With renewables and microgeneration on the rise, Google reckons power grids need a “moonshot” to inject machine learning and more for future systems. Can it succeed? CC-licensed photo by sagesolar 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 9 links for you. Friend or foe? I’m @charlesarthur on Twitter. Observations and links welcome.
|• Why did so many news sites “pivot to video” in 2015?
• Would online discourse be nicer if algorithms weren’t picking content?
• What sort of content do we really respond to?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?
Preorder Social Warming, my forthcoming book, and find answers – and more.
Mike Isaac and Jack Nicas:
At the center of the fight are the two C.E.O.s. Their differences have long been evident. Mr. Cook, 60, is a polished executive who rose through Apple’s ranks by constructing efficient supply chains. Mr. Zuckerberg, 36, is a Harvard dropout who built a social-media empire with an anything-goes stance toward free speech.
Those contrasts have widened with their deeply divergent visions for the digital future. Mr. Cook wants people to pay a premium — often to Apple — for a safer, more private version of the internet. It is a strategy that keeps Apple firmly in control. But Mr. Zuckerberg champions an “open” internet where services like Facebook are effectively free. In that scenario, advertisers foot the bill.
The relationship between the chief executives has become increasingly chilly, people familiar with the men said. While Mr. Zuckerberg once took walks and dined with Steve Jobs, Apple’s late co-founder, he does not do so with Mr. Cook. Mr. Cook regularly met with Larry Page, Google’s co-founder, but he and Mr. Zuckerberg see each other infrequently at events like the Allen & Company conference, these people said.
The executives have also jabbed at each other. In 2017, a Washington political firm funded by Facebook and other Apple rivals published anonymous articles criticizing Mr. Cook and created a false campaign to draft him as a presidential candidate, presumably to upend his relationship with former President Donald J. Trump. And when Mr. Cook was asked by MSNBC in 2018 how he would deal with Facebook’s privacy issues if he was in Mr. Zuckerberg’s shoes, he replied, “I wouldn’t be in this situation.”
Very much the Irish saying of “I wouldn’t start from here”. Notable how the headline assumes everyone knows who Zuckerberg is, but not Cook. (That is, the title on the web page; the headline on the page does say Cook.)
unique link to this extract
Apple started rolling out the latest version of iOS (14.5) on Monday, which could become a watershed moment in the history of privacy on the internet. The new version of iOS makes tracking an opt-in feature. In other words, when you install an app on your iPhone, you will have to allow it to track you across other apps, and if you want, you can stop it from tracking you entirely.
This is the most aggressive pro-privacy step Apple has taken in years. Facebook hates it so much that it has launched a full-on PR campaign claiming this change will hurt small businesses.
For many people on the internet, however, this is actually good news. Ashkan Soltani, an independent privacy and security researcher, told Motherboard that this is “a great step, particularly the global option to block tracking via one step.”
“It’s in line with the Global Privacy Control project that I’ve been working on, and is required under the California Law,” he said in an online chat. “In addition to technical mitigations, I’m looking forward to Apple supporting the legal right for consumers to opt-out so simply.”
“Apple today is turning on enhanced privacy/transparency in apps. This is a great day for European policy ideas for cookie pop-up boxes, now deployed to millions of smartphones all around the world,” Lukasz Olejnik, a security and privacy researcher, wrote on Twitter.
The feature is called App Tracking Transparency and in practice, it means that when you open apps you will be greeted with a new pop-up: “Allow [App Name] to track your activity across other companies’ apps and websites?” The first choice is “Ask App Not to Track” and the second one is “Allow.”
Choosing “Ask App not to Track” makes Apple disable the app from using our Apple device identifier, “a random string of letters and numbers assigned to our iPhones and that is used to track our activities across apps and websites,” and lets the app developer know that you don’t want them to track you in any way
Astro Teller runs Google X, the “moonshot factory”:
Grid operators ensure that the supply of electricity to the grid stays in balance with the amount being used every moment of every day. One of the biggest challenges they face is that they have to manage and make decisions without full visibility into all the new and intermittent power sources coming onto the grid, like wind turbines or solar panels. And they don’t have visibility into how energy is flowing in real time: there is no global map or end-to-end aggregated view that gives every operator a consistent, full view of what’s happening on the grid from power plants down to the solar panels on your roof.
What you see happening on the grid depends on who’s looking and from what vantage point. Everyone who manages, builds, regulates and provides electricity to our power system — from utilities to system operators — uses their own sets of tools and models of the system. And this is becoming a huge hindrance as operators are now orchestrating novel and unpredictable flows of power from billions of new devices both contributing to and drawing from the grid.
Finding a way to create a single view of this splintered system felt to us like a promising entry point for reimagining things. We think machine learning, artificial intelligence and advanced computing tools can help; in the last decade these technologies have powered the creation of virtual representations of real-world environments that act as super-efficient simulated testing grounds for new ideas and software.
“Moonshot factory” is quite the boast, and not really supported by the facts. Google X has tried lots of things and you’d be pushing it to say any has even achieved orbit. This is a big idea, but DeepMind has effectively pulled out of this space already. That’s not encouraging.
unique link to this extract
Fuels from thin air: Prometheus joins the chase to make captured CO2 into net zero hydrocarbon fuels • Biofuels Digest
What is it? Prometheus removes CO2 from the air and turns it into zero-net carbon gasoline and jet fuel at a price that will compete with conventional fossil fuels used in transportation today. Because only renewable energy sources are used in its production, you get net zero carbon fuels.
How does it work? BMW reports that the process uses a solution of liquid water and CO2 that is exposed to an electrified copper plate. This catalyzes a reaction and produces fuel alcohols (mostly ethanol). Closely packed filters made from cylindrical carbon nanotubes embedded in plastic allow ethanol through while blocking water molecules. From there, the more concentrated solution of approximately 95% ethanol can be catalyzed with zeolite to join into more complex hydrocarbons, including gasoline, diesel, or jet fuel. This technique works at room temperature, while traditional methods of extraction require heat to distill it from a solution.
Founder Rob McGinnis speculated that even though the theoretical efficiency of Prometheus’ system was only 50–60%, their less energy-intensive process could nevertheless considerably lower overall cost and be competitive with fossil fuels.
Why is it a big deal? As BMW notes, “The modularity of the approach will enable micro-cells of gasoline production where there is a surplus of renewable energy available.”
What’s difficult here? For one, the CO2 processing — it’s aiming to use atmospheric CO2 not point source. That’s tougher. So, the company has pioneered a DAC technology, that’s short for Direct Air Capture. You have to process about 1600 pounds of air to capture a pound of CO2, so having a very passive process is a must.
As BMW noted that the “salvaged CO2 encounters renewable electricity in an electrochemical stack called the Faraday Reactor. The electricity “charges” the carbon with hydrogen molecules from the water to create long-chain alcohols, releasing pure oxygen.
Petrol contains around 2.3-2.6 kilos of carbon (dioxide? Perhaps) per litre. So this thing has to process 1600 x 2.4 kg = 3.8 tonnes of air to capture the equivalent amount of carbon for a single litre of petrol. Air weighs 1.29kg per cubic metre, so there’s about 3.8kg in 3 cubic metres, hence 3.8 tonnes of air is 3,000 cubic metres. An Olympic swimming pool (50m, 25m wide, 2-3m deep) is 2,500 cubic metres. 120% of that to produce one litre of not-yet petrol.
Let’s say it: that’s insane. Even if it were completely passive, it would still make more sense just to capture the carbon and sequester it, and use the power for electric cars. They’re actually more energy-efficient. (New working hypothesis: things called “Prometheus” are junk.)
unique link to this extract
Ben Smith on how the NY Post is aggrieved that Facebook is blocking sharing of a post because it mentions a house, on which basis Facebook has a policy that the person mentioned can block it:
What Facebook’s clash with The Post really revealed — and what surprised me — is that the platform does not defer, at all, to news organizations on questions of news judgment. A decision by The Post, or The New York Times, that someone’s personal wealth is newsworthy carries no weight in the company’s opaque enforcement mechanisms. Nor, Facebook’s lawyer said, does a more nebulous and reasonable human judgment that the country has felt on edge for the last year and that a Black activist’s concern for her own safety was justified. (The activist didn’t respond to my inquiry but, in an Instagram post, called the reporting on her personal finances “doxxing” and a “tactic of terror.”)
The point of Facebook’s bureaucracy is to replace human judgment with a kind of strict corporate law. “The policy in this case prioritizes safety and privacy, and this enforcement shows how difficult these trade-offs can be,” the company’s vice president for communications, Tucker Bounds, said. “To help us understand if our policies are in the right place, we are referring the policy to the Oversight Board.”
The board is a promising kind of supercourt that has yet to set much meaningful policy. So this rule could eventually change. (Get your stories deleted while you can!)
For now, though, the deletion seems to be an instance of how the company finds itself constantly debating the literal interpretation of its own, made-up rules rather than exercising any form of actual judgment. That came up again this spring in an internal report finding that Facebook hadn’t cracked down on “Stop the Steal” splinter groups because they were all hovering below its “violation threshold.”
I should note that when it comes to the article about the activist’s house, Facebook waded into one of the trickiest areas of online speech, and one of the hardest calls for news organizations today.
Poor Facebook, forced to make decisions over policies that it dreamt up itself. Reddit has a similar policy (see No.3), which it enforces effectively: people who doxx someone (give out their personal details) get banned. Facebook could do that, of course.
Privacy concerns, potential discriminatory categorization of people and data control have some publishers including The Guardian joining web browsers in blocking Google’s cookieless tracking and ad targeting method, FLoC. Meanwhile, The New York Times is among publishers open to testing FLoC, or Federated Learning of Cohorts, a method that categorizes groups of people based on their website visits and enables ad targeting and measurement in aggregate, rather than at an individual level. But as contributors to WordPress, which operates the foundation for millions of websites, also have mulled disabling FLoC, and as European regulators delay trials there, the anti-FLoC chorus grows louder.
“We’ve decided to opt-out of the FLOC trial, for the moment, as we assess the commercial and privacy implications of the technology,” a Guardian News and Media spokesperson told Digiday. “As we learn more about the technology, we may seek to move to trial in the future, but we are also testing a range of other privacy-respecting identity solutions, as we seek to build out our future advertising strategy.”
Google declined to comment on the record for this story.
…headwinds billow against the increasingly controversial FLoC technique. Facing concerns regarding compliance with data use and privacy restrictions in Europe, Google revealed in March that it will not make FLoC available for testing in countries where the General Data Protection Regulation and ePrivacy Directive are in place.
Privacy is suddenly quite a thing this year, isn’t it? Almost as if some sort of tipping point had been reached.
unique link to this extract
Car keys are just one implementation of the Apple U1 chip’s UWB [ultrawideband wireless]. Apple offered us plenty more in a series of patents it filed towards the end of 2019, as reported by Patently Apple.
Diagrams in these filings show a series of Apple U1 trackers used to model the articulation of someone’s body in front of a TV hooked up to some future version of the Apple TV box. Exercise platform Apple Fitness+ is already here, so how about using UWB to tell when you are performing yoga moves correctly in a virtual class, using a virtual on-screen avatar?
Another suggested application is to use UWB to find the fire extinguisher, fire exit or defibrillator in a building, presumably a hospital. Apple has already made major in-roads into healthcare organisations with Apple Health — particularly in the US — so why wouldn’t it work its way deeper into their infrastructure?
The patent filing also details the use of multiple UWB sensors for posture analysis, with Apple U1 sensors dotted across the spine, arms, legs and head.
The contents of patents can never be taken as a guarantee something is going to be released, or is even actively being seriously developed. Some of these sound like a headache, and we don’t fancy clipping a half-dozen sensors on for a yoga class, but the idea Apple U1 is not really for finding things is not our own. And these patents do lay out AirTag use more-or-less exactly as Apple announced at its launch event.
There are more obvious applications that may have already come to mind as you read this, too. Apple U1 devices could be used as objects or play pieces in an augmented reality game or app you play using Apple AR Glasses.
Take that idea further. An Apple U1 chip could sit in every shop shelf, identifying the products that sit on them. Ask Siri where to find the carrots, or a Hermès AirTag holder, and your Apple Glasses highlight them in your view, appearing to see through walls, showing stock levels and the price.
They could be used to map out the play field for virtual reality, AirTags marking out its limits at the corners of the room. HTC Vive uses dedicated laser-tracking boxes to similar effect, at far greater expense.
Your robot vacuum cleaner need never lose track of its home base again with the help of an Apple U1 chip. It could allow for ultimate energy saving home automation, turning off all appliances in rooms in which there are no beating hearts — assuming everyone in your family is Apple-obsessed enough to own an Apple Watch.
It’s nice to dream, but Apple isn’t in to “every shop shelf”. It’s into “let the shops stick UWB chips on their shelves and we’ll leverage what they do”. It’s hard to see it making a robovac either – not computer-y enough, and too crowded a market. It hasn’t even put the U1 chip in its new TV remote. Change is slow.
unique link to this extract
Renee DiResta and Matt DeButts:
If Facebook’s algorithms didn’t relentlessly promote Trump’s messages, they might not have gone viral, or at least not with the same alacrity. Just look at the past few months: by blocking Trump from their platforms, Facebook, Twitter, and other social media giants appear to have significantly reduced the media coverage about him. If social media made Trump newsworthy, we now know, too, that it can make him less so.
To escape from Facebook’s circular logic of “newsworthiness,” we have two choices: we can either redefine the term, or we can rethink how it is deployed as a justification in content moderation. The first option is tricky. Redefining a widely used term is hard, and would likely encounter resistance. Free speech advocates are understandably reluctant to enable platforms (or Facebook’s Oversight Board, for that matter) to determine what is in “the public interest”—a phrase that often appears alongside “newsworthy.” Many are quick to argue that the public interest is highly subjective or, alternately, that a large profit-seeking corporation may not be appropriately positioned to evaluate it. When the courts have been asked to define “legitimate public interest” in the context of libel cases, they have largely demurred, preferring to rely on editors’ judgments and individual context. Unfortunately, that leaves us where we started. The second option—rethinking newsworthiness in content moderation—is more intriguing.
One possibility is to reverse Facebook’s current position. Consider the following thought experiment: A Facebook user writes something to his audience of ten million. What he says is borderline harmful, though it’s not a clear-cut case. Still, the speech is likely to be newsworthy—in our viral age, when ten million people begin talking about something it will probably become “news”—so the potential damage is high. Furthermore, once this speech has reached ten million accounts, it will become increasingly difficult to remove the message should it prove dangerous in the end. If another Facebook user, this person with an audience of ten, writes the same sentence, the speech is no different, but the potential harm is limited. If the message were to incite violence, the scale would be much more restricted, and Facebook would have an opportunity to intervene before it goes viral. In the latter case, someone’s lack of newsworthiness should make the content more permissible, but in the current enforcement structure, paradoxically, the ordinary person with the small audience is more likely to be moderated and removed.
Which is the point: exceptions are made for the big accounts. Because they create “engagement” and hence revenue.
unique link to this extract
Roku says Google is threatening the removal of YouTube TV to force Roku to grant preferential access to its consumer data moving forward.
It says Google has asked Roku to do things that it does not see replicated on other streaming competitors’ platforms, like creating a dedicated search results row for YouTube within the Roku smart TV interface and giving YouTube search results more prominent placement.
Roku says Google has also required it to block search results from other streaming content providers while users are using the YouTube app on Roku’s system. Roku alleges Google has asked it to favor YouTube music results from voice commands made on the Roku remote while the YouTube app is open, even if the user’s music preference is set to default to another music app, like Pandora.
Roku says Google has threatened to require Roku to use certain chip sets or memory cards that would force Roku to increase the price of its hardware product, which competes directly with Google’s Chromecast.
In response to the allegations, a YouTube TV spokesperson says, “We have been working with Roku in good faith to reach an agreement that benefits our viewers and their customers.”
“Unfortunately, Roku often engages in these types of tactics in their negotiations. We’re disappointed that they chose to make baseless claims while we continue our ongoing negotiations.”
Those Google denials aren’t very denial-y. I haven’t seen Roku making claims like that in the past, either. Who gets hurt more if there’s no YouTube on Roku? Roku isn’t bigger than all the other smart TVs. But Roku users are unlikely to abandon their investment in it, which could include paid channels. So Google loses out. A bit. As does Roku.
unique link to this extract
Errata, corrigenda and ai no corrida: Me, on the iTunes “Buy doesn’t mean you own it” row: “amazing concept that ‘buying’ something doesn’t mean you have possession of it.
Alex Barredo: “NFTs :-P”