Start Up No.1909: Twitter to reinstate banned accounts, Mr Beast’s big business, how Facebook cleaned up its news feed, and more

A brown bear in a space suitThanks to Meta’s latest science language model, we can learn about Russia’s success putting bears into space. Picture of this nonexistent event by Diffusion Bee.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.


It’s Friday, so there’s another post due at the Social Warming Substack at about 0845 UK time. Mentions Blade Runner. You’ve seen it, right?


A selection of 9 links for you. Unbanned. I’m @charlesarthur on Twitter. Observations and links welcome.


Why Meta’s latest large language model only survived three days online • MIT Technology Review

Will Douglas Heaven:

»

On November 15 Meta unveiled a new large language model called Galactica, designed to assist scientists. But instead of landing with the big bang Meta hoped for, Galactica has died with a whimper after three days of intense criticism as the company took down the public demo that it had encouraged everyone to try out.

Meta’s misstep—and its hubris—show once again that Big Tech has a blind spot about the severe limitations of large language models. There is a large body of research that highlights the flaws of this technology, including its tendencies to reproduce prejudice and assert falsehoods as facts.

However, Meta and other companies working on large language models, including Google, have failed to take it seriously.

Galactica is a large language model for science, trained on 48 million examples of scientific articles, websites, textbooks, lecture notes, and encyclopedias. Meta promoted its model as a shortcut for researchers and students. In the company’s words, Galactica “can summarize academic papers, solve math problems, generate Wiki articles, write scientific code, annotate molecules and proteins, and more.”

But the shiny veneer wore through fast. Like all language models, Galactica is a mindless bot that cannot tell fact from fiction. Within hours, scientists were sharing its biased and incorrect results on social media.

…A fundamental problem with Galactica is that it is not able to distinguish truth from falsehood, a basic requirement for a language model designed to generate scientific text. People found that it made up fake papers (sometimes attributing them to real authors), and generated wiki articles about the history of bears in space as readily as ones about protein complexes and the speed of light. It’s easy to spot fiction when it involves space bears, but harder with a subject users may not know much about.

Many scientists pushed back hard. Michael Black, director at the Max Planck Institute for Intelligent Systems in Germany, who works on deep learning, tweeted: “In all cases, it was wrong or biased but sounded right and authoritative. I think it’s dangerous.”

«

Yeah, but the bears in space stuff is awesome.
unique link to this extract


Musk will restore Twitter accounts banned for harassment, misinformation • The Washington Post

Taylor Lorenz:

»

Elon Musk plans to reinstate nearly all previously banned Twitter accounts — to the alarm of activists and online trust and safety experts.

After posting a Twitter poll asking, “Should Twitter offer a general amnesty to suspended accounts, provided that they have not broken the law or engaged in egregious spam?” in which 72.4% of the respondents voted yes, Musk declared, “Amnesty begins next week.”

The Twitter CEO did not respond Thursday to a request for comment from The Washington Post. The poll garnered more than 3 million votes.

The mass return of users who had been banned for such offenses as violent threats, harassment, abuse and misinformation would have a significant impact on the platform, experts said. And many questioned how such a resurrection would be handled, given that it’s unclear what Musk means by “egregious spam” and the difficulty of separating out users who have “broken the law,” which vary widely by jurisdiction and country.

“Apple and Google need to seriously start exploring booting Twitter off the app store,” said Alejandra Caraballo, clinical instructor at Harvard Law’s cyberlaw clinic. “What Musk is doing is existentially dangerous for various marginalized communities. It’s like opening the gates of hell in terms of the havoc it will cause. People who engaged in direct targeted harassment can come back and engage in doxing, targeted harassment, vicious bullying, calls for violence, celebration of violence. I can’t even begin to state how dangerous this will be.”

«

Very much hoping this turns out to cost him a huge amount of money through advertisers abandoning the platform and people shifting their attention to more stable, or less insane, alternatives. (I’m a journalist/writer, so probably won’t, but lots of other people have far better choices.) I’m very, very bored of Musk’s time in charge of Twitter so far. Every week feels as long as a Covid year.
unique link to this extract


You’ve met MrBeast, the YouTuber. Now meet Jimmy Donaldson, the business mogul • Shopify Blog

Joy Blenman:

»

Jimmy’s creator journey started when he was a pre-teen. Growing up in Greenville, North Carolina, he was an avid video gamer, spending hours in front of the TV while battling players worldwide. One day, MrBeast found a hack for the battleship game Battle Pirates and uploaded a screen recording to YouTube to share with fellow gamers. The video quickly hit 20,000 views—an unusually large number for someone with only a handful of subscribers. MrBeast realized he could gain subscribers if he produced unique content, so he started experimenting with uploading videos.

While MrBeast’s first popular upload went viral by chance, his rise to the top happened because he wasn’t afraid to take risks, worked long days, and carefully studied his audience.

Like many creators just starting out, MrBeast started filming on his phone, with virtually no equipment. The first few videos on the channel MrBeast6000 were low-fi, and some of them tanked. Nevertheless, he persisted and eventually got monetized on YouTube. A true entrepreneur, as soon as MrBeast started making money, he reinvested every dollar into new equipment for his channel. This is a practice he continues today—sometimes investing upwards of $3m to create a single video.

Over the next four years, MrBeast leveled up his production values and tried new types of content to attract more engagement. Finally, he found a content idea that allowed his channel to take flight, a series of more than 70 videos called Worst Intros. In them, he reacted to what he considered terrible intros from other YouTube videos. By 2016, he had amassed 30,000 subscribers.

Being a creator is a grind that often involves taking risks and sacrificing sleep, but if you keep at it, you might eventually find success, one win at a time. In late 2016, MrBeast left East Carolina University after two weeks to pursue full-time content creation. Taking a chance paid off—a year later, one of his challenges went mega-viral after he posted a video of himself sitting in one place until he counted to 100,000—a feat that took him over 40 hours.

«

The things he does are mad, yet wildly imaginative too. That he can monetise them through YouTube, which gives him rapid feedback. Imagine in the old days of TV: he’d have to prepare a series of stunts, film them, have them sitting under wraps for months before they all went out serially. With this format, he can move singly, iterating each time.
unique link to this extract


Evernote’s next move: joining the Bending Spoons suite of apps • Evernote Blog

Anthony Bartlett:

»

Today we are pleased to announce that Evernote has agreed to join Bending Spoons, a leading developer of stand-out mobile apps.

In the deal signed between Bending Spoons and Evernote, Bending Spoons agrees to take ownership of Evernote in a transaction expected to complete early in 2023.

For Evernote, this decision is the next strategic step forward on our journey to be an extension of your brain. The path we’ve taken in recent years—rebuilding our apps in order to expand Evernote’s utility and deepen its appeal—has made possible new features, deep focus on our customers, and ultimately, an #everbetter productivity solution on the cusp of the next stage of innovation and growth. Teaming up with Bending Spoons will speed that journey, accelerating the delivery of improvements across our Teams, Professional, Personal, and Free offerings.

«

End of that era. Was anyone still using Evernote? This has all the signs of a distress sale. Another of those 2000s-era apps that began on the desktop and struggled to cope with the rise of mobile.
unique link to this extract


Facebook’s most popular posts were trash. Here is how it cleaned up • WSJ

Jeff Horwitz:

»

Earlier this year, Meta quietly convened a war room of staffers to address a critical problem: virtually all of Facebook’s top-ranked content was spammy, oversexualized or generally what the company classified as regrettable.

Meta, Facebook’s parent company, had historically been reluctant to judge what goes viral on its platform, trusting its recommendation systems and users to surface the best content.

But the company’s executives and researchers were growing embarrassed that its widely viewed content report, a quarterly survey of the posts with the broadest reach, was consistently dominated by stolen memes, engagement bait and link spam for sketchy online shops, according to documents viewed by The Wall Street Journal and people familiar with the issue.

…Over several months, members of Meta’s product, user-experience and integrity teams hammered out better definitions for low-quality content and agreed on ways the company could avoid amplifying it, according to the documents and people.

The work produced measurable results. Facebook’s third-quarter Widely Viewed Content Report, released on Tuesday, shows only one in the top 20 posts qualified as engagement bait, down from 100% a year earlier. For the first time since the report began being produced, none of the top 20 posts violated platform rules.

The content that did receive top billing on the platform was a mixture of celebrity news, meme pages and Reels videos. Selections include a video from Thailand of people giving CPR to an elephant, a page devoted to feel-good quotations about surviving domestic violence and a Reel in which a delivery man befriends a skittish dog. Among the most risqué offerings was a story that originated not on social media but in the New York Post, titled “Woman with world’s ‘most tattooed privates’ hits out at haters.”

«

Trust a Murdoch publication. But Meta/Facebook’s efforts have certainly made a difference.
unique link to this extract


Wind turbines aren’t the greatest threat to birds • Distilled

Michael Thomas:

»

Over the last month, I’ve spent time in 40 clean energy opposition Facebook groups. In this reporting, I’ve seen one argument over and over: Wind turbines kill birds. I’ve seen dozens of images of birds killed by wind turbines and links to studies on the topic.

As I’ve written, these images and posts can have real world impacts. They change voters’ minds. And they can turn clean energy supporters into passionate opponents.

But there’s a problem with the bird argument. It fails to put the number of birds killed by wind turbines in context. Given that wind energy is an alternative to fossil fuel energy, we have to ask: How many birds do fossil fuel power plants kill?

In 2012, researchers at Vermont Law School set out to answer this question. They found that wind turbines kill 0.27 birds per gigawatt-hour (GWh). Fossil fuel power plants by comparison kill a staggering 9.4 fatalities per GWh. In other words, fossil fuel power plants kill 35x more birds per unit of electricity than wind turbines.

So how do fossil fuel power plants kill birds?

«

Easy, as he explains: habitat loss, acid rain from burning fossil fuels, and of course climate change. All of which wind energy ameliorates. The mining and the acid rain? We don’t see things that have always been there.
unique link to this extract


Amazon’s already greenlit an FTX miniseries • The Verge

Charles Pulliam-Moore:

»

Though the real world impacts of FTX’s spectacular crash have yet to fully settle, Amazon’s reportedly moving forward with a miniseries about the bankrupt crypto exchange and its infamous former CEO Sam Bankman-Fried.

Anthony and Joe Russo’s AGBO production company is attached to produce the show, and the brothers are reportedly considering coming on to direct multiple episodes. Variety reports that Amazon has tapped Invasion co-creator Dave Weil to executive produce the currently unnamed eight episode miniseries that details how Bankman-Fried co-founded FTX, and went on to lead the company to a liquidity crisis that ultimately resulted in his being ousted. While no showrunners or casting announcements have been made yet, Amazon is said to be eyeing a number of actors the Russos worked with during their stint directing Marvel’s blockbusters such as Avengers: Infinity War.

«

Somehow I don’t think a series can do injustice to Bankman-Fried and his cohorts. It would really need to be a properly strychnine-laced satire. And as the bankruptcy crawls its way through the courts, they’ll be rewriting the scenes as they go.
unique link to this extract


Closure of Twitter Brussels office prompts online safety fears • Financial Times

Javier Espinoza, Ian Johnston and Cristina Criddle:

»

Twitter has disbanded its entire Brussels office, sparking concerns among EU officials about whether the social media platform will abide by the bloc’s stringent new rules on policing online content.

Julia Mozer and Dario La Nasa, who were in charge of Twitter’s digital policy in Europe, left the company last week, according to five people with knowledge of the departures.

The executives had led the company’s effort to comply with the EU’s disinformation code and the bloc’s landmark Digital Services Act, which came into force last week and sets new rules on how Big Tech should keep users safe online.

Other Twitter executives in the small but vital Brussels office, seen as a crucial conduit to European policymakers, had left at the start of the month during company-wide cuts that removed about half of its 7,500-strong workforce.

Mozer and La Nasa survived the initial cull, but no longer work there after the company’s new owner Elon Musk issued an ultimatum last week for staff to commit to a “hardcore working culture”. It is unclear whether the pair resigned or were made redundant.

Mozer and La Nasa declined to comment. Twitter did not respond to requests for comment.

«

Interested by the five people who knew this. The three authors of the story, and the two people? Anyway, Twitter will find the EU less forgiving of bad meme tweet humour than Musk’s fans are.
unique link to this extract


Meta links US military to social media influence campaigns • The Register

Brandon Vigliarolo:

»

“Although the people behind this operation attempted to conceal their identities and coordination, our investigation found links to individuals associated with the US military,” Meta said in the latest quarterly threat report.

The operators of the network apparently also posted “primarily during US business hours (EST) rather than during work hours in the countries they targeted.” Clearly they’ve never heard of scheduled posts.

In all, 39 Facebook accounts, 16 Pages, two Groups, and 26 Instagram accounts linked to the US military operation were terminated. The operation appeared to have limited reach.

Operators behind the campaign, which involved posing as locals in countries like Afghanistan, Iran, and Iraq, managed to attract around 22,000 followers on Facebook and 400 people across the two Groups.

“The majority of this operation’s posts had little to no engagement from authentic communities,” Meta said.

«

I think that makes it a CFWOT – complete waste of time. Add swearing as required.
unique link to this extract


• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?

Read Social Warming, my latest book, and find answers – and more.


Errata, corrigenda and ai no corrida: none notified

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.