Start Up No.1665: a new Facebook whistleblower (with whistle), YouTube’s slow train to truth, inside Apple’s chips, and more

A Facebook account set up to follow right-wing news sources quickly spiralled into being shown wilder, lunatic content, internal research found. CC-licensed photo by Mike MacKenzie on Flickr.

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.

A selection of 11 links for you. Click with care. I’m @charlesarthur on Twitter. Observations and links welcome.

What Facebook knew – and when – about how it radicalized users • NBC News

Brandy Zadrozny:


In summer 2019, a new Facebook user named Carol Smith signed up for the platform, describing herself as a politically conservative mother from Wilmington, North Carolina. Smith’s account indicated an interest in politics, parenting and Christianity and followed a few of her favorite brands, including Fox News and then-President Donald Trump.

Though Smith had never expressed interest in conspiracy theories, in just two days Facebook was recommending she join groups dedicated to QAnon, a sprawling and baseless conspiracy theory and movement that claimed Trump was secretly saving the world from a cabal of pedophiles and Satanists.

Smith didn’t follow the recommended QAnon groups, but whatever algorithm Facebook was using to determine how she should engage with the platform pushed ahead just the same. Within one week, Smith’s feed was full of groups and pages that had violated Facebook’s own rules, including those against hate speech and disinformation.

Smith wasn’t a real person. A researcher employed by Facebook invented the account, along with those of other fictitious “test users” in 2019 and 2020, as part of an experiment in studying the platform’s role in misinforming and polarizing users through its recommendations systems.

That researcher said Smith’s Facebook experience was “a barrage of extreme, conspiratorial, and graphic content.” 


This is going to be a very Facebook-heavy week. Could have filled this edition just with Facebook links. Here are a couple more (besides the ones excerpted below).

Inside Facebook, Jan. 6 violence fueled anger, regret over missed warning signs • Washington Post

Internal alarm, public shrugs: Facebook’s employees dissect its election role • NY Times

Facebook’s internal chat boards show politics often at centre of decision making • WSJ
unique link to this extract

Haugen claims backed by new Facebook whistleblower filing with SEC • The Washington Post

Craig Timberg:


As the company sought to quell the political controversy during a critical period in 2017, Facebook communications official Tucker Bounds allegedly said, according to the affidavit, “It will be a flash in the pan. Some legislators will get pissy. And then in a few weeks they will move onto something else. Meanwhile we are printing money in the basement, and we are fine.”

Bounds, now a vice president of communications, said in a statement to The Post, “Being asked about a purported one-on-one conversation four years ago with a faceless person, with no other sourcing than the empty accusation itself, is a first for me.”

Facebook spokeswoman Erin McPike said in a statement, “This is beneath the Washington Post, which during the last five years competed ferociously with the New York Times over the number of corroborating sources its reporters could find for single anecdotes in deeply reported, intricate stories. It sets a dangerous precedent to hang an entire story on a single source making a wide range of claims without any apparent corroboration.”

…The whistleblower told The Post of an occasion in which Facebook’s Public Policy team, led by former Bush administration official Joel Kaplan, defended a “white list” that exempted Trump-aligned Breitbart News, run then by former White House strategist Stephen K. Bannon, and other select publishers from Facebook’s ordinary rules against spreading false news reports.

When a person in the video conference questioned this policy, Kaplan, the vice president of global policy, responded by saying, “Do you want to start a fight with Steve Bannon?” according to the whistleblower in The Post interview.

Kaplan, who has been criticized by former Facebook employees in previous stories in The Post and other news organizations for allegedly seeking to protect conservative interests, said in a statement to The Post, “No matter how many times these same stories are repurposed and re-told, the facts remain the same. I have consistently pushed for fair treatment of all publishers, irrespective of ideological viewpoint, and advised that analytical and methodological rigour is especially important when it comes to algorithmic changes.”

He added, “There has never been a whitelist that exempts publishers, including Breitbart, from Facebook’s rules against misinformation.”


The Bounds response is not what you’d call a categoric denial. The Facebook statement has no heft at all – whistleblowers are always going to be single sources. The Kaplan response is categoric, but I don’t trust him.
unique link to this extract

For more on how social networks radicalise people (and how much faster it happens than offline), read Social Warming, my latest book – which also has suggestions on how to fix them.

How many users does Facebook have? The company struggles to figure it out • WSJ

Sam Schechner and Jeff Horwitz:


Facebook said in its most recent quarterly securities filings that it estimates 11% of its monthly active users world-wide—which totaled 2.9 billion for its flagship platform in the second quarter—are duplicate accounts, with developing markets accounting for a higher proportion of them than developed ones. On its website for advertisers, Facebook says its estimate for an ad’s audience size depends in part on the number of accounts users have, but it doesn’t quantify the impact.

The internal Facebook research documents on users with multiple accounts—spanning from 2017 through earlier this year—offer a more expansive view into how the company has handled the phenomenon.

Facebook has long called itself a “real identity platform” and bars users from having multiple personal accounts. Unlike Twitter Inc. and other platforms without such rules, the company requires users to have just one master account under a real name. Facebook says the policy helps prevent impersonation and scams.

The memos viewed by the Journal suggest that most instances of multiple accounts stem from users being locked out of their main accounts or making mistakes when signing in, and Facebook says it tries to redirect these users back into their primary accounts. But data cited in the documents show that a quarter to a third of duplicates have been what Facebook calls “persistent SUMA,” where a user continues to operate multiple accounts.

“It’s not a revelation that we study duplicate accounts, and this snapshot of information doesn’t tell the full story,” Facebook spokesman Joe Osborne said, adding that duplicate accounts pose a challenge for many large digital platforms, and that advertisers use Facebook because it gives them desired results.


Facebook is hating these stories. It never quite knows where they’re going to come next, and all it can do is say things aren’t that bad. Makes you wonder when there would ever be something that they said: yes, this is indeed bad.
unique link to this extract

YouTube CEO Susan Wojcicki built a $1 trillion empire by not being Facebook • Business Insider

Hugh Langley and Rob Price:


As if guided by the opposite of Zuckerberg’s famous maxim (“move fast and break things”), Wojcicki’s YouTube tends to move slowly and avoid making waves — at least when it comes to setting the agenda for the rest of the industry.

It wasn’t until more than a month after the polls closed in 2020 that YouTube cracked down on misinformation about the election results. YouTube was also the last of the big social-media companies to ban Trump’s account after the US Capitol riot on January 6.

In September, the company announced a ban on all videos that include misinformation about any vaccines, mimicking a policy Facebook introduced in February. YouTube could have been first: It had consulted with Stanford University and other external bodies over the idea of building an outright vaccine-misinformation policy several years ago, a person familiar with those conversations said, but YouTube demurred. It took a pandemic for the platform to take a stand.

The cautious approach isn’t always unreasonable, given the nature of the issues confronting global platforms like YouTube. Sometimes decisions on high-profile issues don’t fall into neat categories. Spending time to consider the various implications, rather than reacting instantly, is sensible, according to some people inside and outside of YouTube. “A lot of this stuff has never been considered before by society or industry,” said one source.

A YouTube insider described the level of polarized content during the 2020 US elections that the platform needed to deal with as unprecedented. “We were flying blind,” the employee said. “There’s literally no playbook.”


But notice how the platforms, including YouTube, never err on the side of caution about allowing content. They err on the side of caution about *removing* content. Possibly the argument they make to themselves is that this reduces customer complaints. But it also boosts use until they can inch towards a very slow removal.
unique link to this extract

Russia strengthens its internet censorship powers • The New York Times

Adam Satariano and Paul Mozur:


Russia’s boldest moves to censor the internet began in the most mundane of ways — with a series of bureaucratic emails and forms.

The messages, sent by Russia’s powerful internet regulator, demanded technical details — like traffic numbers, equipment specifications and connection speeds — from companies that provide internet and telecommunications services across the country. Then the black boxes arrived.

The telecom companies had no choice but to step aside as government-approved technicians installed the equipment alongside their own computer systems and servers. Sometimes caged behind lock and key, the new gear linked back to a command center in Moscow, giving the authorities startling new powers to block, filter and slow down websites that they did not want the Russian public to see.

The process, underway since 2019, represents the start of perhaps the world’s most ambitious digital censorship effort outside China. Under President Vladimir V. Putin, who once called the internet a “C.I.A. project” and views the web as a threat to his power, the Russian government is attempting to bring the country’s once open and freewheeling internet to heel.


Used against Twitter and, separately, against Putin opponent Alexei Navalny.
unique link to this extract

A look inside Apple’s silicon playbook • Ars Technica

Steven Levy:


[senior hardware technology VP Johny] Srouji explained how his journey at Apple has been one of conscious iteration, building on a strong foundation. A key element of the company’s strategy has been to integrate the functions that used to be distributed among many chips into a single entity—known as SOC, or system-on-a-chip.

“I always fundamentally felt and believed that if you have the right architecture, then you have a chance to build the best chip,” he says. “So we started with the architecture that we believe would scale. And by scaling, we mean scaling to performance and features and the power envelope, whether it’s a watch or iPad or iMac. And then we started selectively figuring the technologies within the chip—we wanted to start owning them one by one. We started with the CPU first. And then we went into the graphics. Then we went into signal processing, display engine, etcetera. Year over year, we built our engineering muscle and wisdom and ability to deliver. And a few years later, when you do all this and you do it right, you find yourself with really good architecture and IP you own and a team behind you that is now capable of repeating that recipe.”

[Senior hardware engineering VP John] Ternus elaborates: “Traditionally, you’ve got one team at one company designing a chip, and they have their own set of priorities and optimizations. And then the product team and another company has to take that chip and make it work in their design. With these MacBook Pros, we started all the way at the beginning—the chip was being designed right when the system was being thought through. For instance, power delivery is important and challenging with these high-performance parts. By working together [early on], the team was able to come up with a solution. And the system team was actually able to influence the shape, aspect ratio, and orientation of the SOC so that it can best nest into the rest of the system components.” (Maybe this helped convince Apple to restore the missing ports that so many had longed for in the previous MacBook.)

Clearly these executives believe the new Macs represent a milestone in Apple’s strategy.


Described like that, you could almost believe that the reason Apple took so long to remedy all the flaws in the MacBook Pros is that it was too busy working on how the new SOC-based systems should integrate everything.
unique link to this extract

British entrepreneur sells company to Twitter • BBC News


British entrepreneur Nick D’Aloisio, who sold the mobile app Summly to Yahoo for $30m (£21.73m) at the age of 17, has sold his latest company to Twitter.

The Sphere group chat app was founded by Mr D’Aloisio and Tomas Halgas. It connects strangers interested in common topics, has been sold for an undisclosed amount and will close in November. Its 20 or so staff will join Twitter to integrate their community features into the social network.

The company started as a question and answer app that allowed users to instantly chat to paid experts. At the end of 2018, almost 500,000 people were using that version of the platform.

However, Mr D’Aloisio said he found himself drawn to the community aspect of the app which brought strangers interested in the same topics together. “What was interesting was that people were talking so often throughout the day, and it wasn’t just talking to their friend on Facebook, but someone they had not met before about something they were interested in,” he told the BBC.

…”[Twitter] are as excited as us to try and figure out this problem around online communities,” said Mr D’Aloisio. “It’s not only a hugely interesting problem, but also a necessity. All groups have the potential to become genuine communities. But most groups suffer from problems in online communication that prevent community-building – things like awkward silences, conversations going off-topic, and vitriol.

“However, we learned over the past two years that a group can transform into a community if its members feel their participation is welcomed.”


I think I briefly met D’Aloisio when he sold Summly; at the time there were lots of doubts about how involved he had really been in its creation. Evidently he’s good at having the right idea just ahead of the right time.
unique link to this extract

Finally, Facebook can say it’s not the most toxic social network • The Guardian

Marina Hyde has Facebook’s PR approach down to a T:


is there a more exciting player in the prebuttal space than Facebook, whose attempts to get out in front of the fortnightly exposés of its behaviour are fast becoming a totally non-ominous part of the early 21st-century powerscape? Of course, many of us have long accepted that when the firm finally causes the apocalypse, the event will be succeeded by a video of Facebook VP Nick Clegg going: “We will do better.” I am now beginning to think the event may even be preceded by a Clegg video announcing: “We will learn from this. Find out from WHAT when the darkness falls next week.”

For now, Facebook is only convincingly troubled by “disinformation” if it’s about itself. We don’t know what will emerge next week, but we can be almost sure how the firm will react to it. The usual MO of Facebook’s chiefs has been to deny they even did the thing they’re being accused of, until the position becomes untenable. At that point, they concede they did whatever it was on a very limited scale, until that position becomes untenable. Next up is accepting the scale was more widespread than initially indicated, but with the caveat that the practice has now come to an end, until that position is the latest to become untenable.

Clear evidence that the practice never came to an end and, in fact, only became more widespread will come with aggressive reminders that it is not and never has been technically illegal. If and when whatever-it-is has been proved to be technically illegal after all, Facebook will accept the drop-in-their-ocean fine, with blanket immunity for all senior officers, and move back to step one in the cycle. We get rinsed; they repeat.


unique link to this extract

US conducts ‘successful’ test of hypersonic missile technology • AFP via Yahoo News


The United States successfully tested hypersonic missile technology, a new weapons system which is already being deployed by China and Russia, the US Navy said Thursday.

The test, conducted Wednesday at a NASA facility in Wallops, Virginia, is a “vital step in the development of a Navy-designed common hypersonic missile,” the navy said in a statement.

“This test demonstrated advanced hypersonic technologies, capabilities, and prototype systems in a realistic operating environment,” it said.

Hypersonic missiles, like traditional ballistic missiles, can fly more than five times the speed of sound (Mach 5).

But they are more maneuverable than their ballistic counterparts and can trace a low trajectory in the atmosphere, making them harder to defend against.


Surprising how quickly this announcement followed the revelation, last week, that China had done the same. A new flavour of Mutually Assured Destruction: never to be used, always to be held as a threat.
unique link to this extract

Exclusive: governments turn tables on ransomware gang REvil by pushing it offline • Reuters

Joseph Menn and Christopher Bing:


According to three people familiar with the matter, law enforcement and intelligence cyber specialists were able to hack REvil’s computer network infrastructure, obtaining control of at least some of their servers.

After websites that the hacker group used to conduct business went offline in July, the main spokesman for the group, who calls himself “Unknown,” vanished from the internet.

When gang member 0_neday and others restored those websites from a backup last month, he unknowingly restarted some internal systems that were already controlled by law enforcement.

“The REvil ransomware gang restored the infrastructure from the backups under the assumption that they had not been compromised,” said Oleg Skulkin, deputy head of the forensics lab at the Russian-led security company Group-IB. “Ironically, the gang’s own favorite tactic of compromising the backups was turned against them.”

Reliable backups are one of the most important defences against ransomware attacks, but they must be kept unconnected from the main networks or they too can be encrypted by extortionists such as REvil.


Fabulous hoisting with own petard for a ransomware group to get caught through compromised servers and backups.
unique link to this extract

Conti Statement 10.22.2021 •

A statement from one of the eastern European ransomware gangs about the US attacking the REvil ransomware gang:


As a team, we always look at the work of our colleagues in the art of pen-testing, corporate data security, information systems, and network security. We rejoice at their successes and support them in their hardships.

Therefore, we would like to comment on yesterday’s important announcement by the US law enforcement about the attack on the REvil group.

We want to remark the following:

First, an attack against some servers, which the US security attributes to REvil, is another reminder of what we all know: the unilateral, extraterritorial, and bandit-mugging behavior of the United States in world affairs.

However, the fact that it became a norm does not presume that it should be treated like one. Unlike our dearest journalist friends from the Twitter brothel, who will sell their own mother for a bone from bankers or politicians, we have the guts to name things as they are. We have a conscience, as well as anonymity, while our skills allow us to say something that many “allied” governments are afraid of saying:

With all the endless talks in your media about “ransomware-is-bad,” we would like to point out the biggest ransomware group of all time: your Federal Government. There is no glory in this REvil attack. First, because REvil has been dead in any case, but secondly, because the United States government acted as a simple street mugger while kicking a dead body.

Let’s break it down point by point. There was an extraterritorial attack against some infrastructure in some countries.


Simply the funniest thing you’ll read all week. (It’s hard even to be sure if they know it’s ironic.)
unique link to this extract

Errata, corrigenda and ai no corrida: Ian C points out that, the site shown last week which lets you (fairly) seamlessly edit pictures online, limits the size of the retouched pictures that you download to about 150KB – OK for a phone, but not to print. (The site’s owner is probably trying to limit bandwidth costs.) Looks like you’ll need that pricey image editor after all.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

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