Lmst

Yoel Roth

Internet sanitation enthusiast and full-time corgi wrangler.

@thisismissem Your feedback and perspectives were so, so, so valuable. Thank you!

Government jawboning of tech platforms is a serious issue, and we need serious jurisprudence that draws appropriate limits on government conduct.

The 5th Circuit, regrettably, hasn't given us that in Missouri v. Biden. I wrote about how the problems with the case start with getting the facts wrong. https://knightcolumbia.org/blog/getting-the-facts-straight-some-observations-on-the-fifth-circuit-ruling-in-missouri-v-biden-1

@bookish @jeffjarvis No comment. 😐

Since 2020, I’ve been at the epicenter of a wide-reaching campaign to intimidate social media platforms into backing away from their investments in safety and security. It worked.

New from me (gift link): https://www.nytimes.com/2023/09/18/opinion/trump-elon-musk-twitter.html?unlocked_article_code=4tdIbFuKLW42ISeaU4acN26WTieKQcsLEoCyhJt1DC8dcAq9yCnJjyrbKLCEWm2hVWmWh-x94MKiw-I_OrqJ8JIYpDsdvQ4BFioWZ_RXCQ4ftJfFamVymL4ZnoK5RUQIhDdY-ZuJck3JBMeNXn5VYxEZ-tp8__DgJ_29osLV2tNCx4SZkrQrNtAyYPdzMK4asGiGrshlttyZF4arTjYH7ObwQo2-GSiVT3z3QovPSQ8Q4L9ggP7frVv1zKmIi4yukMwCGcqmRYnUy8pmnGPw0wWV3c9FMTUKuc6VM7kGy9gMnz_OUsQCiX8LR3v5Ls40VVkp1tb_c7PD4BiQ6lFP2Aw&smid=nytcore-ios-share&referringSource=articleShare

The whole piece is spectacular, but this bit in particular is just a perfect illustration of the axiom that the amount of effort required to refute bullshit is orders of magnitude greater than the effort required to produce it. https://www.theguardian.com/books/2023/aug/26/naomi-klein-naomi-wolf-conspiracy-theories

When looking at the Mirror World, it can seem obvious that millions of people have given themselves over to fantasy, to make-believe, to playacting. The trickier thing, the uncanny thing, really, is that’s what they see when they look at us. They say we live in a “clown world”, are stuck in “the matrix” of “groupthink”, are suffering from a form of collective hysteria called “mass formation psychosis” (a made-up term). The point is that on either side of the reflective glass, we are not having disagreements about differing interpretations of reality – we are having disagreements about who is in reality and who is in a simulation.

For instance, in July 2022, Wolf went on a rightwing podcast carried by something called Today’s News Talk and shared what she described as her “latest thinking”. She had noticed that when she went into New York City, where the vast majority of the population has been vaccinated, the people felt … different. In fact, it was as if they were not people at all.

“You can’t pick up human energy in the same way, like the energy field is just almost not there, it’s like people are holograms … It’s like a city of ghosts now, you’re there, you see them, but you can’t feel them.”

And she had noticed something even more bizarre: “People [who are vaccinated] have no scent any more. You can’t smell them. I’m not saying like, they don’t smell bad or they don’t smell – like I’m not talking about deodorant. I’m saying they don’t smell like there’s a human being in the room, and they don’t feel like there’s a human being in the room.”

This, she explained to the host, was all due to the “lipid nanoparticles” in the mRNA vaccines, since they “go into the brain, they go into the heart, and they kind of gum it up”. Perhaps even the “wavelength which is love” was experiencing this “gumming up … dialing down its ability to transmit”. She concluded, “That’s how these lipid nanoparticles work.”

That is not how lipid nanoparticles work. It is not how vaccines work. It is not how anything works. Also, and I can’t quite believe I am typing these words, vaccinated people still smell like humans.

Yoel Roth boosted:

I had a great conversation with @ethanz which you can read/listen to here.

https://publicinfrastructure.org/podcast/85-timnit-gebru/

Yoel Roth boosted:

IFTAS Issues New Moderator Trust and Safety Needs Assessment Survey

Announcement: https://about.iftas.org/2023/08/09/iftas-federated-trust-and-safety/

Moderator Needs Assessment: https://cryptpad.fr/form/#/2/form/view/thnEBypiNlR6qklaQNmWAkoxxeEEJdElpzM7h2ZIwXA/

#mastoadmin #MastoMods #Moderation #TrustAndSafety @moderation

Nvidia has started asking large GPU buyers who their end users are (per The Information): https://www.theinformation.com/articles/in-an-unusual-move-nvidia-wants-to-know-its-customers-customers?utm_source=ti_app&rc=gomoa3

Their focus seems to be on picking favorites in a crowded market, but this could actually be an interesting hook for "know your customer" diligence to prevent horrific abuses of AI, if they cared to do that.

Yoel Roth boosted:

Me at every work-related conference I've ever been to

Volodymyr Zelenskyy at a summit, standing on his own amongst a crowd, looking apprehensive

This is one of those cool insights at Meta’s scale that also only tells part of the story:

Are 90% of their reports visually similar because they’re failing to find and report stuff that ISN’T visually similar?

It’s easier to chase the threats you’re familiar with than to find novel ones.

Interesting tidbit from Meta staff at #TrustCon23 just now: >90% of the CSAM Meta report to NCMEC is visually similar to content they’ve reported before.

The argument goes: The same bad content circulates again and again, so effective moderation requires you to get very good at similarity detection.

Fascinating decision by the Oversight Board telling Meta to ban the Cambodian Prime Minister for inciting violence: https://oversightboard.com/news/656303619335474-oversight-board-overturns-meta-s-decision-in-cambodian-prime-minister-case/

Meta have said they’ll comply.

It’s worth noting that the content in question was posted in January, and it’s now, *checks notes*, uh, June.

So it’s not clear that the Oversight Board process is totally working (yet?) as a way to make difficult, life-and-death decisions in the timeframe trust and safety work requires.

The Atlantic Council Task Force for a Trustworthy Future Web, led by @rightsduff, put out its comprehensive report today: https://www.atlanticcouncil.org/in-depth-research-reports/report/scaling-trust/

I had the privilege of leading work on one of the report's annexes, specifically focused on securing federated platforms: https://www.atlanticcouncil.org/in-depth-research-reports/report/scaling-trust_annex5/

Bottom line up front: We're missing some key policy, technical, and institutional pieces right now, but these are solvable challenges.

@mmitchell_ai @huggingface @giadap These are so, so good. I’m a huge fan of the clear articulation of consequences in the policy itself, too, and how readable it all is. 👏👏👏

Yoel Roth boosted:

Today's podcast is a fun one with @yoyoel talking about the challenges of doing trust & safety on a decentralized/federated system... Potentially of interest to folks here. https://www.techdirt.com/2023/06/13/techdirt-podcast-episode-354-decentralizing-content-moderation/

Yoel Roth boosted:

Now, a note on the Fediverse: SG-CSAM is not really a thing on here (other kinds are — thanks Japan), but the reason is simple: the Fediverse isn't popular enough to make it profitable. So don't gloat about this just yet, there are massive T&S issues that the Fediverse is going to get hit with that it is extremely ill-prepared for. As far as I know, no instance even has table stakes CSAM protections. Get on it.

It’s unconscionable that Twitter would deploy what remains of its legal team to bully academics into paying an obscene ransom in order to keep access to essential data. https://inews.co.uk/news/twitter-researchers-delete-data-unless-pay-2364535

The best hope for stopping this is regulatory action, particularly under the DSA.

Groups like the Coalition for Independent Technology Research are helping lobby on behalf of researchers. Learn more and get involved: https://independenttechresearch.org/

But in recent weeks, the company has been contacting researchers, asking them to pay $42,000 a month to access 0.3% of all the tweets posted to the platform – something researchers have previously said is totally unaffordable. Previous contracts for access to the data were set as low as a couple of hundred dollars a month.

An email, seen by the i, says researchers who don’t sign the new contract “will need to expunge all Twitter data stored and cached in your systems”. Researchers will be required to post screenshots “that showcase evidence of removal”. They have been given 30 days after their agreement expires to complete the process.

This piece about industrialized catfishing services is absolutely fascinating. Exploitation at every level: of the clients, and of the “freelancers.” https://arstechnica.com/culture/2023/05/this-is-catfishing-on-an-industrial-scale/

$While sex chat and call lines have been around forever, there is generally a mutual understanding that these are fee-for-service arrangements. If users are instead wooed by ads offering a real dating application and then freelancers are paid to continue this charade, “it constitutes real deception,” he says, “and arguably exploitation on both sides of the Internet—both the victims and the workers.” Freelancers working in the industry say they make a fraction of the money users are paying. Workers earn around 7 cents per message, or 2 euro an hour. For the company where Liam freelanced, the shifts were six hours a day, six days a week, “and they want you to be active on screen all that time—no breaks,” he says. The 36-hour weeks got him around 400 euro a month. “It was a pittance,” he says. In the murky, anonymous, globally dispersed industry, even some of the freelancers aren’t who they say they are—and are working for even less. In Lagos, 22-year-old physics student Idris—who asked for anonymity to protect his privacy—started working for Cloudworkers, a Swiss company, in June 2020. The company doesn’t actually hire people in Nigeria, so he has to sublet an account from someone else. He gets 40 percent of the fee paid to the account’s owner. Cloudworkers didn’t respond to requests for comment.$

Yoel Roth boosted:

Okay folks: this morning we're launching something we think is pretty useful. Lots of people have strong opinions on how content moderation should work, but they've never done it. So we built a content moderation mobile game (browser-based): https://moderatormayhem.engine.is/

Details about it are here: https://www.techdirt.com/2023/05/11/moderator-mayhem-a-mobile-game-to-see-how-well-you-can-handle-content-moderation/

@conspirator0 Interesting to see that the fake accounts seem to largely be on mastodon[.]social. I’d have expected them to distribute across more instances.

Client Info

Server: https://mastodon.social

Version: 2025.04

Repository: https://github.com/cyevgeniy/lmst