Three pieces of information are enough to uniquely identify 81% of Americans from an “anonymised” database. CC-licensed photo by Corey Seeman on Flickr.

You can send this link to your friends if they want to get each day’s Start Up post by email. They’ll need to click a confirmation link, so no spam.

A selection of 10 links for you. Mad dogs and Englishmen. I’m @charlesarthur on Twitter. Observations and links welcome.

Mark Zuckerberg’s Ponzi scheme • Aaron Greenspan

»

Zuckerberg’s version [of a Ponzi scheme] is slightly different, but only slightly: old users leave after getting bored, disgusted and distrustful, and new users come in to replace them. Except that as Sam Lessin told us, the “new users” part of the equation was already getting to be a problem in 2012. To balance it out and keep “growth” on the rise, all Facebook had to do was turn a blind eye. And did it ever.

In Singer v. Facebook, Inc.—a lawsuit filed in the Northern District of California alleging that Facebook has been telling advertisers that it can “reach” more people than actually exist in basically every major metropolitan area—the plaintiffs quote former Facebook employees, understandably identified only as Confidential Witnesses, as stating that Facebook’s “Potential Reach” statistic was a “made-up PR number” and “fluff.” Also, that “those who were responsible for ensuring the accuracy ‘did not give a shit.'” Another individual, “a former Operations Contractor with Facebook, stated that Facebook was not concerned with stopping duplicate or fake accounts.”

That’s probably because according to its last investor slide deck and basic subtraction, Facebook is not growing anymore in the United States, with zero million new accounts in Q1 2019, and only four million new accounts since Q1 2017. That leaves the rest of the world, where Facebook is growing fastest “in India, Indonesia, and the Philippines,” according to Facebook CFO David Wehner. Wehner didn’t mention the fine print on page 18 of the slide deck, which highlights the Philippines, Indonesia and Vietnam as countries where there are “meaningfully higher” percentages of, and “episodic spikes” in, fake accounts. In other words, Facebook is growing the fastest in the locations worldwide where one finds the most fraud. In other other words, Facebook isn’t growing anymore at all—it’s shrinking. Even India, Indonesia and the Philippines don’t register as many searches for Facebook as they used to. Many of the “new” users on Instagram are actually old users from the core platform looking to escape the deluge of fakery.

«

Before you say “who is this Greenspan guy anyway?” – he’s the person on whose computer Zuckerberg wrote the original code for thefacebook.com. So he’s known Zuck a little while.
unique link to this extract

---

Facebook deceived users about the way it used phone numbers, facial recognition, FTC to allege in complaint • The Washington Post

Tony Romm:

»

The Federal Trade Commission plans to allege that Facebook misled users’ about its handling of their phone numbers as part of a wide-ranging complaint that accompanies a settlement ending the government’s privacy probe, according to two people familiar with the matter.

In the complaint, which has not yet been released, federal regulators take issue with Facebook’s earlier implementation of a security feature called two-factor authentication. It allows users to request one-time password, sent by text message, each time they log onto the social-networking site.

But some advertisers managed to target Facebook users who uploaded those contact details, perhaps without the full knowledge of those who provided them, the two sources said. The misuse of the phone numbers was first identified in media reports and by academics this year.

The FTC also plans to allege that Facebook had provided insufficient information to users — roughly 30 million — about their ability to turn off a tool that would identify and offer tag suggestions for photos, the sources added.

«

The switcheroo of getting people to supply their phone number, and then using that for advertising – that’s doubly crap: it discourages people who hear about it from securing their account (and perhaps securing it on other platforms because they fear the same), and it gives advertisers access to people that the people haven’t consented to.

I wonder if there’s a GDPR version of that – though it’s not clear whether Facebook did this in Europe.
unique link to this extract

---

Apple dominates App Store search results, thwarting competitors • WSJ

Tripp Mickle:

»

Audiobooks.com, an RBmedia company, largely held the No. 1 ranking in “audiobooks” searches in the App Store for nearly two years. Then last September it was unseated by Apple Books. The Apple app had only recently begun marketing audiobooks directly for the first time.

“It was literally overnight,” said Ian Small, Audiobooks.com’s general manager. He said the change triggered a 25% decline in Audiobooks.com’s daily app downloads. The app at the time had 35,000 customer reviews and a 4.8 on the App Store’s 5-star ranking. The preinstalled Apple Books app, with no reviews or ratings, has since ranked No. 1 in searches for “audiobooks.” It also ranks first in searches for “books” and “reader.”

Apple says the No. 1 position for Books in a “books” search is reasonable, since it is an exact name match. The app was also first for “audiobooks” because of “user behavior data” and the inclusion of “audiobooks” as a keyword associated with the app, a spokesman said.

Apple’s role as both the creator of the App Store’s search engine and the beneficiary of its results has rankled developers. They contend Apple is essentially pinning its apps No. 1, compelling anyone seeking alternatives to consider Apple apps first. Such a tactic would help preserve loyalty to Apple’s mobile operating system—a key to future iPhone sales—and encourage the use of revenue-generating apps such as Apple TV and News, developers say.

«

I’d be surprised if Apple’s apps didn’t do well in searches for music or books. The WSJ graphic on this is pretty impenetrable. Quite what the algorithm is, nobody knows.
unique link to this extract

---

Justice Department announces broad antitrust review of Big Tech • The Verge

Makena Kelly:

»

“Without the discipline of meaningful market-based competition, digital platforms may act in ways that are not responsive to consumer demands,” said Assistant Attorney General Makan Delrahim of the Antitrust Division. “The Department’s antitrust review will explore these important issues.”

The investigation will address broad concerns over whether Big Tech is stifling competition, the Wall St Journal said, and will be separate from the department’s probes of Google and Apple that were reported earlier this summer that are intended to take a closer look at individual potential violations. The review reported today will look into search engines, social media platforms, and retail, but not focus on any individual company or practice.

In a press release, the Justice Department said the review “will consider the widespread concerns that consumers, businesses, and entrepreneurs have expressed about search, social media, and some retail services online.” The Department declined further comment beyond the release.

At Attorney General Barr’s confirmation hearing this past January, he told senators that he would like to see the Justice Department take a harder look at whether companies like Google and Amazon were abusing their market dominance.

«

The press release is super-vague. If it were any more vague it would be written in white text on a white background.
unique link to this extract

---

You’re very easy to track down, even when your data has been anonymized • MIT Technology Review

Charlotte Jee:

»

Researchers from Imperial College London and the University of Louvain have created a machine-learning model that estimates exactly how easy individuals are to reidentify from an anonymized data set. You can check your own score by entering your zip code, gender, and date of birth.

On average, in the US, using those three records, you could be correctly located in an “anonymized” database 81% of the time. Given 15 demographic attributes of someone living in Massachusetts, there’s a 99.98% chance you could find that person in any anonymized database.

“As the information piles up, the chances it isn’t you decrease very quickly,” says Yves-Alexandre de Montjoye, a researcher at Imperial College London and one of the study’s authors.

The tool was created by assembling a database of 210 different data sets from five sources, including the US Census. The researchers fed this data into a machine-learning model, which learned which combinations are more nearly unique and which are less so, and then assigns the probability of correct identification.

This isn’t the first study to show how easy it is to track down individuals from anonymized databases. A paper back in 2007 showed that just a few movie ratings on Netflix can identify a person as easily as a Social Security number, for example. However, it shows just how far current anonymization practices have fallen behind our ability to break them.

«

unique link to this extract

---

Browsers are pretty good at loading pages, it turns out • Carter Sande

»

MDN is a documentation/tutorial website run by the creators of Firefox. They basically wrote the book on how to make websites. So when they created a beta version of their site that used client-side navigation, I asked them why, and one of their developers responded that it was to load pages more quickly. With the normal hyperlink tag, the browser has to download the HTML code for the whole page, but if you write the behaviour yourself in JavaScript, you can make sure to only download the part of the new page that’s different from the old one. I’m usually lucky enough to have a pretty fast Internet connection, so I never really noticed the difference.

But I got the chance to experience the benefits of client-side navigation on a recent trip to Canada… I fired up both versions of MDN on my phone, and clicked a link on each one to see how much faster the JavaScript version would be:

Hang on a second! The JavaScript version wasn’t faster, it was way slower! What gives?

It turns out that browsers have a lot of tricks up their sleeves that help them put pages on the screen more quickly. A big one we’re seeing here is called progressive rendering: browsers download the top part of the page first, then show it on the screen while the rest of the page finishes downloading. The JavaScript version has to wait for the entire JSON response to come back before it can show anything on the screen, so it feels slower.

«

In other words: forget those Javascript people. They’re just complicating things.
unique link to this extract

---

The video game industry can’t go on like this • Kotaku

Joshua Rivera:

»

Ubisoft is an exception, regularly releasing entries in single-player game franchises like Far Cry and Assassin’s Creed. But it buttresses them with aggressive microtransactions and extensive season pass plans. (And the occasional diversion like Trials Rising and South Park: The Fractured But Whole.) The big-budget single-player experience is now almost entirely the domain of first-party studios making marquee games for console manufacturers, which bankroll games like Spider-Man and God of War. The economics of first-party exclusives are totally different—they’re less about making money by themselves and more about drawing players into the console’s ecosystem.

This is worth considering, because as big publishers prioritize live, service-oriented games, the number of games on their schedules has dropped. If you look at the Wikipedia listings for EA, Ubisoft, and Activision games released by year, you’ll get a stark—if unscientific—picture of how each big publisher’s release slate has thinned out in the last five years, relying on recurring cash cows like sports games and annualized franchises and little else. In 2008, those three publishers released 98 games; in 2018 they released just 28, not including expansions.

In short, the single-player game was not sustainable. So why should we think the current model is?

«

Also: how super-expensive it is to make a big video game. The top end is getting eaten from below by simpler games which do all that people want; a classic “disruptive innovation” change, where the cheap games don’t have all the bells and whistles, but people don’t want bells and whistles.
unique link to this extract

---

Barr revives debate over ‘warrant-proof’ encryption • WSJ

Dustin Volz:

»

While [US Attorney General William] Barr offered examples in which he said encryption thwarted criminal investigations, he didn’t provide fresh statistics about the extent of the problem. The FBI suffered a setback last year when it revealed it had accidentally inflated public statistics about the number of encrypted devices investigators were unable to break open, and officials haven’t provided an updated metric since then.

Mr. Barr sought to convince technology companies to work toward a compromise with law-enforcement agencies, lest they are forced to deal with hastily passed laws in the wake of a crisis. “Given the frequency with which these situations are now arising, it is only a matter of time before a sensational case crystallizes the issue for the public,” Mr. Barr said.

The National Security Council convened a deputies meeting from various federal agencies last month to consider options on how to move forward on the encryption issue, but the meeting ended without any clear resolution on how to proceed, according to people briefed on it.

A US official said Mr. Barr’s speech wasn’t aimed at outlining the path forward but reflected consensus within the Trump administration that a solution must be found to address the proliferation of too-tough-to-crack encryption. Critics contend that the FBI and contractors that specialize in bypassing encryption possess the tools to get into many devices they want to unlock.

Sen. Ron Wyden (D., Ore.), a longtime privacy advocate who has vocally opposed government efforts to weaken encryption, called it an “outrageous, wrongheaded and dangerous proposal.”

«

No clue what to do about it, but sure something should be done about it, and picking the wrong thing to do about it: the Trump administration in a nutshell. There’s already been a “sensational case” – the San Bernadino one in 2016 – and the FBI paid an Israeli company about $1m to break into the iPhone in question, to find nothing useful. There was more, and better, data on the terrorists’ Facebook profiles.
unique link to this extract

---

New York City to consider banning sale of cellphone location data • The New York Times

Jeffery Mays:

»

Telecommunications firms and mobile-based apps make billions of dollars per year by selling customer location data to marketers and other businesses, offering a vast window into the whereabouts of cellphone and app users, often without their knowledge.

That practice, which has come under increasing scrutiny and criticism in recent years, is now the subject of proposed legislation in New York. If passed, it is believed that the city would become the first to ban the sale of geolocation data to third parties.

The bill, which will be introduced on Tuesday, would make it illegal for cellphone companies and mobile app developers to share location data gathered while a customer’s mobile device is within the five boroughs.

Cellphone companies and mobile apps collect detailed geolocation data of their users and then sell that information to legitimate companies such as digital marketers, roadside emergency assistance services, retail advertisers, hedge funds or — in the case of a class-action lawsuit filed against AT&T — bounty hunters.

“The average person has no idea they are vulnerable to this,” said Councilman Justin L. Brannan, a Brooklyn Democrat who is introducing the bill. “We are concerned by the fact that someone can sign up for cell service and their data can wind up in the hands of five different companies.”

«

Just me, or is it madness that NYC is only the first, and that this is only “proposed” legislation which, the story says, will be strongly opposed by the ad tech industry “which has a strong presence in the city”. Make their execs’ location data public, let’s see how they feel about it then.
unique link to this extract

---

16in MacBook Pro rumored to launch in October • 9to5Mac

Chance Miller:

»

Apple’s rumored 16in MacBook Pro could launch in October, according to a new supply chain report from the Economic Daily News. The report also says that Apple will release updated versions of the 13in MacBook Pro and Retina MacBook Air in October.

Today’s report corroborates that the 16in MacBook Pro will launch with a 3072×1920 LCD display, which is up from the 2880×1800 panel in the 15in MacBook Pro.

As for pricing, the report says that the 16in MacBook Pro will bring a “new high price for Apple notebooks.” The supply chain industry reportedly expects the laptop to start at around $3,000, with Apple positioning it between the iMac and iMac Pro as a portable option for users with pro needs.

While Apple did just refresh the MacBook Air, the update only added True Tone display technology and left things like the processor the same. A refresh in the fall could bring improved performance among other changes. For instance, Ming-Chi Kuo has said that Apple will shift to a new scissor switch keyboard in the MacBook Air this year. 

«

Blimey, that really is priced for the iMac Pro brigade. Though the pro laptops always used to be; somehow they seemed to have edged down in price over the years. Now they would be jacking up again. Personally, I’ve used a 15in screen for so long that I don’t think I could live with something smaller, but maybe that’s just habit – which can be broken. Scissor switches, though. Going to love hearing Apple’s execs explaining that one, if it comes to pass.
unique link to this extract

---

Errata, corrigenda and ai no corrida: misbehaving HTML meant that yesterday’s pictures of the middle seats on airlines didn’t work. (Now fixed – thanks Timothy.) And the post came live an hour earlier than normal – you didn’t oversleep.