Start Up No.1340: Britain buys satellites, the NHS app cockup, faulty facial recognition policing, focus falls on Wirecard’s auditors, and more

Apple’s iOS 14 will tell you when apps grab data from the clipboard, and has already told on TikTok. CC-licensed photo by 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 10 links for you. Know where you are? I’m @charlesarthur on Twitter. Observations and links welcome.

‘We’ve bought the wrong satellites’: UK tech gamble baffles experts • The Guardian

Alex Hern:


The UK government’s plan to invest hundreds of millions of pounds in a satellite broadband company has been described as “nonsensical” by experts, who say the company doesn’t even make the right type of satellite the country needs after Brexit.

The investment in OneWeb, first reported on Thursday night, is intended to mitigate against the UK losing access to the EU’s Galileo satellite navigation system.

But OneWeb – in which the UK will own a 20% stake following the investment – currently operates a completely different type of satellite network from that typically used to run such navigation systems.

“The fundamental starting point is, yes, we’ve bought the wrong satellites,” said Dr Bleddyn Bowen, a space policy expert at the University of Leicester. “OneWeb is working on basically the same idea as Elon Musk’s Starlink: a mega-constellation of satellites in low Earth orbit, which are used to connect people on the ground to the internet.

“What’s happened is that the very talented lobbyists at OneWeb have convinced the government that we can completely redesign some of the satellites to piggyback a navigation payload on it. It’s bolting an unproven technology on to a mega-constellation that’s designed to do something else. It’s a tech and business gamble.”

Giles Thorne, a research analyst at Jeffries, agreed. “This situation is nonsensical to me,” he said. “This situation looks like nationalism trumping solid industrial policy.”


This satellite company agrees you’d need to upgrade the satellites to get a navigation capability. (But they’d be very resistant to jamming.) And of course there would be satellite broadband, their original purpose, which nobody sensible wants because the latency is terrible.

The problem with LEO (low earth orbit) for GPS is that they go out of visibility so quickly. If you can’t keep a fix on a satellite for long enough, you can’t get the time on its atomic clock, so you can’t figure out your position. The 24 US GPS satellites are in MEO (medium EO) at 20,200km; the OneWeb ones are at 1,200km, and presently in a pole-to-pole orbit, though that can be changed.

This has the potential to be a money pit. Don’t say we weren’t warned.
unique link to this extract

The rise and fall of Hancock’s homegrown tracing app • Financial Times

Pilita Clark, Helen Warrell, Tim Bradshaw and Sarah Neville:


The UK was one of the first western countries to start building a phone app. By mid-March, work was under way at NHSX, a body reporting to the health department. [Health secretary Matt] Hancock, a digital enthusiast, set it up last year to drag the notoriously analogue NHS into the 21st century.

Headed by a former diplomat, Matthew Gould, it soon discovered, as other nations would, that the task was far from simple. 

The app would have to rely on a Bluetooth system that was designed for pairing a smartphone with a wireless headset rather than trigger an alert that a stranger close to you was carrying a potentially fatal virus.

Figuring out how to repurpose this technology without draining a phone’s battery was just one of the early hurdles faced by developers around the world.

In the UK, however, there was another challenge. In other nations, the app was deployed as a back-up in established national systems where widespread testing helped teams of contact tracers track down those infected.

This was impossible in the UK, which was struggling to process more than [the unsatisfactorily low number of] 8,000 tests a day in March.

That meant the British app had to be shaped in a distinctive way, said several people involved in the early stages of its development.

Instead of triggering alerts about people who had tested positive for Covid-19, developers were initially told to build an app based on people reporting virus symptoms.

This had a ripple effect that ate up developers’ time, those familiar with the process said. First, an algorithm had to be built to try to ensure someone who claimed to have a cough or fever was telling the truth.

This, in turn, meant a database had to be built to hold and process huge amounts of confidential patient symptom data that would then have to be encrypted, anonymised and protected from hackers.


This is sure to figure heftily in the surely forthcoming public inquiry. There will be lots of “well it was impossible to know at that time”. Except that Apple and Google’s shift in early April offered the chance to get it right. But Hancock and the rest ploughed on. British exceptionalism: the wrong sort.
unique link to this extract

TikTok App to stop accessing user clipboards after being caught in the act by iOS 14 • MacRumors

Juli Clover:


In a statement to The Telegraph, TikTok said that it accessed the clipboard to identify spammy behaviour:


“Following the beta release of iOS 14 on June 22, users saw notifications while using a number of popular apps.

“For TikTok, this was triggered by a feature designed to identify repetitive, spammy behavior. We have already submitted an updated version of the app to the App Store removing the anti-spam feature to eliminate any potential confusion.

“TikTok is committed to protecting users’ privacy and being transparent about how our app works.”


An update to remove the feature has already been submitted to the App Store , and a download of the new update confirms that TikTok no longer appears to be accessing the clipboard.

TikTok did not say whether the feature would be removed from Android devices, nor whether clipboard data was ever stored or moved from user devices. Other apps have also been called out for reading the clipboard, including Starbucks, Overstock, AccuWeather, several news apps, and more.


That TikTok statement is baldly self-contradictory, and a lie. TikTok isn’t transparent about how the app works, because nobody outside TikTok knew it was grabbing the clipboard until this. The protection of users’ privacy – ditto. I bet there will be some crowdsourced examination of the destination of clipboard contents using an older version. And of the Android version.

Equally, I’m surprised that any app gets access to the iOS system clipboard without the user explicitly invoking it. If I copy something in one app, why should any other app be able to see it without me switching to that app and invoking “Paste”? After all, if I copy a password in an app, does that mean any random app running in the background can see it? (App developers tell me: yes. This blogpost from February pointed out the problem.)

It’s impressive that iOS 14 is only in its first week in beta and that it’s already highlighted this sort of behaviour, but the ability to behave like this at all is peculiar.
unique link to this extract

Wrongfully accused by an algorithm • The New York Times

Kashmir Hill:


The police drove Mr. Williams to a detention center. He had his mug shot, fingerprints and DNA taken, and was held overnight. Around noon on Friday, two detectives took him to an interrogation room and placed three pieces of paper on the table, face down.

“When’s the last time you went to a Shinola store?” one of the detectives asked, in Mr. Williams’s recollection. Shinola is an upscale boutique that sells watches, bicycles and leather goods in the trendy Midtown neighborhood of Detroit. Mr. Williams said he and his wife had checked it out when the store first opened in 2014.

The detective turned over the first piece of paper. It was a still image from a surveillance video, showing a heavyset man, dressed in black and wearing a red St. Louis Cardinals cap, standing in front of a watch display. Five timepieces, worth $3,800, were shoplifted.

“Is this you?” asked the detective.

The second piece of paper was a close-up. The photo was blurry, but it was clearly not Mr. Williams. He picked up the image and held it next to his face.

“No, this is not me,” Mr. Williams said. “You think all black men look alike?”


The algorithm had decided it was him, though. Hill has done such great work on facial recognition – she was the one who revealed how ClearView had swept up as much data as it could, a hundred years ago. Well, in January.
unique link to this extract

Depixellation? Or hallucination? • AI Weirdness

Janelle Shane on an AI system which claims it can reconstruct a face from a pixellated image, but which reconstructed Obama as a white guy:


Biased AIs are a well-documented phenomenon. When its task is to copy human behavior, AI will copy everything it sees, not knowing what parts it would be better not to copy. Or it can learn a skewed version of reality from its training data. Or its task might be set up in a way that rewards – or at the least doesn’t penalize – a biased outcome. Or the very existence of the task itself (like predicting “criminality”) might be the product of bias.

In this case, the AI might have been inadvertently rewarded for reconstructing white faces if its training data (Flickr-Faces-HQ) had a large enough skew toward white faces. Or, as the authors of the PULSE paper pointed out (in response to the conversation around bias), the standard benchmark that AI researchers use for comparing their accuracy at upscaling faces is based on the CelebA HQ dataset, which is 90% white. So even if an AI did a terrible job at upscaling other faces, but an excellent job at upscaling white faces, it could still technically qualify as state-of-the-art. This is definitely a problem.

A related problem is the huge lack of diversity in the field of artificial intelligence. Even an academic project with art as its main application should not have gone all the way to publication before someone noticed that it was hugely biased. Several factors are contributing to the lack of diversity in AI, including anti-Black bias. The repercussions of this striking example of bias, and of the conversations it has sparked, are still being strongly felt in a field that’s long overdue for a reckoning.


unique link to this extract

Woke capitalism • The Bluestocking

Helen Lewis, in absolutely blistering form in her newsletter (of which this is a special Sunday version because I think this topic is bugging her mightily):


Why is [Robin] DiAngelo’s book [“White Fragility”, about why white people are so.. fragile?, and which has gone to the top of the NYT bestseller lists] so popular? Again, look at economics. Every big company has recently poured money into “diversity training”. But it is often no more scientifically grounded than Myers-Briggs: the “implicit bias” test is controversial and the claim that it can predict real world behaviour (never mind reduce bias) is a shaky one. But it looks like something solid and quantitative – ooh! a test score! – and therefore metrics-obsessed modern companies (ie all of them) love it. People have been given the idea that confronting their own bias is the best way to address racism. (Also, let’s be honest, there are probably quite a lot of people who have bought White Fragility because they think of themselves as the kind of person who would read a book like that, without actually wanting to read a book like that. It’s a Social Justice Brief History of Time.)

Anyway, here’s Harvard Kennedy School professor of public policy Iris Bohnet:


“About $8 billion a year is spent on diversity trainings in the United States alone. Now, I tried very hard to find any evidence I could. I looked not just in the United States but also in Rwanda and other post-conflict countries, where reconciliation is often built on the kind of diversity trainings that we do in our companies, to see how this is working. Sadly enough, I did not find a single study that found that diversity training in fact leads to more diversity.”


Eight billion dollars a year! Imagine if you put that money into, say, paying all your junior staff a wage which allows them to live in the big city where your company is based, without needing help from their parents. You’d probably do more good at increasing your company’s diversity. Hell, get your staff to read White Fragility on their own time and give your office cleaners a pay rise.


“A Social Justice ‘Brief History Of Time'”. What a gut punch. (BHOT is famously the book everyone owned but nobody read.) But she’s absolutely right. Lewis ranges far and wide here: since it’s statue-toppling time, she’s noticed plenty of clay feet around the place. I love her comment: “the only question I want to ask big companies who chirrup on about ’empowering the female leaders of the future’ is this one: Do you have a creche?” Actions, not words. Don’t miss her analysis of why Stonewall is emphasising the “T” in “LGBT” either.
unique link to this extract

Wirecard scandal puts spotlight on auditor Ernst & Young • WSJ

Patricia Kowsmann, Paul J. Davies and Juliet Chung:


Ernst & Young, auditor to insolvent German fintech company Wirecard, had questions related to unorthodox arrangements under which the company’s cash was held in bank accounts it didn’t control as far back as 2016, according to emails seen by The Wall Street Journal.

The auditor subsequently signed off on three years of Wirecard’s financial results with those arrangements in place.

Now $2bn that was held in those accounts has disappeared. Wirecard says the money probably doesn’t exist. On Friday, a German shareholder association filed a criminal complaint to the prosecutors’ office in Munich, where Wirecard is based, accusing EY auditors of missing the alleged fraud.

“We feel Ernst & Young’s auditing work was a disaster,” said Marc Liebscher, whose Berlin-based law firm is representing the private Wirecard investors who filed the complaint. “Our clients are convinced, Ernst & Young should stand trial.”

EY said it had been duped along with everyone else. “There are clear indications that this was an elaborate and sophisticated fraud, involving multiple parties around the world in different institutions, with a deliberate aim of deception,” it said.

…Investors who bet Wirecard’s share price would fall have been sending detailed complaints to EY for years, flagging their concerns and media reports that raised questions about the company’s accounting and business practices, based on letters reviewed by the Journal.

“They’ve basically turned a blind eye toward the critics that raised very serious allegations,” said Fraser Perring, who with his former partner, Matthew Earl, published an early report critical of Wirecard in 2016.


If it takes a whistleblower going to journalists for the fraud to unravel, then the auditors self-evidently haven’t been doing their job. The fraud was there to be found; as in so many cases (see also: Enron), it’s already visible in the books. But auditors hate to rock the boat: they’re the classic case of not understanding something if your job relies on not understanding it.
unique link to this extract

Lest we forget the horrors: a catalog of Trump’s worst cruelties, collusions, corruptions, and crimes: the complete listing (so far): atrocities 1-759 • McSweeney’s Internet Tendency

Ben Parker, Stephanie Steinbrecher, Kelsey Ronan, John Mcmurtrie, Sophia Durose, Rachel Villa, and Amy Sumerton do what I had hoped someone would do. (If you close your eyes and wish hard enough, the internet will crowdsource whatever you want.) Their “atrocity key” includes these categories:
– Sexual Misconduct, Harassment, & Bullying
– White Supremacy, Racism, & Xenophobia
– Public Statements / Tweets
– Collusion with Russia & Obstruction of Justice
– Trump Staff & Administration
– Trump Family Business Dealings
– Policy
– Environment

I thought they’d missed one in the “White supremacy, racism & xenophobia” from their “before January 2017” category: his call, in a full-page advert, for the death sentence for the five black suspects who were arrested after the rape of a woman in Central Park on the basis that he thought they were guilty of sexual assault. (They were exonerated on DNA evidence and the real criminal’s confession.) But it turns out Trump reiterated that lie in October 2016.

Needless to say, it’s a long, long read, even though they don’t really get started until he’s in office.
unique link to this extract

Microsoft to permanently close all of its retail stores • The Verge

Chris Welch:


Microsoft is giving up on physical retail. Today the company announced plans to permanently close all Microsoft Store locations in the United States and around the world, except for four locations that will be “reimagined” as experience centers that no longer sell products.

Those locations are New York City (Fifth Ave), London (Oxford Circus), Sydney (Westfield Sydney), and the Redmond campus location. The London store only just opened about a year ago. All other Microsoft Store locations across the United States and globally will be closing, and the company will concentrate on digital retail moving forward. Microsoft says and the Xbox and Windows storefronts reach “up to 1.2 billion monthly customers in 190 markets.”

The company tells The Verge that no layoffs will result from today’s decision. “Our commitment to growing and developing careers from this diverse talent pool is stronger than ever,” Microsoft Store VP David Porter said in a LinkedIn post on the move.

A source with knowledge of Microsoft’s retail operations told The Verge that this plan was originally in place for next year, but was accelerated by the COVID-19 pandemic.


So wonderful how Microsoft calls it “a new approach” to retail. Approach as in full speed retreat. The cost: $450m (“asset writeoffs and impairments”). Began in 2009, had 116 stores in 2018, but the number had fallen to 82 at the time of closure. Sorry, “new approach”.
unique link to this extract

Radioactivity hike seen in Northern Europe; source unknown • The New York Times


Nordic authorities say they detected slightly increased levels of radioactivity in northern Europe this month that Dutch officials said may be from a source in western Russia and may “indicate damage to a fuel element in a nuclear power plant.”

But Russian news agency TASS, citing a spokesman with the state nuclear power operator Rosenergoatom., reported that the two nuclear power plans in northwestern Russia haven’t reported any problems.

The Leningrad plant near St. Petersburg and the Kola plant near the northern city of Murmansk, “operate normally, with radiation levels being within the norm,” Tass said.

The Finnish, Norwegian and Swedish radiation and nuclear safety watchdogs said this week they’ve spotted small amounts of radioactive isotopes harmless to humans and the environment in parts of Finland, southern Scandinavia and the Arctic.

The Swedish Radiation Safety Authority said Tuesday that “it is not possible now to confirm what could be the source of the increased levels” of radioactivity or from where a cloud, or clouds, containing radioactive isotopes that has allegedly been blowing over the skies of northern Europe originated. Its Finnish and Norwegian counterparts also haven’t speculated about a potential source.


It’ll be the Chernobyl Reenactment Group, won’t it.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

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.