A crowded beach might contain a lot of people, but data shows it doesn’t pose a Covid risk.CC-licensed photo by Joe Shlabotnik 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. Use them wisely. I’m @charlesarthur on Twitter. Observations and links welcome.
The European Union is poised to ban artificial intelligence systems used for mass surveillance or for ranking social behavior, while companies developing AI could face fines as high as 4% of global revenue if they fail to comply with new rules governing the software applications.
The rules are part of legislation set to be proposed by the European Commission, the bloc’s executive body, according to a draft of the proposal obtained by Bloomberg. The details could change before the commission unveils the measure, which is expected to be as soon as next week.
The EU proposal is expected to include the following rules:
• AI systems used to manipulate human behavior, exploit information about individuals or groups of individuals, used to carry out social scoring or for indiscriminate surveillance would all be banned in the EU. Some public security exceptions would apply.
• Remote biometric identification systems used in public places, like facial recognition, would need special authorization from authorities.
• AI applications considered to be ‘high-risk’ would have to undergo inspections before deployment to ensure systems are trained on unbiased data sets, in a traceable way and with human oversight.
• High-risk AI would pertain to systems that could endanger people’s safety, lives or fundamental rights, as well as the EU’s democratic processes – such as self-driving cars and remote surgery, among others.
• Some companies will be allowed to undertake assessments themselves, whereas others will be subject to checks by third-parties. Compliance certificates issued by assessment bodies will be valid for up to five years.
• Rules would apply equally to companies based in the EU or abroad.
Let’s not forget it was the EU that gave us the delight of cookie approval on every site you visit. So I’m a little wary of their efforts to make everything wonderful in the world of AI. What counts as “AI systems used to manipulate human behaviour”? Surely that’s Facebook’s and Twitter’s and Instagram’s newsfeed algorithms. And, arguably, YouTube’s algorithm (using AI) which tries to make you watch more videos.
unique link to this extract
Charlie Warzel (with the first of his Substack-sponsored newsletters, lured away from the NY Times):
Even if you don’t know the story, you probably have a sense of what happened next. Her lopsided Twitter poll answers [to the question: is Alien a horror film?] quickly indicated she’d expressed A Bad Opinion. The quote retweets started rolling in. The first few seemed like faux outrage at a particularly spicy genre opinion that kept with the spirit of her post — but that eventually shifted, too. People were mad. Hunt logged off and went to bed.
She woke up to angry emails from strangers. Overnight, friends in the U.K. sent concerned messages asking if she was doing alright. She opened Twitter and found her poll had 120,000 votes. Over 6000 people, including filmmakers like Kevin Smith, had angrily quote tweeted her, many demanding that she apologize — to Film Twitter, to prominent directors, to the medium of space itself. The reason? Once the tweet picked up steam, it was elevated into Twitter’s Trending Topics widget in the U.S and U.K.
The whole affair is a perfect example of context collapse, which generally occurs when a surfeit of different audiences occupy the same space, and a piece of information intended for one audience finds its way to another — usually an uncharitable one — which then reads said information in the worst possible faith (You can read about the origins here from scholar danah boyd).
In this case, the collapse was substantially amplified by Twitter’s Trending widget, which took an anodyne opinion by a verified Twitter user and displayed it to millions of random people as if it was some kind of significant pop cultural event. “My imagined audience when I tweeted this was, ‘oh, we’re all at the bar and having this low stakes debate,” she told me recently. “In retrospect, that was totally naive to think anyone would have taken it that way.”
The point of Twitter’s Trending Topics is ostensibly to surface significant news and Twitter commentary and invite others to ‘join the conversation.’ Left unsaid, of course, is that ‘the conversation’ at scale is complete garbage — an incomprehensible number of voices lecturing past each other.
Twitter has announced a new plan to study the fairness of its algorithms. As part of the effort, which the company has dubbed the “Responsible Machine learning Initiative,” data scientists and engineers from across the company will study potential “unintentional harms” caused by its algorithms and make the findings public.
“We’re conducting in-depth analysis and studies to assess the existence of potential harms in the algorithms we use,” the company wrote in a blog post announcing the initiative.
To start, the company will study Twitter’s image cropping algorithm, which has been criticized as being biased toward people with lighter skin. Twitter will also study its content recommendations, including a “a fairness assessment of our Home timeline recommendations across racial subgroups,” and “an analysis of content recommendations for different political ideologies across seven countries.”
It’s not clear how much of an impact this initiative will have. Twitter notes that in some cases it may change aspects of its platform based on its findings, and other studies may simply result in “important discussions around the way we build and apply ML [machine learning].” But the issue is a timely one for Twitter and other social media platforms.
Twitter, the company, is strange. There’s all this talk about “fairness” of the algorithms, yet they also drive all the wild engagement that causes absurd pile-ons. That’s exactly the intention of the algorithm(s). It’s as though they think they’re painting everything green, but are colour-blind and mix up red and green paints. (Yes, because it’s a male sort of thing.)
unique link to this extract
According to Prof Mark Woolhouse, an epidemiologist at Edinburgh University who sits on the government’s SPI-M committee, the chance of a super-spreader event among the crowds that turned up from Bournemouth to Southend was minimal in theory – and nonexistent in practice.
“Over the summer we were treated to all this on the television news, pictures of crowded beaches, and there was an outcry about this,” he told MPs. “There were no outbreaks linked to public beaches. There’s never been a Covid-19 outbreak linked to a beach, ever, anywhere in the world, to the best of my knowledge.”
If that version of events seems at odds with stern warnings from the health secretary, Matt Hancock, that outdoor exercise could be banned, and an accusation that sunbathers were putting lives at risk, it is wholly consistent with the scientific evidence, other experts agree.
“We have known for some time that only about 10% of transmission events are linked to outdoor activities,” said Dr Müge Çevik, a lecturer in infectious diseases and medical virology at the University of St Andrews.
“Even those events generally involve either prolonged close contact or a mixture of indoor and outdoor time. We had a lot of existing knowledge even when the pandemic began about respiratory viruses and how they transmit in general, and everything directs us to the conditions in people’s homes and workplaces.”
Derek Thompson, after the US Center for Disease Control revised its view of the threat of catching Covid from surfaces down to “less than 1 in 10,000”:
I’ve been writing about our misplaced obsession with surface hygiene since the summer. Like many, I spent the early months of the pandemic dunking my apples and carrots in soap. That was before I read a persuasive essay in the medical journal The Lancet by Emanuel Goldman, a microbiology professor at Rutgers New Jersey Medical School: “Exaggerated Risk of Transmission of COVID-19 by Fomites.” (In medical jargon, fomites are objects and surfaces that can transmit an infectious pathogen.)
This opinion ran contrary to the conventional wisdom of the broader scientific community, and Goldman told me that several journals rejected his essay. But he was not alone in his quest. Writers such as my colleague Zeynep Tufekci and researchers such as Jose-Luis Jimenez, an aerosol scientist at the University of Colorado Boulder, were also outspoken in their insistence that we needed to focus on ventilation rather than surfaces, windows rather than Windex. They were rebuffed, not only by loudmouths on Twitter and on TV, but by other scientists who clung stubbornly to an outdated view of viral spread.
Over the weekend, I caught up with Goldman to ask how it felt to be vindicated by the world’s most famous public-health organization. “On a personal level, I feel great,” he said. “But I’m kind of wondering what took them so long. There is so much inertia in the scientific establishment.”
These days, Goldman is extending his crusade against fomite fear from COVID-19 to other diseases. The old story is that if you make contact with a surface that a sick person touched, and then you touch your eyes or lips, you’ll infect yourself. While Goldman acknowledges that many diseases, especially bacterial diseases, spread easily from surfaces, he now suspects that most respiratory viruses spread primarily through the air, like SARS-CoV-2 does.
“For most respiratory viruses, the evidence for fomite transmission looks pretty weak,” Goldman said. “With the exception of RSV [respiratory syncytial virus], there are few other respiratory viruses where fomite transmission has been conclusively shown.”
Strangely, though the CDC revised guidance (which is very wishy-washy, and doesn’t say “well, treat the risk as zero” – even though 1 in 10,000 is about the risk of dying by self-inflicted accidental injury in the next year) quotes 36 papers, none of them is Goldman or Jimenez. That inertia is substantial. (Via John Naughton.)
unique link to this extract
China should fully lift the family planning policy to ease ageing trend: central bank paper • Global Times
In a rare move, the People’s Bank of China (PBC), the country’s central bank, published on Wednesday a working paper indicating that the country should not only fully lift the family planning policy, but also encourage couples to have more children in an effort to tackle the ageing population trend.
The main contradiction facing China at the moment has changed, from population expansion to the imminent loss of demographic dividend and the growing crisis of population ageing and fewer births. The report, “Understanding and Countermeasures on China’s Population Transition,” published by the PBC highlights that it is necessary to lift the fertility rate and allow people to have more children in a timely manner. Additionally, the pension system should also be improved, the paper noted.
Any delay would miss the valuable window of opportunity to respond to the demographic transition with fertility policies, repeating the fate of developed countries, said the paper.
The demographic transition is a new phenomenon for human beings, and so far, only developed countries have experienced this transition, with an ageing population and fewer children.
Hugely significant. The formal abandonment of the One Child policy in 2015 was the first part, having got through a pinch point in food supply v demographics; now it’s looking to expand again.
unique link to this extract
One big problem with climate economics is that there just isn’t enough of it. Climate change is going to affect every facet of our economy. Quantitatively, it’s vastly more important than any optimal tax calculation or detail of occupational licensing; it’s arguably even more important than the business cycle itself. But the number of papers at top journals dedicated to climate economics is miniscule. In a scathing 2019 article entitled “Why are economists letting the world down on climate change?”, economists Andrew Oswald and Nicholas Stern write:
We are sorry to say that we think academic economists are letting down the world. Economics has contributed disturbingly little to discussions about climate change. As one example, the Quarterly Journal of Economics, which is currently the most-cited journal in the field of economics, has never published an article on climate change…
We suspect that modern economics is stuck in a kind of Nash equilibrium. Academic economists are obsessed with publishing per se and with pleasing potential referees. The reason there are few economists who write climate change articles, we think, is because other economists do not write climate change articles.
…If there is one climate economist who is respected above all others, it’s William Nordhaus of Yale, who won the Econ Nobel in 2018 “for integrating climate change into long-run macroeconomic analysis.” The prize specifically cited Nordhaus’ creation of an “integrated assessment model” for analyzing the costs of climate change. The most famous of these is the DICE Model, used by the Environmental Protection Agency.
But the DICE Model, or at least the version we’ve been using for years, is obviously bananas. As climate writer David Roberts noted in 2018, according to the standard version of Nordhaus’ model, the economic cost of a 6°C increase in global temperatures would only be 10% of GDP. As Roberts notes, climate scientists believe that that level of temperature increase would make the Earth basically unlivable. An unlivable Earth is going to cost a lot more than 10% of GDP.
His point being that Biden’s administration is largely ignoring the bad advice, and going for more aggressive moves on climate economy.
unique link to this extract
Dan Murtaugh, Colum Murphy, James Mayger and Brian Eckhouse:
Three owners of Xinjiang’s polysilicon refineries have been linked to a state-run employment program that, according to some foreign governments and academics, may at times amount to forced labor. China denies such accusations and recently insisted that journalists and diplomats are free to go see for themselves.
That’s why two Bloomberg reporters went to Xinjiang in March, after weeks of unsuccessful requests for factory tours. Such visits aren’t unusual elsewhere in China. But this time a security apparatus sprang into action. Upon our landing in Urumqi, two police officers boarded the plane, one with an automatic weapon slung across his chest and a photo identifying one of the reporters in hand. After questioning on the tarmac, we left the airport. For the next three days agents followed us everywhere, obstructing all attempts to speak to locals and deleting our photos.
The veil over Xinjiang has made the search for answers about the links between China’s labor program and its solar industry a job for outside researchers—who, it turns out, have found potentially telling details just by combing through public records.
The owner of one polysilicon factory, GCL-Poly Energy Holdings Ltd., said in a 2019 report that it had accepted 121 poor minority workers from the Uyghur heartland in southern Xinjiang. Photos posted by the local government in June 2017 show workers, lined up in blue uniforms, about to be sent by the labor program to companies including East Hope Group Co., an aluminum smelter that in recent years also started producing polysilicon in Xinjiang. The previously unreported document was found by Adrian Zenz, a German researcher based in Minnesota who’s become a chief source of data about the labor program in Xinjiang—and thus a focus of China’s wrath.
Kudos to the Bloomberg pair in China (Murphy and Mayger) for trying. From experience, a confrontation with those types is immediately, viscerally concerning.
unique link to this extract
The 2022 iPhone lineup will feature substantial changes to the camera system, according to analyst Ming-chi Kuo in an investor note reported on by MacRumors and AppleInsider. Kuo says the high-end iPhones — presumably the Pro range — will use a larger 48-megapixel sensor for the primary camera. Separately, the analyst suggests that 2023 iPhones may start to use Face ID sensors under the screen as a way to get rid of the notch.
48-megapixel sensors have been common in Android phones for years, but this component will reportedly be larger than most. It’ll be a 1/1.3-inch sensor with a pixel size of 1.25µm, according to Kuo. That’s smaller than the 1.7µm pixels in the iPhone 12 Pro Max, but Kuo says that the equivalent pixel size will be more like 2.5µm when the sensor is used for 12-megapixel images. Kuo also expects the higher-resolution sensor to enable 8K video capture.
The 2022 iPhone lineup is going to see a reduction in screen size options, Kuo says; there won’t be a 5.4in mini any more, and Apple will stick to the 6.1in and 6.7in displays on other models. The mini has reportedly not sold to Apple’s expectations, with the company said to have overestimated demand and cut production orders. The 2021 lineup, however, is expected to keep the same screen sizes.
It always puzzles me what possible use stories like this can be. Might you be holding out to not buy this year’s iPhone because of what there might be in next year’s? And it seems like, for any failings, the iPhone mini will be there this year – which is more than the 5C managed.
unique link to this extract
Suicide trends in the early months of the COVID-19 pandemic: an interrupted time-series analysis of preliminary data from 21 countries • The Lancet Psychiatry
(A huge number of authors):
We sourced data from 21 countries (16 high-income and five upper-middle-income countries), including whole-country data in ten countries and data for various areas in 11 countries). Rate ratios (RRs) and 95% CIs based on the observed versus expected numbers of suicides showed no evidence of a significant increase in risk of suicide since the pandemic began in any country or area. There was statistical evidence of a decrease in suicide compared with the expected number in 12 countries or areas: New South Wales, Australia (RR 0·81 [95% CI 0·72–0·91]); Alberta, Canada (0·80 [0·68–0·93]); British Columbia, Canada (0·76 [0·66–0·87]); Chile (0·85 [0·78–0·94]); Leipzig, Germany (0·49 [0·32–0·74]); Japan (0·94 [0·91–0·96]); New Zealand (0·79 [0·68–0·91]); South Korea (0·94 [0·92–0·97]); California, USA (0·90 [0·85–0·95]); Illinois (Cook County), USA (0·79 [0·67–0·93]); Texas (four counties), USA (0·82 [0·68–0·98]); and Ecuador (0·74 [0·67–0·82]).
This is the first study to examine suicides occurring in the context of the COVID-19 pandemic in multiple countries. In high-income and upper-middle-income countries, suicide numbers have remained largely unchanged or declined in the early months of the pandemic compared with the expected levels based on the pre-pandemic period.
Errata, corrigenda and ai no corrida: none notified