Start Up No.1922: will iOS get sideloading?, more on that fusion result, Binance daily outflow tops $1bn, the truth about strikes, and more

Life as the adjunct of a real-estate AI chatbot can make you ask yourself who’s the robot, and who the human. CC-licensed photo by Matthew Hurst 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 11 links for you. Putting more in than out. I’m @charlesarthur on Twitter. Observations and links welcome.

Apple to allow outside app stores in overhaul spurred by EU laws • Bloomberg via Yahoo

Mark Gurman:


Apple is preparing to allow alternative app stores on its iPhones and iPads, part of a sweeping overhaul aimed at complying with strict European Union requirements coming in 2024.

Software engineering and services employees are engaged in a major push to open up key elements of Apple’s platforms, according to people familiar with the efforts. As part of the changes, customers could ultimately download third-party software to their iPhones and iPads without using the company’s App Store, sidestepping Apple’s restrictions and the up-to-30% commission it imposes on payments.

The moves — a reversal of long-held policies — are a response to EU laws aimed at leveling the playing field for third-party developers and improving the digital lives of consumers. For years, regulators and software makers have complained that Apple and Google, which run the two biggest mobile app stores, wield too much power as gatekeepers.

If similar laws are passed in additional countries, Apple’s project could lay the groundwork for other regions, according to the people, who asked not to be identified because the work is private. But the company’s changes are designed initially to just go into effect in Europe.

Even so, the news bolstered shares of companies that offer dating services and other apps. Match Group Inc. jumped as much as 10% and Bumble Inc. was up as much as 8.6% — a sign investors think the companies could get a break from Apple’s commissions.


Will only apply in Europe initially, and Apple is doing it because of the Digital Markets Act, Gurman says. Hell of a scoop. There’s been a feeling that this is inevitable, but now the question is what subtle (or not-so-subtle) obstacles Apple will put in the way of those sideloaders. There’s a lot of money in those 30% fees.
unique link to this extract

A US nuclear fusion test inches us closer to a clean energy holy grail • Fast Company

Alex Pasternack:


The ignition the NIF produced in an experiment last week amounted to 3.15 megajoules of energy, a gain of about 54% over the roughly 2 megajoules that the reaction consumed from the lasers, the lab’s analysis suggests. The test built on a record the lab set last August, in a fusion experiment that yielded over 1.3 megajoules. The Financial Times first reported the breakthrough on Monday, and the Dept. of Energy confirmed it during a press conference on Tuesday.

…The NIF’s approach—intended for weapons research specifically—is a terribly inefficient way of producing electricity. NIF’s laser, the world’s largest, loses up to 99% of its energy in the process of heating up the pellet. In an actual reactor, there’s also the energy lost to waste heat and noise, which typically means a thermal efficiency of less than 50%. To be a viable commercial energy source, fusion reactors must be able to draw significantly more net energy from the reaction, possibly a hundred times more.

Another approach, magnetic confinement confusion, which uses magnetic fields to heat the plasma inside donut-shaped tokamak reactors, likely holds more promise for commercial energy production.


As suspected: it’s encouraging, and shows that the science is solid, but it’s absolutely miles from application. Plus it takes about a day (minimum) to set up each experiment, whereas – as one scientist pointed out – you need to be doing it at least every second (that’s three orders of magnitude faster). There are so many challenges in so many directions.
unique link to this extract

Becoming a chatbot: my life as a real estate AI’s human backup • N+1 Magazine

Laura Preston:


The position was at a company that made artificial intelligence for real estate. They had developed a product called Brenda, a conversational AI that could answer questions about apartment listings. Brenda had been acquired by a larger company that made software for property managers, and now thousands of properties across the country had put her to work.

Brenda, the recruiter told me, was a sophisticated conversationalist, so fluent that most people who encountered her took her to be human.

But like all conversational AIs, she had some shortcomings. She struggled with idioms and didn’t fare well with questions beyond the scope of real estate. To compensate for these flaws, the company was recruiting a team of employees they called the operators. The operators kept vigil over Brenda 24 hours a day. When Brenda went off-script, an operator took over and emulated Brenda’s voice. Ideally, the customer on the other end would not realise the conversation had changed hands, or that they had even been chatting with a bot in the first place. Because Brenda used machine learning to improve her responses, she would pick up on the operators’ language patterns and gradually adopt them as her own.

…Brenda was more efficient than the most industrious human agent. She could cross-reference a vast database of property information in an instant and field messages faster than any human at a keyboard. She could deal with calls at all hours of the day and night, didn’t need a lunch break and could work weekends and holidays. When the leasing agents arrived in the office each morning, their tour schedules were neatly arranged, as if by elves in the night.

Meanwhile, we operators, with our advanced degrees in the humanities, had aptitudes Brenda lacked. We were intuitive, articulate and sensitive to the finer points of delivery. At $25 an hour we also cost almost nothing to employ, by corporate standards. Under the Brenda-operator alliance, everyone came out ahead: the operators got paid better than they would as adjunct professors, and Brenda became more likable, more convincing, more humane. Meanwhile, Brenda’s corporate clients were satisfied knowing they had not replaced their phone lines with a customer-service bot. What they were using, instead, was cutting-edge AI backed by PhDs in literature.


But, but, but. Read all of it: a fantastic description of modern life under the thumb of chatbots, and humans acting like machines.
unique link to this extract

ChatGPT is coming for your job. And I do mean your • Jabberwocking

Kevin Drum:


As good as it is, ChatGPT right now is only a curiosity and a warning. It’s a curiosity because even a modest effort exposes it as an idiot savant, full of on-point facts but not really able to draw sophisticated conclusions from them. It’s a warning because it’s probably only a few years away from having the knowledge and verbal abilities of a PhD student.

How do we respond when that happens? I’m not sure, but I’ll say this: we currently live in a world full of lawyers and professors and journalists who are able to calmly accept the prospect of millions of unemployed truck drivers when AI fills the world with self-driving trucks. But they will probably be a wee bit more upset at the prospect of millions of unemployed lawyers, professors, and journalists.

Maybe this is a good thing. The only way we’ll get a serious response to AI is if either (a) it affects the working class in numbers so big it creates riots, or (b) it reduces the income of the ruling class by 1% or so. Both would be considered problems of about the same magnitude and would provoke roughly the same energy toward finding a solution.


Are lawyers, professors and journalists actually the ruling class? Surely that group consists of the very very rich (if there really is a “ruling” class, tbd) and they won’t notice if the work gets done by ChatGPT or a real person.
unique link to this extract

Binance suffers $1bn outflow in one day as crypto jitters spread • Financial Times

Joshua Oliver and Scott Chipolina:


Changpeng Zhao, chief executive of Binance, said the exchange had experienced $1.14bn in net withdrawals on Tuesday as users continued to remove their assets from his marketplace.

Net withdrawals over seven days had topped $3.6bn, according to an analysis conducted by blockchain research group Nansen.

The outflows underscored the nervousness swirling around crypto markets since the collapse last month of FTX, one of the largest companies in the industry. FTX founder Sam Bankman-Fried was charged with fraud in the US on Tuesday.

Nansen said the rapid withdrawals were the largest since June, when the digital assets market was embroiled in an unprecedented market crash that saw popular tokens plummet in value.

Zhao played down the scale of these redemptions, insisting it was “Business as usual for us.” “Some days we have net withdrawals; some days we have net deposits,” Zhao said on Twitter.

Investors have pulled record levels of bitcoin from crypto exchanges in the past month on concerns over the safety of their assets. Crypto exchanges like Binance take custody of clients’ assets alongside offering a trading venue.

Binance said it has more than $60bn in assets, sufficient funds to honour withdrawals. However, the company’s disclosures do not include its liabilities making it difficult to ascertain its financial health. The exchange told the FT all client deposits are backed by corresponding assets and that its “capital structure is debt free.”


Pretty much nobody believes this apart from the people who have money in Binance. Bitcoin’s price actually rose on Tuesday amid the withdrawals, perhaps because people are getting out of other junkcoins towards a relatively safe haven. The next stage will be if those buyers convert back to fiat, dumping bitcoin.
unique link to this extract

Rise of open-source intelligence tests US spies • WSJ

Warren P. Strobel:


As Russian troops surged toward Ukraine’s border last fall, a small Western intelligence unit swung into action, tracking signs Moscow was preparing to invade. It drew up escape routes for its people and wrote twice-daily intelligence reports.

The unit drafted and sent to its leaders an assessment on Feb. 16, 2022, that would be eerily prescient: Russia, it said, would likely invade Ukraine on Feb. 23, US East Coast time. 

The intelligence shop had just eight analysts and used only publicly available information, not spy satellites and secret agents. It belonged to multinational chemicals company Dow Inc., not to any government.

“I’m leading an intelligence center that accurately predicted the invasion of Ukraine without any access to sensitive sources,” said John Robert, Dow’s director of global intelligence and protection, whose unit helps the company manage business risk and employee safety.

Supercharged by the Ukraine war, the rise of open-source intelligence, or OSINT, which comprises everything from commercial satellite imagery to social-media posts and purchasable databases, poses revolutionary challenges for the Central Intelligence Agency and its sister spy agencies, according to former senior officials who spent decades working in those agencies’ classified spaces.

Dow is just one of a fast-growing number of companies, nonprofit groups and countries transforming publicly available data into intelligence for strategic and economic advantage. China has the largest, most focused effort, while US spy agencies, with deeply ingrained habits of operating in the shadows, have been slow to adapt to a world in which much of what is important isn’t secret, according to dozens of officials and many studies.

The CIA is simultaneously dealing with a closely related challenge: It is pivoting from two decades focused on terrorism toward spying on a new primary intelligence target, China. But some officials say the technological tsunami facing US intelligence agencies poses a more fundamental challenge than merely swapping priorities.

…But by some estimates, more than 80% of what a U.S. president or military commander needs to know comes from OSINT, and not from foreign agents, spy satellites or expensive eavesdropping platforms.


OK but it’s that other 20% which is where the CIA and the rest earn their money, isn’t it?
unique link to this extract

Meta kills Facebook Connectivity • Light Reading

Mike Dano:


Meta Connectivity, which launched in 2013, sought to develop innovative connection technologies – from solar-powered drones and fibre-laying robots to low-Earth orbit satellites – in order to extend the company’s social network to more users.

Last year, the company estimated that more than 300 million people got access to faster Internet services thanks to Meta Connectivity.

The closure of Meta Connectivity stems from the massive round of layoffs Meta’s Mark Zuckerburg announced earlier this month. He said he would reduce the size of the company by roughly 13%, laying off more than 11,000 employees. Meta has struggled to focus on metaverse products while its core social networking advertising business faces threats from the likes of Apple and others.


This has the feeling of a company realising that it’s not actually going to defeat the laws of physics. Solar-powered drones lose out to simpler solutions such as solar-powered mobile phone masts. Sure, Musk’s Starlink is still pushing low-Earth orbit satellites for connections, but let’s see what there is left 10 years from now.
unique link to this extract

The biggest myths about this week’s strikes in the UK • New Statesman

Anoosh Chakelian:


Every time workers threaten to go on strike in the UK, a little ritual ensues. The average wage of the sector in question is googled (just look at how searches for “rail salary” and “train driver salary” spiked when the rail strikes began in June). The googler in question – possibly based on how this compares with their own income, whether strikes inconvenience them personally, plus a dash of “could I do that job myself?” – then decides whether or not the industrial action is justified.

This psychological process, in addition to how government ministers frame trade union leaders, all serves to shape public sentiment towards strikes.

One of the biggest and most basic public misconceptions in relation to strikes is about average salaries. The average annual pay in the UK is £33,000. But we, as the British public, tend to assume it’s actually lower. When the New Statesman asked British voters what they thought the average salary was, we found a third of them pegged it to £20,001-£30,000 – the most chosen salary bracket in the poll.

This may sound like a minor underestimation, but it means we therefore assume certain workers are better paid (or paid closer to the average) than they actually are.


This begins the problem, yet what’s noticeable about the strikes happening in the UK this month (and there are a lot) is that they have broad public support. As much as anything, that’s because people also don’t buy the argument that public sector pay awards will drive inflation (which is anyway untrue; Chakelian explains why).
unique link to this extract

The unbearable lightness of hydrogen • Bloomberg New Energy Futures

Michael Liebreich:


While clean hydrogen will be needed to decarbonize a number of use cases in industry, and perhaps for long-duration storage, I found it hard to identify any role for it in applications like land transportation or space heating. Since then, as I have done more work on industrial heat, I have even come to believe it has a limited role even there.

If my intention at the time was to inject some reality into discussions about hydrogen, I clearly failed. Rhetoric around hydrogen has become ever more overblown.

According to lobbying group the Hydrogen Council, citing a series of reports commissioned from McKinsey over the past three years, hydrogen can be expected to contribute more than 20% of emissions reductions needed for the world to reach net-zero emissions – a figure repeated by politicians and journalists seemingly without the slightest critical examination.


No critical examination? Incredible!


In October this year the Hydrogen Council and McKinsey released another report entitled Global Hydrogen Flows, predicting long-distance transport of 400m tonnes of clean hydrogen and its derivatives (calculated on a hydrogen-content basis) by 2050, out of total global production of 660m tonnes of hydrogen. It is worth bearing in mind that today, 94m tonnes of hydrogen are used annually, virtually all of it made from fossil fuels, creating 2.3% of global emissions. The vast bulk of today’s hydrogen never leaves the compound on which it is made, let alone cross an international border.


I mean, it might change, but the inertia against such dramatic reconstitution is colossal.
unique link to this extract

Former members of Twitter’s safety council voice concerns over Musk’s acquisition • NPR


[Eirliani] Rahman [who resigned from Twitter’s third-party safety council]: in terms of average number of tweets per day, in the first two weeks, antisemitic posts went up by 61% – against gay men, the corresponding number is 58%. I find that highly unacceptable.

Rachel Martin, NPR: And all this data you’re quoting is in the time period since Musk took over?

Rahman: Completely correct. And these are the data that my fellow former peers put forward. And they were part – they are part of the council and still are in there. And the other red line that he crossed was when previously banned accounts were reinstated – so for example, the ones that led up to what happened January 6 here in the US. So for me, all these were highly, highly problematic. We were hoping that our – with our resignations, it would prompt a rethinking within Twitter, within the council, but also just generally within Twitter headquarters to reconsider what’s happening to the content moderation and to make it a safer space for the public.

Martin: Anne, in your resignation letter, you said Twitter is moving toward automated content moderation. Why is that risky in your view?

Collier: Content moderation is very complex and highly nuanced. It’s also very contextual. It’s very, very hard for algorithms to determine what truly is harmful without any context whatsoever. Human beings are needed to do that. And we know that Twitter staff is massively reduced. And Twitter has to be reliant on automated content moderation more. So – and that’s an announcement directly from Twitter itself.


unique link to this extract

The Trumpification of Elon Musk • Wired

Gideon Lichfield:


The point is that the focus on Musk is a mistake. Arguably not as much of a mistake as it was with Trump; an owner-CEO has more power over their company than a president does over their country. But trying to report on what’s happening by expecting either his abject failure or resounding success and then using his most attention-grabbing tactics as evidence for that thesis is not doing anyone a service. 

As with Trump, the real story is often what’s going on below the level of newsmaker in chief. It’s about the actual numbers around Twitter’s advertising, not Musk’s claims that advertisers are coming back. It’s about who’s actually joining and leaving Twitter, not about who’s threatening to leave. It’s about Twitter’s role in the world—its importance to natural-disaster management or to any number of communities for whom it’s a store of social wealth—rather than just how much money it will lose. Musk and Trump subvert the ability to focus on such nuances by making the story all about themselves. The very same tactic that draws their fans ensnares their critics. And we, by which I mean everybody, but especially the media, fall for it every time.

Just before Musk bought Twitter, I tweeted a prediction that “not much [will] change. Trump et al will come back, trolling will increase somewhat, rest of us will block and mute more and engage less but still use it for publishing—more web 1.0, less 2.0.” As foolish as it is to make predictions, and as crazy as the past six weeks have been, I still think this is as plausible a long-term outcome as any other. It’s neither the destruction of Twitter nor a turnaround, but a bet that the platform is too important to too many people to disappear altogether and will hobble along, however dysfunctionally, in some form. This prediction could be utterly wrong, but its chief quality is that it’s boring. People should make boring predictions more often.


Absolutely this. Though there has been a definite disengagement just in the past week by some significant (to me) accounts; once the air starts to go out of the balloon, you can’t get it back in.
unique link to this extract

• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?

Read Social Warming, my latest book, and find answers – and more.

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.