Start Up No.2181: Apple fined €1.84bn after Spotify complaint, AI chatbots give bad tax advice, 4K TV too good?, and more


A group of British scientists have discovered the secret of getting media coverage by referencing popular film topics. CC-licensed photo by Hervé Simon 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.


There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.


A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.


Apple fined €1.84bn in one of Europe’s largest antitrust actions • WSJ via MSN

Kim Mackrael:

»

The European Commission, the EU’s executive body, said it found the company violated antitrust rules by restricting app developers from telling users about alternative ways to subscribe to music-streaming services. The commission said it ordered Apple to change its practices.

“Apple’s conduct, which lasted for almost 10 years, may have led many iOS users to pay significantly higher prices for music streaming subscriptions,” the commission said Monday.

Apple said it plans to appeal the decision, which it said was reached “despite the Commission’s failure to uncover any credible evidence of consumer harm.”

Monday’s fine is the culmination of a multiyear investigation into Apple’s App Store practices and represents one of the largest antitrust penalties ever imposed by the Commission against a single company. Google has faced larger fines—of €4.3bn and €2.4bn—in two separate cases that the tech company has appealed.

Apple’s fine of €1.84bn, equivalent to about $2bn, was larger than some antitrust lawyers had anticipated. The EU’s guidance for calculating an antitrust fine allows it to increase the baseline calculation for what the fine should be to deter a company from its behaviour.

“I think it’s important to say that if you are a company who’s dominant, and you do something illegal, you will be punished,” said Margrethe Vestager, the EU’s competition czar. The size of the fine should help demonstrate the bloc’s resolve in tackling anticompetitive behavior, she added.

«

Obviously this follows from Spotify’s complaint about not being able to tell people in the Spotify app about the option of subscribing in Spotify itself. Apple’s response of not being able to show consumer harm is sort of true, but also irrelevant. One can calculate the counterfactual where people could be told that they’d pay less by following a link to the Spotify site: there are plentiful well-paid experts around Brussels who make a good living working out the numbers for hypotheticals like that.
unique link to this extract


Scientists unearth mysteries of giant, moving Moroccan star dune • The Guardian

Steven Morris:

»

They are impressive, mysterious structures that loom out of deserts on the Earth and are also found on Mars and on Saturn’s biggest moon, Titan.

Experts from universities including Aberystwyth in Wales have now pinpointed the age of a star dune in a remote area of Morocco and uncovered details about its formation and how it moves across the desert.

Prof Geoff Duller of the department of geography and earth sciences at Aberystwyth said: “They are extraordinary things, one of the natural wonders of the world. From the ground they look like pyramids but from the air you see a peak and radiating off it in three or four directions these arms that make them look like stars.”

The team, which was also made up of University of London academics, travelled to the south-east of Morocco to study a 100-metre high and 700-metre wide dune in the Erg Chebbi sand sea known as Lala Lallia, which means the “highest sacred point” in the Berber language.

«

This item has been on radio, TV and other media all day, and I wondered why, because it barely makes any sense and it’s about something that I’ve never worried about and didn’t even when I read it. (Still don’t.) Then someone pointed out that it contains the word “dune”, same as a big film that’s just been released. Scientists timing the announcement about their work to catch some hot SEO? Why of course. That’s much more fun to observe.
unique link to this extract


TurboTax and H&R Block’s AI chatbots are giving bad tax advice • The Washington Post

Geoffrey Fowler:

»

This year, TurboTax and H&R Block added artificial intelligence to the tax-prep software used by millions of us. Now while you’re doing your taxes online, there are AI chatbots on the right side of the screen to answer your burning questions.

But watch out: Rely on either AI for even lightly challenging tax questions, and you could end up confused. Or maybe even audited.

Here’s one example: Where should your child file taxes if she goes to college out of state? When I asked, TurboTax’s “Intuit Assist” bot offered irrelevant advice about tax credits and extensions. H&R Block’s “AI Tax Assist” bot gave me the wrong impression she has to file in both places. (The correct answer: she only files in the other state if she has earned income there.)

Question after question, I got many of the same random, misleading or inaccurate AI answers.

…After I shared my results with TurboTax maker Intuit, the company changed some of how the bot picks its answers. But its new version of Intuit Assist was still unhelpful on a quarter of the questions.

H&R Block’s AI gave unhelpful answers to more than 30% of the questions. It did well on 529 plans and mortgage deductions, but confidently recommended an incorrect filing status and erroneously described IRS guidance on cryptocurrency.

“I feel that my job as a tax professional is very secure,” said Beverly Goodman​​​​, a tax manager at EP Wealth who helped me analyze the AI advice.

«

This was so predictable. So very predictable.
unique link to this extract


Here’s the full AI-generated script from the Willy Wonka disaster • Gizmodo

Thomas Germain:

»

An event based on Willy Wonka and the Chocolate Factory made international news over the weekend after a promised world of imagination turned into a full on disaster. “Willy’s Chocolate Experience” in Glasgow, Scotland was promoted with elaborate AI-generated images of lollipop forests and jellybean waterfalls. But when families arrived, they were greeted by a filthy, barely decorated warehouse, prompting parents to call the police (see the photos here). Now, Gizmodo has a copy of the event’s unhinged AI-generated script.

The script was shared in a Facebook group organized after the event called “House of Illuminati Scam,” named after the company behind the production. An actress named Cara Lewis posted the document, saying actors were given two days to memorize it and then told to abandon the text and improvise as the fiasco unfolded.

Gizmodo reached out to Lewis and a number of the other actors but didn’t hear back, and with no response from House of Illuminati, we can’t fully guarantee the script’s authenticity. However, Lewis was clearly present at Willy’s Chocolate Experience, and the script matches descriptions from other actors and people who attended the event. Based on our reporting, it seems like the real thing.

The script has all the hallmarks of AI, including the nonsensical decision to include lines for audience members and descriptions of the crowd’s reaction, as though it’s happening in real-time. You’ll also notice the code names for Willy Wonka and the Oompa Loompas. House of Illuminati said on its website that any resemblance to existing characters is “coincidental” and the event is unrelated to the copyrighted Willy Wonka and the Chocolate Factory.

«

AI is the warming water, and we’re the frogs.
unique link to this extract


A WordPress ‘firehose’ allows AI companies to buy access to a million posts a day • 404 Media

Jason Koebler and Samantha Cole:

»

In September 2023, WordPress.com quietly changed the language of a developer page explaining how to access a “Firehose” of roughly a million daily WordPress posts to add that the feeds are “intended for partners like search engines, artificial intelligence (AI) products and market intelligence providers who would like to ingest a real-time stream of new content from a wide spectrum of publishers.” Before then, this page did not note the AI use case. 

This is notable because of the fervor and confusion that has arisen this week after we broke the news that Automattic, which owns WordPress.com and Tumblr, was preparing to send user data to OpenAI and Midjourney. Since then, there has been much discussion about which WordPress blogs would be included, which would not, whether data was already sent, and whether people who opt out would have their data redacted retroactively. 

We still do not know the answers to all of these questions, because Automattic has repeatedly ignored our detailed questions, will not get on the phone with us, and has instead chosen to frame a new opt-out feature as “protecting user choice.”

Update: After this article was published, Automattic told 404 Media that it is “deprecating” the Firehose: “SocialGist is rolling off as a firehose customer this month and the remaining customers are winding down in the coming months, both things that were already in motion for different reasons,” an Automattic spokesperson said. “We’re in the process of updating our developer page to indicate that we have been deprecating the old firehose for several months.” The spokesperson did not answer the original questions we posed to them about the data supply chain for the Firehose.

«

Cool – so all this nonsense of mine is getting indexed? I’m going to be immortal? (Though to quote Woody Allen: “I don’t want to be immortal through my works, I want to be immortal through not dying.”)
unique link to this extract


Your TV is too good for you • The Atlantic

Ian Bogost:

»

Last fall, when Netflix hiked the cost of its top-tier Ultra HD plan by 15%, I had finally had enough: $22.99 a month just felt like too much for the ability to see Jaws in 4K video resolution. A couple of weeks later, I heard that Max was pushing up the fee of its own 4K streaming by 25%. Now I wasn’t just annoyed, but confused. Super-high-res televisions are firmly ensconced as the next standard for home viewing of TV and movies. And yet, super-high-res content seems to be receding ever further into a specialty consumer niche. What happened?

4K certainly is ubiquitous; you won’t find many sets with lower resolution for sale at Best Buy. In practice, though, the technology is rarely used. Cable signals are generally mere HD, as are the standard plans on most streaming services. And the fancy new displays, as they’re placed and viewed in people’s homes, may never end up looking any sharper than the old ones, no matter what Netflix plan you have. In short, the ultra-high-definition future for TV has turned out to be a lie.

A relentless narrative of progress brought us to this point, but it did not begin in 2012, when the first 4K televisions were brought to market at roughly the price of a Honda Accord. Rather it extends back into the early days of TV, with the idea that picture quality can and always will be improved: first with the introduction of color sets, then with bigger screens, then with added pixels.

But sometimes progress ends. The peak of television-picture quality, as actually seen by TV viewers, was reached 15 years ago, and we’ve been coasting ever since. Forget the cable signals and the streaming plans. Most people just can’t sit close enough to today’s televisions to make full use of their picture.

«

But.. but the Vision Pro! Which maybe goes up to 8K, or something comparable. No problem sitting close enough to that.
unique link to this extract


Fury after Exxon chief says public to blame for climate failures • The Guardian

Dharna Noor and Oliver Milman:

»

The world is off track to meet its climate goals and the public is to blame, Darren Woods, chief executive of oil giant ExxonMobil, has claimed – prompting a backlash from climate experts.

As the world’s largest investor-owned oil company, Exxon is among the top contributors to global planet-heating greenhouse gas emissions. But in an interview, published on Tuesday, Woods argued that big oil is not primarily responsible for the climate crisis.

The real issue, Woods said, is that the clean-energy transition may prove too expensive for consumers’ liking.

“The dirty secret nobody talks about is how much all this is going to cost and who’s willing to pay for it,” he told Fortune last week. “The people who are generating those emissions need to be aware of and pay the price for generating those emissions. That is ultimately how you solve the problem.”

Woods said the world was “not on the path” to cut its planet-heating emissions to net zero by 2050, which scientists say is imperative to avoid catastrophic impacts of global heating. “When are people going to willing to pay for carbon reduction?” said Woods, who has been Exxon’s chief executive since 2017.

“We have opportunities to make fuels with lower carbon in it, but people aren’t willing to spend the money to do that.”

«

Yes, definitely our fault that ExxonMobil chose not to invest in renewable energy decades ago and drive the prices down.
unique link to this extract


How the Pentagon learned to use targeted ads to find its targets—and Vladimir Putin • WIRED

Byron Tau:

»

As he would explain in a succession of bland government conference rooms, [US government contractor and technology Mike] Yeagley was able [in 2019] to access the geolocation data on Grindr users through a hidden but ubiquitous entry point: the digital advertising exchanges that serve up the little digital banner ads along the top of Grindr and nearly every other ad-supported mobile app and website. This was possible because of the way online ad space is sold, through near-instantaneous auctions in a process called real-time bidding. Those auctions were rife with surveillance potential. You know that ad that seems to follow you around the internet? It’s tracking you in more ways than one. In some cases, it’s making your precise location available in near-real time to both advertisers and people like Mike Yeagley, who specialized in obtaining unique data sets for government agencies.

Working with Grindr data, Yeagley began drawing geofences—creating virtual boundaries in geographical data sets—around buildings belonging to government agencies that do national security work. That allowed Yeagley to see what phones were in certain buildings at certain times, and where they went afterwards. He was looking for phones belonging to Grindr users who spent their daytime hours at government office buildings. If the device spent most workdays at the Pentagon, the FBI headquarters, or the National Geospatial-Intelligence Agency building at Fort Belvoir, for example, there was a good chance its owner worked for one of those agencies. Then he started looking at the movement of those phones through the Grindr data. When they weren’t at their offices, where did they go? A small number of them had lingered at highway rest stops in the DC area at the same time and in proximity to other Grindr users—sometimes during the workday and sometimes while in transit between government facilities. For other Grindr users, he could infer where they lived, see where they traveled, even guess at whom they were dating.

Intelligence agencies have a long and unfortunate history of trying to root out LGBTQ Americans from their workforce, but this wasn’t Yeagley’s intent. He didn’t want anyone to get in trouble. No disciplinary actions were taken against any employee of the federal government based on Yeagley’s presentation. His aim was to show that buried in the seemingly innocuous technical data that comes off every cell phone in the world is a rich story—one that people might prefer to keep quiet.

«

A really great story.
unique link to this extract


The Guardian’s new “Deeply Read” article ranking focuses on attention, not just clicks • Nieman Journalism Lab

Laura Hazard Owen:

»

The No. 1 most-clicked story on The Guardian’s US site last Wednesday: “Alabama IVF ruling leaves Republicans stuck between their base and the broader public.” The most “deeply read” story, however, was on a very different topic: “Dune v Dune: do Denis Villeneuve’s films stay true to the book?“

“Deeply Read,” a feature launched Wednesday, “uses attention time to surface a wider range of journalism that other readers are spending more time with,” The Guardian said:

»

It appears on our regionalised home pages and reflects the interests of the region’s audience.

Not all of these pieces are long. To power the list we created a metric that looks at the attention time from readers compared with the length of the piece. This means that the list is diverse in terms of topic, length and format.

«

With news publishers increasingly relying on subscription revenue rather than advertising, engagement is becoming a more important metric. Expanding the kinds of “top” lists can also help publishers promote discovery within their own sites. The Guardian’s ranking gauges “active time spent” on a story, Chris Moran, the Guardian’s head of editorial innovation, explained to me via Twitter DM.

“The metric is a long-term internal one in Ophan [The Guardian’s internal analytics system] called the attention benchmark and it’s very simple,” he said. “It takes active reading time, takes into account the length of the article, and gives us a score out of five clocks. So five clocks is ‘this is a great reading time for this length!’ and one clock is ‘this isn’t great for this length.”

“We’ve had this for a number of years internally to help us see less reach-y pieces that really work with a smaller audience,” he added. “And for many years I’ve wanted to share it with readers because it highlights such great journalism and little off the beaten track of trending topics. To be clear it still matters to show people what is popular, but we love showing them something more.”

«

Chris is a very smart guy, and this is a typically clever thing to intrigue passers-by.
unique link to this extract


An update on Facebook News • Meta

»

In early April 2024, we will deprecate Facebook News – a dedicated tab in the bookmarks section on Facebook that spotlights news – in the US and Australia. This follows our September 2023 announcement that we deprecated Facebook News in the UK, France and Germany last year.   

This is part of an ongoing effort to better align our investments to our products and services people value the most. As a company, we have to focus our time and resources on things people tell us they want to see more of on the platform, including short form video. The number of people using Facebook News in Australia and the U.S. has dropped by over 80% last year. We know that people don’t come to Facebook for news and political content — they come to connect with people and discover new opportunities, passions and interests. As we previously shared in 2023, news makes up less than 3% of what people around the world see in their Facebook feed, and is a small part of the Facebook experience for the vast majority of people.

The changes affecting the Facebook News feature will not otherwise impact Meta’s products and services in these countries. People will still be able to view links to news articles on Facebook. News publishers will continue to have access to their Facebook accounts and Pages, where they can post links to their stories and direct people to their websites, in the same way any other individual or organization can. News organizations can also still leverage products like Reels and our ads system to reach broader audiences and drive people to their website, where they keep 100% of the revenue derived from outbound links on Facebook. 

«

It may be true that news is less than 3% of what people around the world see in their Facebook feed, but it’s 100% of what news organisations produce, and nobody has ever tried to measure how much of their output contributes to the content “people around the world” see when they’re not on Facebook. So while that 3% figure may be true, it may also be a distraction.
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

2 thoughts on “Start Up No.2181: Apple fined €1.84bn after Spotify complaint, AI chatbots give bad tax advice, 4K TV too good?, and more

  1. what is it with Americans screwing up the language? “Deprecate” means ‘belittle’ or ‘denigrate’ – not whatever nonsense Fb et al like to pretend. If they mean ‘wind down’ or ‘close’ why not just say that?

  2. I read the dune article without thinking about the connection to the film. What I found fascinating was the description of luminescence dating, which I had never heard of previously. The idea that minerals can tell us when they last saw the sun is magical.

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.