From this summer, you won’t have to put your laptop in here in a growing number of US airports. CC-licensed photo by Rakesh A on Flickr.
A selection of 10 links for you. Was your April Fool’s joke funny? I’m @charlesarthur on Twitter. Observations and links welcome.
Douglas Preston on a find – in south west North Dakota (try saying it out loud) – of fossils from the day when the asteroid struck and wiped out the dinosaurs:
[Robert] DePalma returned to do a preliminary excavation of the site. “Almost right away, I saw it was unusual,” he told me. He began shovelling off the layers of soil above where he’d found the fish. This “overburden” is typically material that was deposited long after the specimen lived; there’s little in it to interest a paleontologist, and it is usually discarded. But as soon as DePalma started digging he noticed grayish-white specks in the layers which looked like grains of sand but which, under a hand lens, proved to be tiny spheres and elongated droplets. “I think, Holy shit, these look like microtektites!” DePalma recalled. Micro tektites are the blobs of glass that form when molten rock is blasted into the air by an asteroid impact and falls back to Earth in a solidifying drizzle. The site appeared to contain micro tektites by the million.
As DePalma carefully excavated the upper layers, he began uncovering an extraordinary array of fossils, exceedingly delicate but marvellously well preserved. “There’s amazing plant material in there, all interlaced and interlocked,” he recalled. “There are logjams of wood, fish pressed against cypress- tree root bundles, tree trunks smeared with amber.” Most fossils end up being squashed flat by the pressure of the overlying stone, but here everything was three-dimensional, including the fish, having been encased in sediment all at once, which acted as a support. “You see skin, you see dorsal fins literally sticking straight up in the sediments, species new to science,” he said. As he dug, the momentousness of what he had come across slowly dawned on him. If the site was what he hoped, he had made the most important paleontological discovery of the new century.
The thought at the back of one’s mind is always what struck one of the first people to realise what wiped out the dinosaurs: one day, this could easily happen to us. A 300-metre object would end world agriculture.
unique link to this extract
Air passengers at a growing number of US airports will no longer need to remove electronics, liquids, and other items from their carry-on luggage at security checkpoints as the Transportation Security Administration rolls out new technology.
The TSA took a major step in a broader plan to revamp its overall screening process with faster, more advanced technology when it signed a contract Thursday for hundreds of new carry-on baggage screening machines, Administrator David Pekoske said on a press call Friday. The agency has tested the new technology at more than a dozen airports since 2017, along with the relaxed protocols that allow passengers to leave items such as laptops and toiletries inside their luggage.
The rollout of the computed tomography, or CT, machines will begin this summer, Pekoske said. The $97m contract will buy 300 machines, but the list of airports receiving them has yet to be made final, Pekoske said.
The technology creates 3-D images of bags’ contents and will eventually be able to detect items automatically that the TSA now asks passengers to remove, he said.
“It’s not a little bit better, it’s a lot better,” Pekoske said of the technology.
This is going to be introduced over the next eight years – so it’s going to be “do I need to..?” all over the place. By the time it’s everywhere, we’ll only notice the places where it’s slow.
unique link to this extract
[Elina] Berglund won’t divulge many details about [period-tracking/pregnancy likelihood predictor] Natural Cycles’ technology, lest competitors seek to copy it. But she does say it reliably predicts ovulation by taking into account a user’s menstruation dates, fluctuations in her body temperature, and data on the cycles of hundreds of thousands of women. It also adapts to each user: The app will err on the side of caution by showing additional red days when it doesn’t have enough information. The more data a user inputs, the more precise its red day-green day predictions become.
Clinical studies show Natural Cycles is 93% effective at preventing pregnancy with typical use, meaning that after a year, seven women out of 100 users will become pregnant. (With perfect use, Natural Cycles is 99% effective, according to its research.) That puts it about on par with hormonal birth control pills (91%) and beats condoms (82%) and the rhythm method (76%). But it’s less effective than long-acting reversible contraceptives such as intrauterine devices (almost 100%). Even though Natural Cycles wasn’t developed with proponents of so-called natural family planning methods such as the Catholic Church in mind, it’s won praise from those quarters because it isn’t “artificial” birth control that divorces sex from procreation.
However, it then ran slap bang into a PR crisis. Side note: Berglund was on the team at CERN which discovered the Higgs Boson. Now she’s cofounded an app which has several hundred thousand users paying $99 per year. That’s serious money.
unique link to this extract
“This group [of outside people chosen for Google’s external advisory board on AI] will consider some of Google’s most complex challenges that arise under our AI Principles, like facial recognition and fairness in machine learning, providing diverse perspectives to inform our work,” the company said in an announcement. The board, called the Advanced Technology External Advisory Council (ATEAC), included recognized experts in AI research who had worked in the field for years.
But some members of the new board drew immediate scrutiny, especially Kay Coles James, president of the conservative Heritage Foundation. On social media, some characterized the decision as an attempt to cater to conservatives at the expense of true expertise in the field. By Saturday, one AI expert who was invited to the board had dropped out, vaguely noting that it may not be “the right forum” for the work.
Privately, several Google employees were also livid about the decision to include James, according to sources familiar with the discussions. On internal message boards, employees described James as “intolerant” and the Heritage Foundation as “amazingly wrong” in their policies on topics like climate change, immigration, and, particularly, on issues of LGBTQ equality. A person with James’ views, the employees said, “doesn’t deserve a Google-legitimized platform, and certainly doesn’t belong in any conversation about how Google tech should be applied to the world.”
There’s also a Medium petition by Google employees. The Heritage Foundation is the sort of bonkers institution that could only grow up in the US. Why not ask a group that represents minorities or women, since they’ll be at far more risk from any inequity introduced by AI?
unique link to this extract
To assess the impact of a fully deployed AppleNews+ I did the following calculation.
• In the United States, the magazine industry generates annual revenue of $27B, a loss of more than 40% in ten years.
• Divided by 225 million readers of magazines (according to the trade association), the Average Revenue per User (ARPU) amounts to $120 per reader and per year, all sources included.
• The revenue promised by Apple News+ is $9.9 a month => $119 a year. Minus Apple’s 50% cut, it gives a net income per reader of $59.
➜ By joining Apple News+, the US magazine industry will lose 50% of its revenue per reader.
Of course, we are talking of transfer here: magazine readers who will join Apple News+ will inevitably cancel their subscription to its preferred publication. (I will carefully review my personal subscription portfolio that amounts to $1500/year, although my most expensive subs — digital newspapers — won’t be in Apple News+, for a good reason)…
To put it differently, for each magazine reader switching to Apple News+, the platform would need to recruit one additional subscriber, only to preserve the size of the sector. The real uncertainty here is the ability of Apple to nearly double the number of people paying for a magazine in the United States where most subscriptions are already dirt cheap (only 13% of the magazines’ circulation revenue come from digital).
These are pretty brutal numbers, though I think there’s a counter-argument that each subscriber *is* a new subscriber; that most Apple News+ users will be those who haven’t previously subscribed, rather than “churners”.
unique link to this extract
[Rafat Ali commented of News+:] “They’ll just cherry pick what they want via News+, and Apple will shave off a few cents for the publisher while owning all the data, customer relationship and power.”
That would appear to be the primary concern of the two major News+ holdouts: The New York Times and The Washington Post. Apple badly wanted to lock down at least one of them, and it began a vigorous courtship of the papers last spring, not long after the Texture deal closed and Apple’s plans for its content bundle were beginning to materialize, according to people familiar with the matter. “They put a tremendous amount of pressure on,” one source said. “Eddy Cue was in and out of their offices really trying to woo them.” Cue’s elevator pitch, according to people familiar with the discussions, was, “We’ll make you the most-read newspaper in the world.”
In multiple meetings with top brass at both newspapers, Apple made it clear that they wanted the whole shebang, as opposed to a pared down offering or a specialized sliver of stories. “They didn’t want to have limitations in terms of content,” according to a person with knowledge of the talks. But Apple dangled flexible terms governing the duration after which they could pull out, as well as exclusivity. “You’d be protected against a competitor coming in,” the same source said. “If this thing was really successful and everyone else came back to the table, there was a period where you’d have exclusivity.”
But the Times and the Post couldn’t be swayed. Over the past several years, both publications have developed substantial digital subscription businesses that are now vital moneymakers, helping to offset the industry’s advertising collapse. Those businesses continue to grow, and the Times and the Post, put simply, want their own subscribers, not Apple’s subscribers—and they certainly don’t want Apple subscribers if Apple is going to keep a 50% cut of the revenues.
It will still be a few days before I can publish my full review of the P30 Pro, but I spent this past weekend comparing its camera against Google’s Pixel 3 and struggling to believe my eyes. The Pixel 3’s Night Sight mode is algorithmic magic, granting that camera something akin to superhuman night vision. It requires up to six seconds of exposure time, during which you have to hold the phone steady to obtain a sharp image. Huawei has a similar night mode, but I find that completely unnecessary with the P30 Pro: this camera shoots better low-light photos than Google Night Sight without the need for a long exposure.
Let’s dive into some examples. This first one includes the output from the default Google Pixel camera to give you an idea of what the human eye sees. It’s also an accurate representation of what you’ll be able to obtain using an iPhone without the help of either the flash or RAW image processing. Even adapted to the pre-sunrise darkness in the room, my eyes couldn’t discern any color. Google’s Night Sight image is the best, I’m confident in saying, that any smartphone before the P30 Pro could achieve in the circumstances. And the P30 Pro makes that shot look like a splotchy mess.
I’d love to know how Huawei is doing this; one would have thought that camera sensors were pretty much equal everywhere, and that Google was taking it further by its use of AI. But Huawei is pulling in photons that others lose. One for iFixit to answer, at least in part?
I was going to say that sometimes you want a night shot to look like a night shot, but of course you can just darken it in the edit.
unique link to this extract
Rather neat: finds cheap flights from nearby you to various points around the world. The sort of thing that could be enjoyable around the Easter break. Apologies to Australians and New Zealanders, though then again you’ve already got it good.
unique link to this extract
[Skype co-founder Jaan] Tallinn warns that any approach to AI safety will be hard to get right. If an AI is sufficiently smart, it might have a better understanding of the constraints than its creators do. Imagine, he said, “waking up in a prison built by a bunch of blind five-year-olds.” That is what it might be like for a super-intelligent AI that is confined by humans.
The theorist Yudkowsky found evidence this might be true when, starting in 2002, he conducted chat sessions in which he played the role of an AI enclosed in a box, while a rotation of other people played the gatekeeper tasked with keeping the AI in. Three out of five times, Yudkowsky – a mere mortal – says he convinced the gatekeeper to release him. His experiments have not discouraged researchers from trying to design a better box, however.
The researchers that Tallinn funds are pursuing a broad variety of strategies, from the practical to the seemingly far-fetched. Some theorise about boxing AI, either physically, by building an actual structure to contain it, or by programming in limits to what it can do. Others are trying to teach AI to adhere to human values. A few are working on a last-ditch off-switch. One researcher who is delving into all three is mathematician and philosopher Stuart Armstrong at Oxford University’s Future of Humanity Institute, which Tallinn calls “the most interesting place in the universe.” (Tallinn has given FHI more than $310,000.)
Armstrong is one of the few researchers in the world who focuses full-time on AI safety. When I met him for coffee in Oxford, he wore an unbuttoned rugby shirt and had the look of someone who spends his life behind a screen, with a pale face framed by a mess of sandy hair. He peppered his explanations with a disorienting mixture of popular-culture references and math. When I asked him what it might look like to succeed at AI safety, he said: “Have you seen the Lego movie? Everything is awesome.”
The University of California study identified several traits that were associated with being more likely to view the video: being male, Christian and unemployed, watching a lot of TV, having a pre-existing heightened fear of terrorism and having previously being exposed to violence (such as having been the victim of assault or domestic violence, or having lost a loved one to suicide or murder).
Crucially, they found that even two years after the beheading videos went viral, those who watched them were more fearful of future events, including potential terrorist attacks. In this way then, when large numbers of people watch terrorist videos it helps further militants’ central aim: to spread terror.
Authoritarian regimes have long understood that public executions are an effective form of social control because they spread fear, terrorist groups such as Isis have learned that you don’t necessarily have to force people to witness such atrocities – many of us will seek them out.
Sarah Redmond, one of the authors of the report and a PhD student at the University of California, acknowledged that a different demographic might be attracted to graphic footage posted by far-right terrorists, and that we can’t confidently extrapolate much information from the Isis study about the types of people most likely to watch the mosque attacks online. But the study does underline why it’s crucial for internet companies to develop effective ways to block content posted by terrorist groups, or else risk aiding militants.
It also offers lessons for the media: the authors suggest that by publishing screen shots of the beheading and warning that the footage was too graphic to share, the media inadvertently stoked interest in the original footage, the horrifying images working in the manner of a film trailer
Errata, corrigenda and ai no corrida: none notified
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.