New research shows that the universe’s accelerating expansion is slowing down, which means that… things are complicated. CC-licensed photo by European Southern Observatory 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 8 links for you. Not expanding or contracting. I’m @charlesarthur on Twitter. On BlueSky: @charles.arthur. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
Elon Musk’s Twitter takeover is now the worst buyout for banks since the financial crisis • WSJ
Alexander Saeedy and Dana Mattioli:
»
The $13bn that Elon Musk borrowed to buy Twitter has turned into the worst merger-finance deal for banks since the 2008-09 financial crisis.
The seven banks involved in the deal, including Morgan Stanley and Bank of America, lent the money to the billionaire’s holding company to take the social media platform, now named X, private in October 2022. Banks that provide loans for takeovers generally sell the debt quickly to other investors to get it off their balance sheets, making money on fees.
The banks haven’t been able to offload the debt without incurring major losses—largely because of X’s weak financial performance—leaving the loans stuck on their balance sheets, or “hung” in industry jargon. The resulting writedowns have hobbled the banks’ loan books and, in one case, was a factor that crimped compensation for a bank’s merger department, according to people involved with the deal.
The value of the loans to Musk quickly soured after the $44bn acquisition was completed. But new analysis shows how their persistent underperformance has put the deal in historic territory.
According to data from PitchBook LCD, the Twitter loans have been hung longer than every similar unsold deal since the 2008-09 financial crisis for which the research firm has complete records. There were many more hung deals back then, but banks in that period typically were still able to sell or write off most of their hung debt within roughly a year after they issued the loans. One hung deal—a $20bn all-debt acquisition in 2007—was bigger than Twitter but wound up in bankruptcy about 12 months after banks wrote the check.
…The banks—which also include Barclays, Mitsubishi UFJ Financial Group, BNP Paribas, Mizuho and Société Générale—have been able to collect hefty interest payments from the X loans. They are generally for seven to eight years and carry rates several percentage points above the benchmark for investment-grade companies. And the banks could still ultimately be made whole if X is able to cover its interest obligations and repay the principal when the loans mature.
«
Hard to see anything making the company’s value rise, which means that the debt is going to be hard to sell.
unique link to this extract
National Public Data confirms breach exposing Social Security numbers • Bleeping Computer
Ionut Ilascu:
»
Background check service National Public Data confirms that hackers breached its systems after threat actors leaked a stolen database with millions of social security numbers and other sensitive personal information.
The company states that the breached data may include names, email addresses, phone numbers, social security numbers (SSNs), and postal addresses.
In the statement disclosing the security incident, National Public Data says that “the information that was suspected of being breached contained name, email address, phone number, social security number, and mailing address(es).”
The company acknowledges the “leaks of certain data in April 2024 and summer 2024” and believes the breach is associated with a threat actor “that was trying to hack into data in late December 2023.”
NPD says they investigated the incident, cooperated with law enforcement, and reviewed the potentially affected records. If significant developments occur, the company “will try to notify” the impacted individuals.
It is worth noting that BleepingComputer’s testing revealed that access to NPD’s statement on the security incident has been blocked for IP addresses in numerous locations in the US as well as regions outside the country. More than a dozen captures of the page exist on the Internet Archive, though.
«
At this point might as well just publish all this stuff online rather than make hackers go to the work of hacking it. Takes away their ability to threaten exposure, doesn’t it. (And demonstrates again that there are only two classes of company, those which have been hacked and those which are going to get hacked.)
unique link to this extract
Waning dark energy may evade ‘swampland’ of impossible universes • Quanta Magazine
Steve Nadis:
»
Since 1998, we’ve known that the expansion of the universe is accelerating. Dark energy is the name given to the accelerant. In the standard theoretical model of the cosmos, dark energy has a simple form: It is spread uniformly in space, maintaining a constant density at all times. Dark energy of this type, known as the cosmological constant, would not get diluted as the universe expands. Rather, as space grows, the amount of dark energy grows with it. And so this accumulating energy expands the universe with ever-increasing gusto.
But the constancy of dark energy is merely a hypothesis, one that the DESI experiment set out to check. The collaboration reported that they had so far mapped and analyzed the locations of 6.4 million galaxies to determine how fast the universe has expanded as a function of time. (More distant galaxies reveal the younger universe.) If dark energy is a cosmological constant, then the acceleration should hold steady. DESI’s data set wasn’t big enough to test this by itself, though the team will be able to do so after mapping a total of 40 million galaxies during the five-year survey. But when the scientists pooled their data with other astronomical observations, the combined data sets favored an evolving dark energy.
The findings fall short of the level of statistical certainty needed to claim a discovery. Right now, DESI is calling it a “hint” that dark energy might be weakening in strength. But according to Nathalie Palanque-Delabrouille, a cosmologist at Lawrence Berkeley National Laboratory and one of the leaders of DESI, “that hint gets stronger as we start combining different data sets. And all these data sets seem to be pointing in the same direction.”
«
This is a good, clearly written article about quite where cosmology and physics have got to in the puzzle about the universe’s expansion.
unique link to this extract
Competing in search • Benedict Evans
After the decision in favour of the DOJ finding Google breached the antitrust law, Evans ponders the aftermath:
»
what comes next is not clear at all. It’s obvious that the court will order Google to stop its TAC payments and abandon the contracts (or at a minimum radically reduce their scope), which means it saves $30bn in cash each year and Apple loses $20bn. But then what?
The court might order ‘choice screens’ in user-downloaded Chrome (20% of queries), but it has no power to order a choice screen in Safari, since Apple isn’t a party, and it’s not clear whether it can do that on Android, given that Samsung and Motorola, and the telcos, aren’t parties either. Meanwhile, it’s also unclear whether choice screens actually work, especially where (as the judgment laid out) Google has a much better brand than Bing, let alone any of the other alternatives. A lot of people given a choice screen would just choose Google anyway, as indeed appears to have happened in the EU, which already mandated this on Android. Meanwhile, the DoJ’s head of competition, Jonathan Kanter, was clear in an interview last week that he wants any remedies to change the situation, not just remove the illegal conduct:
»
Q: If this ends up with Google just not having to pay Apple but still being the default, have you accomplished anything?
We want to make sure that remedies in any case, whether it’s this one or any other, are meaningful and meet the markets where they are today, not where they were 15 years ago
«
Then, there are press reports that the DoJ is considering asking for Android and Chrome to be spun out. Spinning out Android seems irrelevant except as a purely punitive step: if it was independent, that wouldn’t change how Samsung sets defaults. If Chrome was independent, it could choose a different default too. But would it?
This is really the first big puzzle: what will Apple, Samsung and Motorola, and a hypothetically independent Chrome Inc, choose as the default if Google can’t pay to be the default?
«
Apple’s Eddy Cue told the judge that Microsoft’s Bing would have to pay more than 100% of revenue share because he sees its quality as being so inferior to Google’s. But things can change, I suppose.
unique link to this extract
Scrolling through online videos increases feelings of boredom, study finds • The Guardian
Nicola Davis:
»
Browsing videos on TikTok or YouTube can be a hit-and-miss affair, with gems lurking amid mediocre efforts. But researchers have found that switching to another video, or skipping forwards and backwards in the same one, actually makes people more bored.
Dr Katy Tam at the University of Toronto Scarborough, the lead author of the research, said boredom was closely linked to attention.
“We feel bored when there’s a gap between how engaged we are and how engaged we want to be,” she said. “When people keep switching through videos, they become less engaged with the videos and they are looking for something more interesting. This can lead to increased feelings of boredom.”
The results appear to chime with other studies: as the team notes, previous research has suggested that while boredom relief is a driver for people to use social media or smartphones, the use of such technology appears to make the feeling worse.
Writing in the Journal of Experimental Psychology: General, Tam and colleagues report how they carried out seven experiments involving a total of more than 1,200 participants.
The first, involving 140 participants, revealed that people tended to switch between videos more when they rated the content more boring, while the second – an online survey involving 231 participants – suggested people thought having the option to skip through a video or switch to another would make viewing a video less boring.
However, the team’s subsequent experiments suggest this is not the case.
«
I’m sure TikTok will shut down forthwith.
unique link to this extract
Former CEO of failed bank sentenced to prison • United States Department of Justice
»
A Kansas man was sentenced to 293 months in prison for using his former position as chief executive officer (CEO) of a bank to embezzle tens of millions of dollars in a cryptocurrency scheme which led the bank to fail at a complete loss of equity for investors.
According to court documents, Shan Hanes, 53, of Elkhart pleaded guilty to one count of embezzlement by a bank officer.
While the CEO of Heartland Tri-State Bank (HTSB) in Elkhart, Kansas, Hanes initiated 11 outgoing wire transfers between May 2023 and July 2023 totalling $47.1m of Heartland’s funds to a cryptocurrency wallet in a cryptocurrency scheme referred to as “pig butchering.” The funds were transferred to multiple cryptocurrency accounts controlled by unidentified third parties during the time HTSB was insured by the Federal Deposit Insurance Corporation (FDIC). The FDIC absorbed the $47.1m loss. Hanes’ fraudulent actions caused HTSB to fail and the bank investors to lose $9m.
A federal judge ordered that restitution be finalized at a separate hearing within the next 90 days.
“Hanes’ greed knew no bounds. He trespassed his professional obligations, his personal relationships, and federal law. Not only did Shan Hanes betray Heartland Bank and its investors, but his illegal schemes also jeopardized confidence in financial institutions,” said US Attorney Kate E. Brubacher. “Today’s sentence is a measure of justice for the victims, and a statement that the U.S. Department of Justice will hold those accountable who violate positions of trust for their own gain.”
«
He has a Twitter profile: “father and banker who loves sports“. His last tweet was in September 2022, when he was in Washington meeting legislators “to discuss banking concerns”. Such as embezzlement? In classic manner, he was approached via WhatsApp. Don’t answer those random messages!
unique link to this extract
Google makes your Pixel screenshots searchable with Recall-like AI feature • The Verge
Emma Roth:
»
Google has announced Pixel Screenshots, a new AI-powered app for its Pixel 9 lineup that lets you save, organize, and surface information from screenshots. Pixel Screenshot uses Google’s private, on-device Gemini Nano AI model to analyze the content of an image and make it searchable.
During a demo at its Pixel launch event, Google showed how you can take a screenshot and then save it to a collection, like “gift ideas.” You can also search through all your other screenshots by typing in a keyword, like “bikes” or “shoes.” Pixel Screenshots will then pull up all relevant results.
Additionally, Pixel Screenshots can give you information about what’s inside an image. So, if you’re looking for the price of a shirt you screenshotted, you can type in “t-shirt price,” and Pixel Screenshots will extract the information from your screenshots to surface an answer. The app will only be available on Pixel 9 devices.
«
Will probably trickle down to other Pixel devices, and then to other Android devices, in time. Though I thought Apple Photos could do something vaguely similar in the searching for text within a picture or screenshot.
unique link to this extract
Colleges still don’t have a plan for AI cheating • The Atlantic
Ian Bogost:
»
No matter how small the assignments, many students will complete them using ChatGPT. “Students would submit ChatGPT responses even to prompts like ‘Introduce yourself to the class in 500 words or fewer,’” [one writing professor] said.
If the first year of AI college ended in a feeling of dismay, the situation has now devolved into absurdism. Teachers struggle to continue teaching even as they wonder whether they are grading students or computers; in the meantime, an endless AI-cheating-and-detection arms race plays out in the background. Technologists have been trying out new ways to curb the problem; the Wall Street Journal article describes one of several frameworks.
OpenAI is experimenting with a method to hide a digital watermark in its output, which could be spotted later on and used to show that a given text was created by AI. But watermarks can be tampered with, and any detector built to look for them can check only for those created by a specific AI system. That might explain why OpenAI hasn’t chosen to release its watermarking feature—doing so would just push its customers to watermark-free services.
Other approaches have been tried. Researchers at Georgia Tech devised a system that compares how students used to answer specific essay questions before ChatGPT was invented with how they do so now. A company called PowerNotes integrates OpenAI services into an AI-changes-tracked version of Google Docs, which can allow an instructor to see all of ChatGPT’s additions to a given document. But methods like these are either unproved in real-world settings or limited in their ability to prevent cheating. In its formal statement of principles on generative AI from last fall, the Association for Computing Machinery asserted that “reliably detecting the output of generative AI systems without an embedded watermark is beyond the current state of the art, which is unlikely to change in a projectable timeframe.”
«
The arms race between cheating and catching the cheats continues. But it also raises the question: what are you testing the students about? What’s the purpose of the out-of-classroom work? And eventually, won’t this be solved by having exams in internet-free rooms – perhaps using pen and paper?
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
The Overspill: when there's more that I want to say 