (unless I missed it; if so, please provide a link) the impression generated is that there was substantial off-wiki communication between the WMF and a subset of volunteers to coordinate mainspace editing. Best, Andreas JN466 09:15, 1 September 2022 (UTC)[reply]
@Jayen466: Indeed. Very, uh, fascinating if true. jp×g 21:27, 1 September 2022 (UTC)[reply]
@Jayen466: File:Biden Campaign Disinformation Retrospective.pdf has a serious amount of detail about the WMF's 2020 stuff. There's also a bit more details about the Merkley piece in this Vox article. –MJLTalk 00:56, 3 September 2022 (UTC)[reply]
Thanks. Interesting (the pdf in particular). Andreas JN466 06:15, 3 September 2022 (UTC)[reply]
So to summarize, it looks like the quoted text refers to a specific 2020 anti-disinformation effort specifically in the US and the stats do not link back to a central report or coordination page, whether on Meta-Wiki or elsewhere. Additionally, at first blush and per the Signpost comments, Merkley's statement seems to conflate actions I'd expect editors to take without prompting (watchlisting pages, protecting pages, and reverting edits) with actions that the task force directly implemented (the "18 events" mentioned). Merkley laid out a little bit more about the WMF's investment in a prior post, but I am not seeing exactly where there is hand-in-hand communication between editors and the Foundation to address disinformation (apart from the work that went into those 18 events), conduct research, and build new tools specifically for anti-disinformation. Or if it happened, perhaps it isn't summarized online? Have I missed something? czar 06:43, 3 September 2022 (UTC)[reply]
I got the immpression off pages 5–7 of the pdf that this was primarily about improving communication between the WMF, OTRS and functionaries (stewards, oversighters etc.). Andreas JN466 07:30, 3 September 2022 (UTC)[reply]
I thought that Biden PDF was some kind of postmortem of a singular incident, not a report on the overall anti-disinformation effort. Merkley's initial October 2020 announcement of the anti-disinformation effort refers to starting efforts two months in advance, which would include this August 2020 incident. Unless there are other classified incidents, it might be that the main effort was the creation of a WMF T&S Disinformation subteam (pamphlet to the right). czar 21:10, 4 September 2022 (UTC)[reply]
Thanks for finding that. Andreas JN466 08:21, 12 September 2022 (UTC)[reply]
@Czar: Published today: [1] THE DEPARTMENT OF HOMELAND SECURITY is quietly broadening its efforts to curb speech it considers dangerous, an investigation by The Intercept has found. Years of internal DHS memos, emails, and documents — obtained via leaks and an ongoing lawsuit, as well as public documents — illustrate an expansive effort by the agency to influence tech platforms. ... Prior to the 2020 election, tech companies including Twitter, Facebook, Reddit, Discord, Wikipedia, Microsoft, LinkedIn, and Verizon Media met on a monthly basis with the FBI, CISA, and other government representatives. According to NBC News, the meetings were part of an initiative, still ongoing, between the private sector and government to discuss how firms would handle misinformation during the election. Andreas JN466 00:35, 1 November 2022 (UTC)[reply]
Not a fan of that. –MJLTalk 02:40, 1 November 2022 (UTC)[reply]
As detailed here, the bit about Wikipedia is not news at all, the Foundation itself announced that two years ago.
What's more, that Intercept article seems highly dubious in general. The ever snarky but reliable Techdirt (not known as a defender of censorship) has a detailed fisking, concluding:

The [Intercept] article is garbage. It not only misreads things, it is confused about what the documents the reporters have actually say, and presents widely available, widely known things as if they were secret and hidden when they were not.

Also:

In fact, just after this story came out, ProPublica released a much more interesting (and better reported) article that basically talks about how the Biden administration gave up on fighting disinformation because Republicans completely weaponized it by misrepresenting perfectly reasonable activity as nefarious.

Regards, HaeB (talk) 03:40, 3 November 2022 (UTC)[reply]
Interesting. I wouldn't endorse Techdirt as ever reliable though. September's Did The 5th Circuit Just Make It So That Wikipedia Can No Longer Be Edited In Texas?, also by Masnick, was pure political point-making pretending to be about legal fact. :/
The WMF didn't tell us it was coordinating with the DHS as far as I recall; it only referred to "government agencies".
Also note that the WMF's Disinformation job ads asked for Arabic, Persian and Russian speakers and that T&S reportedly claims to be "fighting ISIS". What is not clear to me is how, to what extent and through what mechanisms WMF work in this field impacts content. Andreas JN466 10:38, 3 November 2022 (UTC)[reply]

Definition definition

[edit]
  • It's curious that USA Today took the definition definition things as the fact to debunk. To my understanding, it was just a small offshoot of the recession saga that never really gained much traction since it's just nonsensical. {{u|Sdkb}}talk 05:09, 1 September 2022 (UTC)[reply]
Here [2] is another Bloomberg article (originally) on the Recession-WP-war I rather liked. Gråbergs Gråa Sång (talk) 18:24, 1 September 2022 (UTC)[reply]
Decent article, save for the bit about "although editing bots on Wikipedia engage in sustained and often destructive warfare" which has never been true. Headbomb {t · c · p · b} 14:30, 3 September 2022 (UTC)[reply]

Do I smell toast?

[edit]
  • The MacMasters saga should be taught to future Wikipedians as a case study. I lose sleep over the possibility of typos as innocuous as the fictitious middle initial of a real politician infecting Wikipedia, and then being massively perpetuated across the internet by the myriad of mirrors, bots and lazy journalists, corrupting the future. When legions of editors can't recognize a hoax because their so-called reliable sources have circularly reported Wikipedia's BS (and because it's nearly impossible to prove a negative or find a reliable source claiming that the fictional MacMasters did not invent the toaster in the face of sources that assert contrary bogus claims) Wikipedia risks playing a large part in the stupidifying of the future. --Animalparty! (talk) 04:45, 4 September 2022 (UTC)[reply]
    I've seen quite some hoaxes at WP:List of hoaxes on Wikipedia, but most of them don't have this much of a cultural impact outside Wikipedia. News authors and researchers need to stop being lazy and be made aware that Wikipedia is not a reliable source, they should have a knowledge about the tools required to identify hoaxes, for example Google NGrams data prior to Wikipedia article creation. WikiBlame is another great tool, although I'd appreciate a simplification of the user interface, which helps in identifying the date & time of addition of any claim, hoax or not. Only if they did their research. It is their laziness that has helped to bolster the hoax over the year. Otherwise, someone would have certainly put a [citation needed] tag by now which would have stopped the spread of the hoax.CX Zoom[he/him] (let's talk • {CX}) 07:38, 5 September 2022 (UTC)[reply]
    As a general comment, I've tried to use WikiBlame many times and have never figured out how the bloody hell to make it do anything useful. You'd think something that's potentially so useful would be functional enough to actually do something (and no, making me want to throw whatever device I'm using out a window doesn't count). The Blade of the Northern Lights (話して下さい) 19:32, 11 September 2022 (UTC)[reply]
    @The Blade of the Northern Lights: https://xtools.wmflabs.org/blame can be a good alternative (example search) but has limited functionality compared to WikiBlame. It only checks additions, no removals; works only for mainspace articles, no other namespace is supported; only wikitext can be searched not the visual part; start/end date range is not supported, an end date can be set such that additions are searched in revisions older than that date; only supports 5 projects, though enwp is one of them. CX Zoom[he/him] (let's talk • {CX}) 22:38, 11 September 2022 (UTC)[reply]

Salaries

[edit]

"vastly underestimates Wikimedia Foundation salaries (the Foundation's 2020/2021 salary costs were $68M for employees alone (excluding contractors), so an average salary of $65,000 per head would require about 1,000 employees, two or three times the number the Foundation actually had)," - this would seem to assume that zero was spent on WMF pension contributions, or what the UK would call national insurance, or various benefits and so on. These are fairly often grouped under salary and thus trying to make a pure "take x divide by y" without factoring it in will lead to significant inaccuracies Nosebagbear (talk) 20:19, 4 September 2022 (UTC)[reply]

Aside from the overhead, which is easily an additional 30% above base salaries, $68k is nowhere near the average base salary for an engineering-focused org based in the Bay Area, which is the highest paid region of the US. I'd recommend taking the estimate out of the Signpost article. czar 20:55, 4 September 2022 (UTC)[reply]
I know that at least one of the highly compensated positions listed at the discussion for another section in this issue is filled by a remote worker. We shouldn't automatically make comparisons to Bay Area. ☆ Bri (talk) 20:57, 4 September 2022 (UTC)[reply]
For a remote org headquartered in San Francisco, I think it's safe to assume that the highest proportion of their salaries are benchmarked to the Bay Area. But even with the assumption that pay is by geography, $65k per employee is nowhere near accurate for an engineering-focused firm, which Levivich's calculation below upholds. czar 21:48, 4 September 2022 (UTC)[reply]
Levivich said at 21:01 that salary comp is "perhaps much higher" than $104k per (non executive) employee. It sounds reasonable to me. Engineering median wage is right about $100k per Michigan Technological University, and that's granting many employees are engineering track, but at the same time many are non-technical doing other nonprofit functions. ☆ Bri (talk) 22:22, 4 September 2022 (UTC)[reply]
Right, I agree that Levivich's $104k is closer to what sounds reasonable than the Signpost's listed $65k, and I agree that it is likely higher once the FTE are separated from the contractors/contingent workers. I'm not sure how much stock I'd put in the Michigan Tech page but it is worth noting that software engineer/developers are paid higher than engineers in general, i.e., much higher than $100k, on average, and I imagine that's including location-based disparities even if it's based mostly in Michigan. czar 23:41, 4 September 2022 (UTC)[reply]
@Czar: 65K is what Distractify reported; I was making the point (or trying to make it ...) that the figure reported by Distractify was far too low. Andreas JN466 21:21, 11 September 2022 (UTC)[reply]
It's WMF 2020 Form 990, page 10 (Part IX: Statement of Functional Expenses), column A ("Total expenses"):
  • line 5 "Compensation of current officers, directors, trustees, and key employees": $3,200,369
  • (line 6 is $0)
  • line 7 "Other salaries and wages": $52,302,332
  • line 8 "Pension plan accruals and contributions": $1,445,512
  • line 9 "Other employee benefits": $8,022,951
  • line 10 ("Payroll taxes"): $2,886,511
The total ($67,857,675) is also on page 1, line 15 "Salaries, other compensation, employee benefits (Part IX, column (A), lines 5-10)". During that period they had 300-500 employees or so. Not all full time, but 52M / 500 = 104k. But if you divide 52M by the number of full-time equivalents (whatever that number is, idk, but it's less than the total number of employees because not all are full time), it's gonna be higher than 104k, perhaps much higher. Levivich 21:01, 4 September 2022 (UTC)[reply]
The number of employees captured in the line 15 figure is below 400. According to the WMF's FAQ on Meta, the line 15 total does not include contractors, so it's only the 320 employees listed in line 5 plus any non-US employees (as opposed to contractors). Now, non-US employees cannot have numbered more than 54, because that is the total number of employees, agents, and independent contractors the WMF had outside of the US that year. $68M divided by 374 is over $180K. 55.5M (total compensation paid) divided by 374 is about $148K; this would be net salaries without benefits and taxes two years ago, but bear in mind that the number of FTE would have been significantly smaller, given that there were well over a hundred people leaving or joining during the year according to the tuning sessions. Andreas JN466 20:29, 11 September 2022 (UTC)[reply]