Dear Tumblr: Banning “Adult Content” Will Harm Sex-Positive Communities

Social media platform Tumblr has announced a ban on so-called “adult content,” a move made, it seems, in reaction to Tumblr’s app being removed from the Apple app store. But while making the app more available is in theory good for Tumblr users, in practice what’s about to happen is mass censorship of communities that have made Tumblr a positive experience for so many people in the first place.

On December 3, Tumblr CEO Jeff D’Onofrio posted a lengthy missive about a new policy, titled, apparently unironically, “A better, more positive Tumblr.” Instead of laying out a vision that is better and positive, D’Onofrio’s post lays bare the problems with the ban on so-called “adult content.” First of all, the policy is confusing and broad, leaving users in the lurch about what they can and can’t do on Tumblr. Second, according to D’Onofrio, enforcement of the policy will be reliant on automated tools, the use of which is—and always has been—rife with problems. Third, the people who will end up punished aren’t pornbots or sex traffickers but already-marginalized groups who have built sex- and body-positive communities on Tumblr. And finally, all of these things come together to show just how many ways platforms and tech companies can get in between users and their freedom of expression.

In D’Onofrio’s post, he explains that “in order to continue to fulfill [Tumblr’s] promise and place in culture, especially as it evolves, we must change,” going on to say that as part of that evolution, “adult content” will no longer be allowed on the platform. He further explains:

“We recognize Tumblr is also a place to speak freely about topics like art, sex positivity, your relationships, your sexuality, and your personal journey. We want to make sure that we continue to foster this type of diversity of expression in the community, so our new policy strives to strike a balance.”

On the face of it, this is literally contradictory. Saying adult content is banned, but that “diversity of expression” related to all those listed topics isn’t is impossible to parse for the average user. Tumblr’s FAQ “clarifying” the definition of adult content (that is, that which includes “photos, videos, or GIFs that show real-life human genitals or female-presenting nipples, and any content—including photos, videos, GIFs and illustrations—that depicts sex acts”) likewise compounds this problem.

The new policies rule out almost all forms of nudity. “Female-presenting nipples” in particular is a phrase that has come under ridicule, because, among other things, it polices bodies for what they look like, based on a specific conception of gender, and a body part that only some cultures—but certainly not all!—prohibit showing in public.

On the other hand, the very next question has Tumblr claiming that “female-presenting nipples” can be shown in some contexts, that written erotica, “political” nudity, and “art” are permitted. These are all subjective categories that leave a lot of people on uncertain ground. Just look at Facebook, which has similar rules regarding nudity. In the past few years, we’ve seen Copenhagen’s Little Mermaid statue, a famous illustration of a woman licking an ice cream cone, a classic French painting, and a 16th-century statue of the Roman god Neptune taken down by Facebook’s content moderators under the restrictive policy.

Tumblr has also decided that the way to make these subjective calls about what is “art” and what is “adult content” is by using automated tools. D’Onofrio basically admits that these tools don’t work properly, saying in his post that “We’re relying on automated tools to identify adult content and humans to help train and keep our systems in check. We know there will be mistakes”.

That is an understatement. Filters don’t work. We’ve seen this in the copyright context many times. For example, YouTube’s Content ID system works by checking newly uploaded material against a database of copyrighted material and notifying copyright holders if there’s a match. And it resulted in five copyright claims being filed against a video of white noise. Five people claimed they literally owned exclusive rights to static.

And that’s just when it comes to checking for copyrighted material. It’s rather brazen of Tumblr to suggest it has the ability to develop and train a program to determine if something is porn; after all, there is literally a famous Supreme Court quote about the difficulty of defining obscenity! And so far, as any informed observer would have predicted, Tumblr’s system is failing miserably. Among the items flagged are a picture of Pomeranian puppies, selfies of fully-clothed individuals, images of raw chicken, and much much more. And, despite D’Onofrio’s statement that art, discussion of sexuality, and politics wouldn’t violate the terms, all of those categories have been hit.

When we look to groups outside the dominant culture, the problem is especially pernicious. Already, an image of a video game character on a pride flag, a selfie with the word “lesbian,” and someone talking about a family death due to AIDS have all been flagged. Tumblr may think it’s creating a “better” community, but it’s destroying what made it great in the first place.

In his post, D’Onofrio defends the policy by saying that the bottom line is that “there are no shortage of sites on the internet that feature adult content.” Indeed, the Internet is full of porn, the overwhelming majority of which caters to heterosexual men. But on Tumblr, people created sex-positive spaces on Tumblr that don’t exist elsewhere. People created portfolios of their work, all of it, on the platform. Those spaces are going to vanish.

A business decision?

Three paragraphs into his better, more positive manifesto, D’Onofrio states “posting anything that is harmful to minors, including child pornography, is abhorrent and has no place in our community. We’ve always had and always will have a zero tolerance policy for this type of content” and asks that no one confuse that with this new policy. Child exploitation imagery is both vile and illegal, and the fact that Tumblr apparently wasn’t eliminating it shows that it needed to hire people to enforce its existing policy, not outsource the job to algorithms. So why create this new, wholesale ban?

It’s impossible to divorce the new policy from the fact that, just a month prior to the announcement, Tumblr disappeared from the Apple App Store. And that, when asked about it, a Tumblr spokesperson responded with nearly the same words that D’Onofrio also used in his post.

Apple’s App Store has long acted as censor and gatekeeper to the Internet. In 2010, Steve Jobs once said that the iPad offered “freedom from porn” and that there was a “moral responsibility to keep porn off the iPhone.” Apple has consistently enforced draconian rules for app developers, exerting control over how its users get to experience the Internet. The company’s rules have even had the effect of silencing the press, as in 2010 when a large-scale removal of apps containing nudity impacted several mainstream German news publications.

We don’t know if Apple is the sole reason for these new rules. Tumblr also got banned this year in Indonesia because of pornography, for example, and may just want to make itself as non-controversial as possible. And it’s notable that Tumblr’s new policy is largely in line with that of peers Facebook, Microsoft, and YouTube, all of which heavily restrict so-called “adult content.”

The end result, though, is that companies and governments are changing how users get to express themselves on the Internet. The multi-billion dollar corporate porn industry won’t go away; rather, what will are places for people to talk frankly, openly, and safely about sex and sexuality. Groups that are pushed out of mainstream discussions or find themselves attacked in mainstream spaces are once again losing their voices.

O2 outage: 31m mobile customers unable to get online

More than 30 million mobile customers of providers including O2, Tesco Mobile and Sky Mobile have been unable to get online or use 3G and 4G services after a technical fault caused a UK-wide outage.

The outage hit O2 and also affected companies that use its network including its subsidiary Giffgaff and Lycamobile. It began at about 4am on Thursday, according to the website DownDetector.

O2, which has about 25 million UK customers, said that voice calls were still working but advised people to seek out wifi if they needed to get online while the outage persisted.

“We’d encourage our customers to use wifi wherever they can and we apologise for the inconvenience caused,” said an O2 spokesman.

TfL Bus Alerts (@TfLBusAlerts)

Countdown Systems – there are reports that live bus updates to our Countdown Systems has gone down. The company that updates the data to our network of Countdown Systems is currently experiencing difficulties in providing this service. This issue is being investigated.

December 6, 2018

The company said that the problem stemed from a global software issue at a third-party supplier, understood to be Ericsson, which has also affected other mobile operators around the world.

“We’re aware that our customers are unable to use data this morning,” said a spokesman for O2, which is owned by the Spanish telecoms giant Telefonica.

“One of our third-party suppliers has identified a global software issue in their system which has impacted us. We believe other mobile operators around the world are also affected. Our technical teams are working with their teams to ensure this is fixed as quickly as possible.”

Sign up to the daily Business Today email or follow Guardian Business on Twitter at @BusinessDesk

In total, about 31 million mobile phone users have been affected as a number of third-party mobile providers that rely on the O2 network, including Tesco Mobile, Sky Mobile, Lycamobile and O2’s Giffgaff have also been affected.

The outage has also affected services such as Transport for London’s live updates of bus arrival times at stops across the capital, which relies on O2’s network for data updates.

In October, O2 apologised for a major network outage affecting customers across the UK. The company said at the time that the issue was fixed within 40 minutes.

O2’s rivals including Vodafone, Three and EE have also suffered nationwide outages affecting customers in the past year.

In 2012, O2 was at the centre of a major network outage when customers were not able to access services for 25 hours.

ARM1 Gate-level Simulation

Pics Chip geeks have produced an interactive blueprint of the ARM1 – the granddaddy of the processor cores powering billions of gadgets today, from Apple iPhones to Raspberry Pis, cameras, routers and Android tablets.

The peeps behind the fascinating blog normally reverse-engineer chips by pulling the silicon out of their plastic packages, photographing the circuitry using ultra-high-resolution imaging systems, and visualizing the electronics after painstakingly analyzing the semiconductor gate construction – here’s the 6502 CPU from the BBC Micro, Commodore 64, Nintendo Entertainment System, Apple IIe, and many other computers.

In the case of the ARM1, to celebrate the 25th anniversary of Brit chip architects ARM, the team have managed to lay their hands on the original designs of the 32-bit RISC processor core, and visualized it for the web: you can single-step, or race through, instructions, and watch how the ARM1’s transistor gates operate as code and data is clocked through the circuitry in your browser. You can also zoom in and out of the blueprints, and rove around the layout, exploring the core’s subsystems.

The ARM1 is a milestone in British engineering: in late 1983, Steve Furber and Sophie Wilson – two engineers based at Acorn Computers in Cambridge, UK – visited and observed chip designers in America, and decided processor design wasn’t as scary as they first imagined. So they set about formally sketching out a 32-bit RISC microprocessor to power future Acorn microcomputers, at one point modeling the instruction set in BBC BASIC.

Eventually, about 18 months later, they produced the ARM1 – a tiny, low-power, efficient CPU fabricated by VLSI Technology with roughly 25,000 transistors using a 3,000nm (3μm) process. Today, a quad-core Intel Skylake Core i7 processor, with builtin GPU, has 1,350,000,000 transistors using a 14nm process.

Youtube video

The ARM1 was essentially a prototype, and was quickly followed by the ARM2, which was used to launch Acorn’s Archimedes family of computers. The ARM2 had pretty much the same instruction set as the ARM1, although featured new multiplication and (later) atomic swap instructions. Next came the ARM3 with an integrated data and instruction cache that really made programs run like butter off a hot knife.

Then we jumped to the ARM6 (an ARMv3 chip) that moved the architecture to a full 32-bit address space, and the rest is history: today the ARMv7 and ARMv8 architectures in Cortex-A cores power the world’s smartphones, tablets and other devices. Their Cortex-M cousins are used as microcontrollers, and Cortex-R cores are used in real-time systems, such as hard drive controllers.

More or less, the 32-bit RISC ARM instruction set hasn’t changed too much since the mid-1980s; a more compact instruction set called Thumb was introduced by ARM for handhelds and small embedded gizmos, and the 64-bit ARMv8 instruction set, launched in 2011, resembles MIPS64.

One very nice thing about the 32-bit instruction set is its pervasive conditional execution, which helps one avoid branching over code. For example, this sequence of instructions resets the register r0 to 0 if its value is equal to or less than zero, or forces its value to 1 if its value is greater than zero:

 CMP r0, #0 ; if (r0 <= 0) MOVLE r0, #0 ; r0 = 0; MOVGT r0, #1 ; else r0 = 1

Without the conditional moves (MOVLE and MOVGT) after the compare (CMP), you’d have to branch after the compare, which is wasteful.

Die hard … the ARM1’s silicon in all its glory (Credit:

ARM was spun off from Acorn in 1990 before the latter imploded in the late 1990s; today, ARM is a £250m-a-year business with 3,300 employees licensing processor core designs to chipmakers worldwide. The first proper operating system developed for ARM, Acorn’s RISC OS, continues as a community-run project, powered by modern 32-bit ARM cores.

Close up … the semiconductor layout for the ARM1

Here’s what the team have to say about the visualized ARM1:

Designed by Sophie Wilson and Steve Furber before there was an ARM Ltd., the Acorn RISC Machine was the first of a line of processors that power our cell phones and tablets today. Unlike our projects based on microscope images, the Visual ARM was created from a resurrected .cif chip layout file, used under our license agreement with ARM. We also photographed one of the few ARM1 chips at very high resolution, and our photograph is featured at the Centre for Computing History in Cambridge.

Credit goes to ARM founding engineers John Biggs for inspiring the project, discovering the tape, and recovering a usable .cif file, Lee Smith for spotting the variable record format used to encode the file (an artifact of the VMS on Acorn’s VAX that at first appeared to be widespread corruption of the file), to Cambridge University Computing Services for reading the Exabyte tape, and to ARM founder Dave Howard for help unraveling the VLSI CIF dialect. Our chip simulation and visualization was developed by Barry Silverman, Brian Silverman, Ed Spittles, and Greg James.

Evaluating options … the prototype ARM1 as a BBC Micro coprocessor (Source)

You can read more about Sophie Wilson and Steve Furber’s contribution to Acorn, ARM, and computing worldwide here and here. Today, Furber is an ICL Professor of Computer Engineering, working on neural systems, and Wilson is a top Broadcom engineer. ®

Many thanks to Computerphile for the video interview with Prof Furber.

Sponsored: Five steps to dealing with the insider threat

Commandeering Australian citizens to become spies

Welkom thuis!

Je besteedt het grootste gedeelte van je tijd op deze tijdlijn. Je krijgt hier direct updates over zaken die belangrijk voor je zijn.

Werken deze Tweets niet voor je?

Houd je muis boven de profielfoto en klik op de knop Volgend om een account te ontvolgen.

Verspreid het nieuws

De snelste manier om de Tweet van een ander te delen met je volgers is met een Retweet. Tik op het pictogram om hem meteen te versturen.

Praat mee

Stuur een antwoord om te laten weten wat je van een Tweet vindt. Zoek een onderwerp waarin je geïnteresseerd bent en doe meteen mee met het gesprek.

Vind meer van wat je leuk vindt

Volg meer accounts om meteen updates te krijgen over onderwerpen die je belangrijk vindt.

Mis geen enkel Moment meer

Blijf op de hoogte van de beste verhalen terwijl ze worden verteld.

Serial Port SDR

To say that the RTL-SDR project was revolutionary might be something of an understatement. Taking a cheap little USB gadget and exploring the radio spectrum from the tens of megahertz up to into gigahertz frequencies with the addition of nothing more than some open source tools may go down as one of the greatest hacks of the decade. But even in the era of RTL-SDR, what [Ted Yapo] has manged to pull off is still pretty incredible.

With a Python script, a length of wire attached to the TX pin, and a mastery of the electron that we mere mortals can only hope to achieve, [Ted] has demonstrated using a common USB to serial adapter as a SDR transmitter. That’s right, using the cheap little UART adapter you’ve almost certainly got sitting in your parts bin right now and his software, you can transmit in the low megahertz frequencies and even up into VHF with some trickery. The project is still very much experimental, and though this may be the first time, we’re willing to bet this isn’t the last time you’ll be hearing about it.

The basic idea is that when sending certain characters over the UART serial line, they can combine with the start and stop bits to produce a square wave burst at half the baud rate. [Ted] found that sending a string of 0x55 at 19200 baud would generate a continuous square wave at 9600 Hz, and if he turned the baud rate all the way up to 2,000,000 where these USB adapters top out, that signal was transmitted at 1 MHz, right in the middle of the AM dial.

A neat trick to be sure, but alone not terribly useful. The next step was to modulate that signal by sending different characters over UART. [Ted] explains at great length his experiments with multi-level quantization and delta-sigma schemes, and each step of the way shows the improvement of the transmitted audio signal. Ultimately he comes up with a modulation scheme that produces a impressively clean signal, all things considered.

This alone is impressive, but [Ted] isn’t done yet. He realized that this method of transmission was generating some strong frequency harmonics which extended far beyond the theoretical maximum 1 MHz frequency of his UART SDR. In his experimentation he found he was able to pick up a signal from all the way out to 151 MHz, though it was too poor to be of any practical use. Dialing back the expectations a bit, he was able to successfully control a cheap 27 MHz RC toy using the 43rd harmonic of a 631 kHz signal at a range of about 10 feet with a FT232RL adapter, which he notes produces the cleanest signals in his testing.

[Ted] is still working on making transmissions cleaner and stronger by adding filters and amplifiers, but these early accomplishments are already very promising. His work reminds us of a low frequency version of the USB to VGA adapter turned GHz SDR transmitter, and we’re very eager to see where it goes from here.

Protection Against Reconstruction and Applications in Private Federated Learning

Abstract: Federated learning has become an exciting direction for both research and practical training of models with user data. Although data remains decentralized in federated learning, it is common to assume that the model updates are sent in the clear from the devices to the server. Differential privacy has been proposed as a way to ensure the model remains private, but this does not address the issue that model updates can be seen on the server, and lead to leakage of user data. Local differential privacy is one of the strongest forms of privacy protection so that each individual’s data is privatized. However, local differential privacy, as it is traditionally used, may prove to be too stringent of a privacy condition in many high dimensional problems, such as in distributed model fitting. We propose a new paradigm for local differential privacy by providing protections against certain adversaries. Specifically, we ensure that adversaries with limited prior information cannot reconstruct, with high probability, the original data within some prescribed tolerance. This interpretation allows us to consider larger privacy parameters. We then design (optimal) DP mechanisms in this large privacy parameter regime. In this work, we combine local privacy protections along with central differential privacy to present a practical approach to do model training privately. Further, we show that these privacy restrictions maintain utility in image classification and language models that is comparable to federated learning without these privacy restrictions.

Submission history

From: Ryan Rogers [view email]
[v1] Mon, 3 Dec 2018 18:59:16 UTC (133 KB)

The Swede created a $400,000 Indiegogo-scam

Two years ago, Breakit wrote about a successful Indiegogo campaign called Bioring.

The people behind the campaign had managed to attract thousands of individuals from around the world to invest in a “smart ring” that the company supposedly had developed. The campaign raised around $400,000.

At the time, Breakit interviewed one of the people behind the campaign over the phone, a Swedish man who called himself Michael Johnson.

We did not know at the time that the founders of Bioring soon thereafter would disappear with the money without delivering any products.

Similarly, it later turned out that the PR firms they had hired to spread the campaign were never paid.

Instead, Bioring sent them a message over Skype, stating that the PR firms would never be able to track down the person behind the campaign, who now called himself “Big Mike”, and that the money was in an offshore account. Reported at the time by the site Vocative.

The story could have ended there. But Breakit has, with the help of a range of different sources and formal documents, successfully been able to connect the people behind Bioring to another campaign on Kickstarter and Indiegogo, that raised over a million dollar this year.

The new potentially fraudulent campaign has purportedly created a “smart blanket” called Zen Blanket, which promises better sleep for those who use it by lowering stress hormones.

We were given a new lead

When we interviewed “Michael Johnson” in 2016 he sent us a photo of himself. Now, two years later, a reader got in touch to tell us who the man on the photo really was. His name was something else, and he was a former student at the prestigious Stockholm School of Economics, SSE.

Our initial thoughts were that “Michael Johnson” simply had stolen the image he sent to us in 2016. But it soon turned out that there were a number of clues which led us to tie the person on the photo — whose name we have decided to not publish — to Bioring.

The American freelance journalist John Lewis from Dallas is part of a network that works to reveal scams on mainly Indiegogo and Kickstarter.

He tells us that one person, with the same name as the former SSE student, has worked hard to erase his association to Bioring ever since the campaign vanished into thin air.

According to John Lewis, the student contacted several individuals who had criticised or described the campaign as a fraud online. He asked them to delete his name, more or less menacingly.

The person in question has not had a taxed income in Sweden for the past five years. He has not been penniless, however. Documents that we have acquired access to show that a large amount of money, with connection to him, has been transferred to Sweden.  

The Connection to Zen Blanket

What is this person’s connection to Zen Blanket, then?

The registered company behind the successful Zen Blanket campaign, which was launched on Kickstarter in May, is a startup from Gothenburg, Sweden. The campaign was moved to Indiegogo during the summer and raised in total $1.2 million from approximately 5,000 backers from around the world.

The blankets were supposed to have been delivered in August. When they were not delivered as promised, several backers became suspicious: many have criticised the campaign and a few have threatened to file police reports.

Zen Blanket proclaims in a post on their Kickstarter page that the delays are due to the manufacturer not being able to handle the size of the order, and that they are working on a solution. Any reimbursements have not been mentioned.

In a document from the Swedish Companies Registration Office, we have discovered that the SSE student is listed as the founder of the Swedish trading company behind the Zen Blanket campaign, although he is no longer a representative of the company. He is also listed as the founder of two American companies, one in Delaware and the other in Wyoming, which are the co-owners of the aforementioned Swedish trading company.

Furthermore, the student has set up an unrelated company in Sweden with the person listed as the creator of the initial Zen Blanket campaign on Kickstarter.

Breakit finds an address

For the past two years, the SSE guy has had his residency registered in a Stockholm suburb. Since we have not been able to get in touch with him in any other way, we decided to go there and knock on his door. When he opened, we immediately recognised the person from the photo that we had received a couple of years ago.

So what did he say about our accusations?

He claimed that he did not know what we were talking about. And he was not able to explain how his signature had ended up on the documents at the Companies Registration Office.

“That is not my phone number, not my email address and not my signature. And that someone would have sent you my photo, that’s just really strange”, he said when we showed him the material.

He also declared that he had never heard about Bioring or Zen Blanket.

This is how far we get.

Who is guilty? And is it even a fraud?

We will leave it up to others to decide whether the Bioring case was a fraud or just a failure. It is impossible to know, for sure.

It is equally difficult to say with certainty that Zen Blanket is a fraud, perhaps the people behind the campaign actually aim to produce the blankets. They stated that they will do so during the second half of 2019.

That the former student had his identity stolen for years, by people who have managed to create a Swedish company in his name, and used his image, is similarly a possibility in theory.

How, and the extent to which, he is involved in this is impossible for us to say.

Both the Police Finance Department and the Swedish Economic Crime Authority refer to the preliminary investigation confidentiality and have neither been able to confirm, nor retract, if they know about the case and if they are investigating it.

According to Stefan Lundberg, Chief Prosecutor at the Swedish Economic Crime Authority, this type of crime is difficult to control.

“The culprit is often in one country, and those investing money in another. They might even be in several countries. That causes problems for the different countries’ legal systems. Which country should take on the case? It is part of the crime plan: to make it difficult for the authorities”, he says.

What does Indiegogo say?

Since Breakit contacted Indiegogo and Kickstarter a couple of weeks ago with our information about the connections between Bioring and Zen Banket, Indiegogo has started to look into the campaign. Kickstarter has still not replied.

Breakit has also spoken to a person in Gothenburg who bought two blankets. He decided to go and see for himself whether the company really existed on the given address. He found a number of companies and offices, but no Zen Blanket.

During the course of our investigation, Zen Blanket has changed its hometown to Stockholm on the campaign page.

Translation: Amanda Ekström

Facebook’s Very Bad Month Just Got Worse

Who could have imagined that a creepy little app that scoured Facebook for pictures of women in bikinis might be the instrument that skewers the behemoth social network? Who, that is, besides Facebook executives and their lawyers? Until a little over a week ago, the company had successfully sequestered internal e-mails, which were obtained by the legal team of Ted Kramer, the founder of the app company Six4Three, during the discovery process in a 2015 lawsuit. At issue was Facebook’s policy of allowing third-party app developers to access the data of Facebook users’ friends—the very policy that enabled Cambridge Analytica to buy the data of eighty-seven million unwitting users on behalf of the Trump campaign. In Kramer’s case, his Pikinis app relied on that access; once Facebook changed its policy, in 2014, the app no longer worked. Kramer cried foul and sued Facebook for breach of contract. At the company’s request, the judge in the case ordered the records sealed to keep them, ironically, private.

That changed on November 20th, when a parliamentary sergeant-at-arms showed up at Kramer’s London hotel room and escorted him to the halls of Westminster, where the Tory M.P. Damian Collins, the chairman of the Digital, Culture, Media and Sport Committee, demanded that Kramer turn over all his files in the case under threat of arrest. Kramer, who was in London on unrelated business, said that he panicked and moved a number of files from a Dropbox account to a USB drive, which he turned over to the M.P. How Collins knew that Kramer was in London has been traced to the tireless Guardian reporter Carole Cadwalladr, who has spent more than two years unravelling the connections between Cambridge Analytica, Facebook, Donald Trump, Steve Bannon, Robert Mercer, and Brexit. After arranging to meet with Kramer, she appears to have tipped off Collins to Kramer’s whereabouts. But how Collins knew that Kramer was in possession of the documents—which were not his to have—remains a mystery. There are reports that the two men had been in communication for the past few months, at Cadwalladr’s prompting, which suggests that the frog march to Parliament was just for show. (When Kramer’s lawyers found out that he was in possession of the discovery documents and that he had given them to Collins, they attempted to drop him, but, over the weekend, a California judge demanded that they continue to represent Kramer until the matter of the purloined files is adjudicated.)

Collins, for his part, claimed that whatever seal was in force in California was irrelevant in the U.K. On November 27th, he used the snatched e-mails to grill Richard Allan, the Facebook executive who was sent to speak for the company at an unprecedented international hearing on fake news and disinformation that Collins convened in Parliament. Mark Zuckerberg, who said he was too busy to appear—making this the fourth time that he has refused a request from Parliament—was represented by a nameplate positioned in front of an empty chair; every time the television cameras panned to Allan, they also showed his absent boss. Allan admitted that the optics were “not great,” and then went on to do his best Zuckerberg imitation, dodging and feinting as the lawmakers sought to put him on the spot. The most damning claim to emerge was that Facebook either gave, or considered giving, favored access to its users’ data to companies that spent at least two hundred and fifty thousand dollars in advertising on the platform. There also seemed to be evidence that Facebook had been informed, in 2014, that computers with Russian I.P. addresses were siphoning three billion data points a day from Facebook accounts, a claim that Allan refused to address in the hearing.

On Wednesday, Collins published the full cache that he seized from Kramer. The two hundred and fifty pages of internal Facebook documents show, irrefutably, that the company did, indeed, whitelist a number of lucrative business partners, including Netflix, Lyft, and Airbnb, allowing them continued and unfettered access to the accounts of Facebook users and their friends after the company claimed that it had stopped the practice. The documents also reveal that, in 2015, a permissions update for Android devices, which users were required to accept, included a feature that continuously uploaded text messages and call logs to Facebook.

The fallout from Ted Kramer’s London misadventure was just one debacle of many during Facebook’s terrible, horrible, no good, very bad month. Back at Facebook headquarters, in Menlo Park, Zuckerberg was quite busy after all, doing damage control after a scathing New York Times exposé. “Delay, Deny and Deflect: How Facebook’s Leaders Fought Through Crisis,” which was published on November 14th, created one of the biggest crises in the company’s history. As if to prove the reporters’ claims, Zuckerberg publicly denied their veracity. But the Times piece was deeply sourced and contained the disturbing revelation that the company had hired a right-wing opposition research group, Definers Public Affairs, to dig up dirt on George Soros, after Soros gave a blistering speech at the World Economic Forum in Davos decrying the power of social media, especially Facebook and Google, for their “far-reaching adverse consequences on the functioning of democracy, particularly on the integrity of elections.”

Definers Public Affairs had previously pushed out misinformation about the Apple C.E.O., Tim Cook, on behalf of the tech company Qualcomm, which wanted to undermine Cook’s relationship with the Trump Administration. For Facebook, Definers launched a P.R. campaign—propagated through the conservative and alt-right media ecospheres—that traded on anti-Semitic innuendo to suggest that Soros was the deep pockets behind a grassroots anti-Facebook group. News of Facebook’s campaign against Soros came out three weeks after Cesar Sayoc, a fervent Trump supporter, sent a pipe bomb to Soros—and also to the Clintons, the billionaire Tom Steyer, and other prominent Democrats.

And then it got worse. On November 15th, shortly after the Times exposé was published, Facebook claimed that “The New York Times is wrong to suggest that we ever asked Definers to pay for or write articles on Facebook’s behalf or to spread misinformation.” The next day, Facebook’s chief operating officer, Sheryl Sandberg, who is in charge of the company’s communications team, said, somewhat intricately, “I did not know about or hire Definers or any firm. . . . Definers was hired, we have lots of firms. They were hired not to smear anyone.” But then, the night before Thanksgiving, in a Facebook blog post that the company perhaps thought would be lost in the holiday shuffle, Sandberg allowed that “I did receive a small number of e-mails where Definers was referenced.” A week later, after another damning piece in the Times, a company spokesperson admitted that it was Sandberg herself who had initiated an investigation into whether Soros had a financial interest in criticizing Facebook—that is, if the veteran financier had engaged in what is called “short and distort,” trash talking a company after betting that its stock will lose value. Not surprisingly, he hadn’t.

Soros’s complaints about Facebook were hardly original: former Facebook employees like Chamath Palihapitiya, who served as its vice-president for growth, and investors like Roger McNamee have made similar criticisms. Nor was Soros being speculative. At the time of his Davos talk, the company’s role in enabling Russian operatives to exploit social divisions in the United States through Facebook ads and user pages, and helping the Trump campaign suppress the African-American vote, were well known. Indeed, in a two-part “Frontline” documentary, “The Facebook Dilemma,” which was released at the end of October, a collection of former Facebook employees expressed concerns about the unwillingness of Zuckerberg and Sandberg to take meaningful responsibility for Facebook’s immense power and capacity to do harm. “My concerns, at the time, were that I knew that there were all these malicious actors who would do a wide range of bad things given the opportunity, given the ability to target people based on this information that Facebook had,” Sandy Parakilas, who had been the company’s platform operations manager in 2011, told the filmmakers. “So I started thinking through what are the worst-case scenarios of what people could do with this data . . . and I shared that with a number of people, both people in privacy and some senior executives. And the response was muted, I would say . . . I got the sense that this just wasn’t their priority. They weren’t that concerned about the vulnerabilities that the company was creating.”

Those vulnerabilities were on full display on November 5th, the day before the midterm elections. In an effort to prevent the kinds of malign interferences that had dogged the Presidential election two years earlier, Facebook had set up what it was calling a “war room” to monitor its network. The company had already removed thirty Facebook pages, which it believed were designed by Russians to interfere with the elections. The war room was both a public declaration that Facebook was taking fake news and propaganda seriously and an admission that the platform remained susceptible to manipulation. “This is going to be a constant arms race,” Katie Harbath, Facebook’s global politics and government outreach director, told the Verge. “This is our new normal. Bad actors are going to get more sophisticated in what they’re doing, and we’re going to have to get more sophisticated in trying to catch them.” Since elections happen around the world throughout the year, the company planned to keep the war room open “for the foreseeable future.”

That same day, Business for Social Responsibility, a nonprofit organization, issued a report that Facebook had commissioned to address concerns about the platform’s role in the Rohingya genocide in Myanmar. While the Facebook platform has been used in other countries to cultivate ethnic and sectarian violence—Nigeria, Germany, Egypt, India, and Sri Lanka among them—its co-optation by the Myanmar army was especially heinous. Adopting the playbook of Russian operatives, the army created hundreds of seemingly innocuous Facebook accounts about celebrities, entertainment, and beauty, which it used to plant incendiary narratives about the country’s Muslim minority. In its response to the B.S.R. report, Alex Warofka, Facebook’s product policy manager, said that the company had already implemented a number of its recommendations but that “there is more to do.”

Shortly after the Myanmar report was issued, Facebook was compelled to take down an inflammatory, racist video ad, paid for by the Trump campaign, on the grounds that the ad “violates Facebook’s advertising policy against sensational content.” The spot, which targeted voters in Florida and Arizona, blamed Democrats for an undocumented immigrant who had twice been deported before killing two sheriff’s deputies, in Sacramento, in 2014. Facebook’s policy disallows ads that are “dehumanizing or denigrating [to an] entire group of people, and using frightening and exaggerated rumors of danger.” The company removed the ad only after it had already been seen on Facebook by as many as five million people. And here’s the rub: while the ad “cannot receive paid distribution,” the company declared, “the video is allowed to be posted on Facebook.” So there you have it, the fecklessness of Facebook in a single sentence.

November has passed, but Facebook’s troubles have not. A departing African-American Facebook employee charged, in a Facebook post, that the company was “failing its black employees and its black users”; Facebook took down his post. Wired reported that the company was abandoning charities—which it had lured to the platform—once they’d been hacked. The election war room has been dismantled, leading some to suggest that it was a publicity stunt. Definers Public Affairs is no longer working for Facebook, but the damage to Sheryl Sandberg’s credibility and carefully coiffed public persona may prove irreversible—as Michelle Obama said the other day, “it’s not always enough to lean in, because that shit doesn’t work all the time.” And that creepy Pikinis app, courtesy of Damian Collins, has now supplied a portrait of naked greed and corporate dissembling. That empty seat in Parliament, where Mark Zuckerberg chose not to sit to answer the concerns of the global community, has suddenly got much hotter.

China backs bold plan to tear down journal paywalls


Officials pledge support for European-led ‘Plan S’ to make research papers immediately free to read — but it’s unclear whether China will adopt its policies.

Search for this author in:

The audience at OpenAccess 2020 in the Harnack House of the Max Planck Society in Berlin

In a huge boost to the open-access movement, librarians and funders in China have said that they intend to make results of publicly funded research free to read immediately on publication.

The move, announced at an open-access meeting this week in Berlin, includes a pledge of support for Plan S, a bold initiative launched in September by a group of European funders to ensure that, by 2020, their scientists make papers immediately open.

It is not yet clear when Chinese organizations will begin implementing new policies, or whether they will exactly adopt Plan S’s details, but Robert-Jan Smits, the chief architect of Plan S, says the new stance is a ringing endorsement for his initiative. “This is a crucial step forward for the global open access movement,” he says. “We knew China was reflecting to join us — but that it would join as so soon and unambiguously is an enormous surprise.”

In three position papers, seen by Nature, China’s National Science Library (NSL), its National Science and Technology Library (NSTL) and its Natural Science Foundation, a major research funder, all said that they “support the request of the OA2020 initiative and Plan S to transform, as soon as possible, research papers from publicly funded projects into immediate open access after publication, and we support a wide range of flexible and inclusive measures to achieve this goal.”

“… We demand that publishers should not increase their subscription prices on the grounds of the transformation from subscription journals to open access publishing,” the papers say.

The government will now urge Chinese funders, research organizations and academic libraries to make the outcome of publicly funded research free to reach and share as soon as possible, says Xiaolin Zhang, chair of the Strategic Planning Committee of the National Science and Technology Library (NSTL) at the Ministry of Science and Technology in Beijing. He told the meeting that the NSFC, NSTL and NSL will all support the government’s request to make research papers open immediately after publishing, and that implementation policies should follow soon. He says he expects that funders will now push all researchers in China to follow suit.

Zhang also told the Open Access2020 conference, convened by Germany’s Max Planck Society, that any notion that open access has little traction in China is misleading. Funders and research institutions in China have since 2014 encouraged — and funded — scientists to publish their papers in open-access formats, and to archive manuscripts openly online.

But, he added, much of China’s scientific output is still locked behind paywalls. “NSFC funds about 70% of Chinese research articles published in international journals, but China has to buy these back with full and high prices,“ he says. “This is simply wrong — economically and politically.”

He called on publishers at the meeting to start negotiating without delay transformative deals with Chinese library consortia. Such ‘read and publish’ agreements, which have been struck by a number of European national library consortia, and which the University of California system is also hoping to negotiate, cover the subscription costs of paywalled journals, but also allow corresponding authors at eligible institutions to publish their work open access.

Clear signal

China’s commitment to ending subscription publishing took publishers at the meeting by surprise. “This is the first clear signal I received from China on this matter,” Daniel Ropers, chief executive officer of Springer Nature, said following a heated question and answer session on the second day of the conference. “We were under the impression that open access isn’t quite as urgent an issue in China as it is in Europe and the United States. If it is indeed, we are more than happy to engage.”

Springer Nature, he says, already offers a broad range of open-access journals and would consider developing the portfolio further in all disciplines of science. But he says a viable solution is still needed for highly selective subscription journals, including Nature, to satisfy Plan S. (Nature’s news team is editorially independent of its publisher Springer Nature.)

As it stands, the plan would bar scientists from publishing their work behind a paywall after 2020, unless they can also archive their accepted manuscript immediately online with a liberal publishing licence (which few subscription journals currently permit). Many subscription journals do now offer an open-access option, but Plan S will only fund publishing papers via that ‘hybrid’ route in some cases, and will review this policy in 2023.

Two other non-European countries are expected to sign up to the plan in the coming weeks, said Smits. Smits is also seeking support among public science funders in the United States, where the private health-sciences charity the Bill and Melinda Gates Foundation is currently the only funding body to have signed Plan S.

Some scientists at the meeting said they were anxious about what the changes might mean for the evaluation of science and, ultimately, for their careers. “We’re very much in favour of open science,” says Koen Vermeir, a Dutch historian of science who heads the Global Young Academy in Halle, Germany. “But then, publishing in high-quality journals is crucial for our careers. If we can’t publish in Nature or Science anymore it would totally change the equation for us.”

Nature is seeking comment from other publishers on the announcement.

doi: 10.1038/d41586-018-07659-5
Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up

British Telecom bars Huawei’s 5G kit from core of network

Huawei 5GImage copyright Reuters
Image caption Huawei and its rivals are competing to sell 5G equipment to mobile phone networks

BT has said it will not use Huawei’s equipment within the heart of its 5G mobile network when it is rolled out in the UK.

The British firm, however, still plans to use the Chinese company’s phone mast antennas and other products deemed not to be at the “core” of the service.

BT also confirmed that it was stripping out Huawei equipment from the core of its existing 3G and 4G networks, as revealed by the Financial Times (FT).

Huawei faces security concerns.

The US-China Economic and Security Review Commission recently issued a report saying that Beijing could force Huawei and other Chinese 5G equipment-makers to “modify products to perform below expectations or fail, facilitate state or corporate espionage, or otherwise compromise the confidentiality, integrity, or availability” of networks that used them.

Last week, New Zealand became the latest country to bar a local network from using Huawei’s 5G gear.

And on Monday, intelligence service MI6’s chief Alex Younger said Britain needed to decide how comfortable it was using Chinese-owned technologies within its communications infrastructure.

Huawei denies having any ties to the Chinese government beyond those of being a law-abiding taxpayer.

However, critics point out that its founder, Ren Zhengfei, was a former engineer in the country’s army and joined the Communist Party in 1978. There are also questions about how independent of state influence any large Chinese company can be.

Long-lived partnership

BT has has long made use of Huawei’s equipment within its fixed-line network, having signed a pioneering contract with the supplier in 2005.

However, the BBC understands that BT introduced an internal policy a year later that restricted use of Huawei’s equipment to the “periphery” of its infrastructure.

This kit is sometimes referred within the industry to being “the dumb stuff at the end of the pipes”.

It meant in theory that were Huawei’s equipment to be compromised, some customers would be affected but the wider network would not collapse.

Image copyright Reuters
Image caption Huawei can still bid to supply BT with phone mast antennas and other equipment

In 2016, BT also acquired the mobile phone firm EE, which had been using the Shenzhen-based firm’s kit to push about its customers’ data.

Shortly after, BT began removing Huawei equipment that it determined to be at the “control plane” – or core – of the network.

Although this has been going on for two years, it was only publicly disclosed in response to a question by the FT.

“In 2016, following the acquisition of EE, we began a process to remove Huawei equipment from the core of our 3G and 4G networks, as part of network architecture principles in place since 2006,” said a spokesman for BT.

“We’re applying these same principles to our current RFP [request-for-proposal bid requests] for 5G core infrastructure.

“As a result, Huawei has not been included in vendor selection for our 5G core. Huawei remains an important equipment provider outside the core network, and a valued innovation partner.”

One of the benefits to BT of having kept the restriction private until now was that other equipment-makers might have been encouraged to have tendered lower bids if they believed they were competing against the Chinese firm.

Huawei has previously promoted its partnership with BT, and as recently as February issued a press release that said the two were working together on the “development and live trials” of 5G core network technologies.

However, the BBC understands the Chinese firm did not expect to win any such contracts from BT and had been focused instead on selling 5G equipment for use elsewhere in its network.

“Since it acquired EE in 2016, the BT Group has been actively bringing EE’s legacy network architecture in line with this long-standing agreement,” said a spokesman for Huawei.

“This is a normal and expected activity, which we understand and fully support.

“Working together, we have already completed a number of successful 5G trials across different sites in London, and we will continue to work with BT in the 5G era.”