Gabe” on Twitter [sic]:

it’s cool how every google search now starts with a wall of LLM slop that is completely useless and takes up half the screen

Via Simon Willison:

Not all promotional content is spam, and not all AI-generated content is slop. But if it’s mindlessly generated and thrust upon someone who didn’t ask for it, slop is the perfect term for it.

Because it is such a buzzy field and an indicator of a company’s technological savviness, it serves many businesses to loudly trumpet how well they are adapting. Masking human help and borderline fraud are useful for looking cleverer than one actually is. Pushing A.I. — especially by using that terminology — is also very important for them.1

Google’s answer was probably fully generated by a computer; I do not think it is engaging in fraud. But it is advantageous to show that Google is really, really doing “A.I.”, even if this specific example provides little advantage over a typical Google snippet, and ensuring the AirPods marketing page and some third-party reviews are the first results.


  1. From a purely language-based perspective, one will note Apple is no longer holding on tomachine learning” and has eagerly embraced “A.I.” in its products. ↥︎

David Bushell:

I loathe what WordPress development has become. If you haven’t kept up with Gutenberg and full-site editing (FSE) you may be surprised at how radically different modern WordPress themes are — and not in a good way.

Modern WordPress theme development is a series of hacks which purportedly simplify parts of a theme by drastically overcomplicating other things. For years, it has not been possible to load fonts from providers like Adobe in WordPress’ block editor. This is not an edge case issue — everybody wants web fonts — nor is it a particularly unique example of how broken the WordPress theming environment is. If you want to use a web font loaded as a CSS file, you need to do it the old fashioned way.

I am sure there are good reasons to migrate away from the HTML-in-PHP standard of how WordPress themes used to work. But when I am building something for WordPress, I find myself fighting its newer structure constantly. Every time I do, the simplest and most predictable solution is to resort to twenty-year-old techniques — not only because I know them well, but because they are frequently the only things which work.

What a loss, and what a life.

A trip through Steve Albini’s catalogue as a recording engineer — never “producer” — is a varied and lengthy excursion. Yes, he recorded Nirvana’s “In Utero” and it is very good, and he also worked with Cloud Nothings, and Sunn O))), and Low, and black midi. He recorded the Stooges, a joint Robert Plant and Jimmy Page album, and I think his interpretation of Fugazi’s “In on the Kill Taker” remains a highlight, but he also made albums with artists like — a personal favourite — Hot Little Rocket.

A new record from his own band, Shellac, is out next Friday. It will be bittersweet.

Rebecca Kern, Politico:

TikTok and its parent company ByteDance sued Tuesday to challenge a law President Joe Biden signed to force the sale or ban of the video sharing app.

The petition filed with the U.S. Court of Appeals for the District of Columbia Circuit claims the law violates the First Amendment rights of its 170 million American users. It says the law shuts down the platform based on “speculative and analytically flawed concerns about data security and content manipulation.”

Nilay Patel on Threads:

This TikTok complaint is startlingly weak on first read — it basically handwaves through “a sale is impossible, let’s just assume it’s a ban” and begins the First Amendment argument from there

TikTok frames a jettisoning from ByteDance as something which would treat the United States as its own distinct company but, surely, an alternative interpretation of the U.S.’ intent is for the entire TikTok enterprise worldwide to be distinct from ByteDance.

Either way, there is plenty of confirmation in the suit from TikTok that its draft agreement with CFIUS includes all sorts of extraordinary stipulations. Why the two parties could not be satisfied with signatures on that document is a great question. TikTok, for its part, alleges the problems are with the U.S. government, and it says it is preemptively implementing several parts of the draft.

Kudos to Mark Gurman — Apple really did introduce the M4 SoC in the new iPad Pro models. The M4 comes just six months after Apple launched the M3, which is currently used in half the Mac lineup. The other half — the Mac Mini, the Mac Studio, and the Mac Pro — all use processors in the M2 family. Two of those models were only just launched at WWDC last year.

None of this makes sense to anybody outside of Apple. Perhaps it is not supposed to: any given processor is perhaps good enough that you do not need to worry. But Apple itself set up this comparison when it decided to use the same processors in iPads and Macs, and name them to clearly show which ones are newer. I am sure there are legitimate and plentiful performance improvements in each generation of new processors but it is a dizzying set of choices from a buyer’s perspective. Maybe there will be updates to the three Mac desktops at WWDC this year.

Update: Jason Snell, Six Colors:

Why the M4 now? It mostly has to do with Apple shifting chip production at TSMC (the company that fabs Apple’s chips) from the first-generation 3nm process to a new, more efficient second-generation 3nm process. There’s a whole backstory about TSMC’s change in 3nm processes that’s not worth getting into here, but suffice it to say that the first-generation process is largely a dead end, and the company is moving to a new set of 3nm processes.

That is the kind of backstory I would be interested in. However, as I wrote above, this is the kind of explanation which is logical for Apple but produces a confusing result for the rest of us.

Ashley Belanger, Ars Technica:

Near the end of the second day of closing arguments in the Google monopoly trial, US district judge Amit Mehta weighed whether sanctions were warranted over what the US Department of Justice described as Google’s “routine, regular, and normal destruction” of evidence.

[…]

According to the DOJ, Google destroyed potentially hundreds of thousands of chat sessions not just during their investigation but also during litigation. Google only stopped the practice after the DOJ discovered the policy. […]

It is entirely reasonable for individuals to conduct themselves privately and off-the-record, but an official corporate policy built around specific topics seems like a different matter. That Google kept it up even after the DOJ got involved is particularly shady.

Cheng Ting-Fang and Lauly Li, Nikkei:

Apple is deepening its ties with China even as it further expands production in Southeast Asia and India, highlighting the balancing act the iPhone maker is striking between politics and business.

Apple increased its China-headquartered suppliers and Chinese manufacturing sites in 2023 while using fewer suppliers from Taiwan, the U.S., Japan and South Korea, a Nikkei Asia analysis of Apple’s latest official list of suppliers shows.

Ben Jiang, South China Morning Post:

China’s central Henan province, home to the world’s largest iPhone manufacturing complex in its capital Zhengzhou, reported a 60 per cent year-on-year drop in smartphone exports in the first quarter, showing the impact of Apple’s moves to diversify production outside the mainland.

Malcolm Owen, AppleInsider:

In a China Observer report released on Monday, footage of a Foxconn industrial park in Nanning is shown to be deserted. Once employing 50,000 people, it’s now practically an empty shell. AppleInsider has learned that as Apple’s operations have moved elsewhere, manufacturing capacity was freed elsewhere in China, leading to this plant’s closure.

As Apple’s dependence on manufacturing in China increasingly becomes a liability, and many call for its extrication from the country, it can sometimes be easy to forget just how many people have jobs which depend on this supply chain. It is an unfathomably huge industry made up of hundreds of thousands of individuals. This is too often the story of outsourced labour: rich companies move around or exit entirely, leaving others holding the bag.

Satya Nadella, in a memo to Microsoft employees since posted on the company’s blog:

Today, I want to talk about something critical to our company’s future: prioritizing security above all else.

Microsoft runs on trust, and our success depends on earning and maintaining it. We have a unique opportunity and responsibility to build the most secure and trusted platform that the world innovates upon.

Charlie Bell, Microsoft’s executive vice president of security, expanded upon the company’s specific goals and priorities, and explained a particular incentive:

We will mobilize the expanded [Secure Future Initiative] pillars and goals across Microsoft and this will be a dimension in our hiring decisions. In addition, we will instill accountability by basing part of the compensation of the company’s Senior Leadership Team on our progress in meeting our security plans and milestones.

The obvious point of comparison for these memos is Bill Gates’ ‘Trustworthy Computing’ memo from 2002:

Trustworthiness is a much broader concept than security, and winning our customers’ trust involves more than just fixing bugs and achieving “five-nines” availability. It’s a fundamental challenge that spans the entire computing ecosystem, from individual chips all the way to global Internet services. It’s about smart software, services and industry-wide cooperation.

There is a sort of MBA-type wordiness in Nadella’s memo that is not present in the more direct Gates memo despite the latter being considerably longer, but both have similar goals. Microsoft’s poor track record, especially recently, is corroding the trust of its enterprise and government customers but — and this is the catch — where are they going to go?

Thanks to Listen Later for sponsoring this week’s posts at Pixel Envy.

Some sponsors provide their own Friday posts, but I was asked to write a little something of my own for Listen Later and I am happy to do so. I only accept sponsorships for products I actually like and would use. Listen Later is just such a service.

ListenLater

I use Listen Later to read long articles to me while I work, when I cook, and as I clean up after dinner. I have also used it to translate news stories. Maybe my favourite use is as a drafting tool for when I am writing a longer article and need to hear it read back to me. The text-to-speech quality is excellent, and having it show up in my podcast app alongside episodes from the shows I listen to makes it easy to keep track of new articles.

You know that article you swear you were going to read but still have not got around to? Yeah, the one in that ancient browser tab. Try hearing it with Listen Later’s free trial instead.

Raghav Mendiratta, of Stanford University’s Center for Internet and Society, in March 2021:

Under Rule 4(2), it is mandatory for a significant social media intermediary providing messaging services to identify the first originator of a message if a competent court or executive authority orders that it is necessary to do so for the purposes of investigation and prosecution of certain offences punishable with imprisonment for a term not less than five years. Technical experts say that compliance with this requirement is not possible unless end-to-end encryption on messaging services such as WhatsApp is broken.

WhatsApp sued over these rules the same month and then, last week, threatened to leave India if it is required to comply with policies that threaten encryption.

Indu Bhan, Economic Times:

WhatsApp LLC on Thursday told the Delhi High Court that the popular messaging platform will end if it is made to break encryption of messages.

“As a platform, we are saying, if we are told to break encryption, then WhatsApp goes,” counsel Tejas Karia, appearing for WhatsApp, told a Division Bench comprising Acting Chief Justice Manmohan and Justice Manmeet Pritam Singh Arora.

This is a familiar threat from WhatsApp, but it feels particularly weighty in India owing to the its extraordinary popularity in the country. I have to wonder if WhatsApp is bluffing. Would it really abandon the hundreds of millions of users in its most popular geography?

Apple:

[…] Today, we’re introducing two additional conditions in which the CTF is not required:

  • First, no CTF is required if a developer has no revenue whatsoever. […]

  • Second, small developers (less than €10 million in global annual business revenue*) that adopt the alternative business terms receive a 3-year free on-ramp to the CTF to help them create innovative apps and rapidly grow their business. […]

Two fundamental issues remain with the Core Technology Fee — namely, that developers still need to pay Apple even if their app is distributed exclusively outside the App Store and in-app payments are handled by a third-party processor, and the fee is an unknown and surprising future charge. One marvels at how the Mac could remain such a successful developer platform for so long without the support of a per-install fee.

But I was wrong. This is a meaningful relaxation of terms for entirely free apps, like the young developer example raised by Riley Testut during the March DMA compliance workshop.

Zach Ocean:

Encountered AI music in the wild today

Motown-style tracks straight from Suno/Udio with … interesting … titles and lyrics

Recommended by Spotify via Discover Weekly

These “interesting” songs include instant classics like “My Arms Are Just Fuckin’ Stuck Like This” and “It’s Time To Take a Shit on the Company’s Dime”. Classic Happy Bunny-style humour.

Ryan Broderick explains in Garbage Day:

The story behind the page is interesting. Obscurest Vinyl started as a Facebook page that would photoshop fake album covers for classic records that didn’t exist. The page recently shifted into posting AI songs to go with the fake album covers. As one commenter noted, you can tell the songs are AI because most of them feature bass and drum parts that don’t repeat in any discernible pattern. The account also regularly fights with users on Instagram who gripe about it using AI.

Truly embarrassing for Spotify that it is recommending stuff like this, and not for the first time.

Timothy B. Lee, writing in Asterisk:

Over the last decade, Silicon Valley elites have grown increasingly frustrated with media coverage of their industry. And they aren’t wrong that coverage has grown increasingly negative. But I think they’re wrong to assume this reflects a hostility toward Silicon Valley in particular.

A more banal explanation is that companies like Google, Facebook, and Uber aren’t startups anymore. It no longer makes sense to publish positive profiles introducing readers to these companies. So reporters have switched to treating Silicon Valley giants like other big companies, which means mostly writing about them when they do something wrong.

Lee is right that tech journalism often consists of thin stories built off press releases and simplistic narratives — but so, too, does most general audience journalism. While there is the occasional nuanced story with correct weighting given to affirming and dissenting views, it is far more common to see misapplied view from nowhere journalism. But, critically, this is true of all beats. Erwin Knoll once said “everything you read in the newspapers is absolutely true except for the rare story of which you happen to have firsthand knowledge”, and that includes knowledge of media itself. Given how pressured journalists are, as Lee is careful to note, it is not difficult to see why stories across a range of topics become either pure boosterism or damp scandals.

Apple:

Starting May 1, 2024, new or updated apps that have a newly added third-party SDK that‘s on the list of commonly used third-party SDKs will need all of the following to be submitted in App Store Connect:

  1. Required reasons for each listed API

  2. Privacy manifests

  3. Valid signatures when the SDK is added as a binary dependency

Jesse Squires:

Historically, Apple has rarely, if ever, explicitly acknowledged any third-party SDK or library. It took years for them to even acknowledge community tools like CocoaPods in Xcode’s release notes. Thus, it is interesting to see which SDKs they have deemed important or concerning enough to explicitly mandate a privacy manifest. And, in typical Apple fashion, I’m pretty sure SDKs authors were not notified about this in advance. We all learned which SDKs need privacy manifests at the same time — when the list was published.

When this requirement was announced at WWDC last year, I assumed this list would be dominated by SDKs for analytics, authentication, logging, advertising, and other potentially sensitive use cases. After all, it came on the heels of reporting by the Markup and the Wall Street Journal about SDKs invisible to end users and implicated in mass surveillance, with one such software package — X-Modebanned by Apple and Google.

This list of SDKs contains seemingly few such packages. As of writing, there are 87 SDKs on Apple’s list and fully one-quarter of them — by my count — are Flutter packages intended to simplify cross-platform development. I can see how there could be risks to file and photo pickers, for example, but this list sure looks more like it is comprised of popular SDKs, not necessarily ones of privacy concern. Kits from Facebook and Snap are on the list, but TikTok’s is nowhere to be found. Several Google SDKs are on the list, including Firebase analytics, but Google’s standalone ads framework is not; Unity is on the list, but not Unity’s ad kit.

As Squires writes, any documentation about why these SDKs are on Apple’s list would be helpful. I would even take a sentence fragment.

Robert Simmon, Nightingale:

The launch of Ikonos was one of a handful of developments that allowed newsrooms to expand from reporting on rocket launches and satellite hardware, to using remote sensing data as an essential tool to help tell stories. A wide variety of satellite data are now used to provide context to the news, to document events, and as a tool for investigation.

It still blows my mind that I — a nobody — can open Google Earth any time I want and see aerial photography with a level of detail that would have been classified not too long ago. Years of imagery are available, too, so if I want to see how an area has changed, it is just a few button clicks away. I appreciated Simmon’s look at how capabilities like these have allowed journalists at places like Bellingcat and Buzzfeed News to document events in ways that would not have been possible before widespread consumer satellite photography.

Welcome to a world where your reading list becomes your listening playlist. We are thrilled to introduce you to ListenLater.net — an innovative service that turns articles into podcasts, making your favourite reads accessible in your favourite podcast app.

ListenLater

Have you ever stumbled upon an article that piqued your interest but didn’t have the time to delve into it? Whether you’re commuting, exercising, or relaxing, ListenLater.net simplifies the process. Just email the article’s URL, and presto! Listen Later takes over, converting it into an engaging podcast episode. The narration is so lifelike, you’ll find yourself immersed in the listening experience.

But the innovation doesn’t end with articles. Listen Later enhances your entire email experience. Forward any email, and it will transform its content and attachments into a podcast episode just for you — whether it’s work reports, newsletters, or those lengthy reads you’ve postponed.

Worried about language barriers? Listen Later has you covered! ListenLater.net offers multilingual translation and narration services. Simply select your preferred language, and immerse yourself in a vast array of content, all available to your ears.

Join us in embracing convenience and versatility. Listen Later brings your articles, emails, and documents to life in audio format. Head over to ListenLater.net and start your free trial today. Let’s redefine the way we consume information, one podcast at a time.

Rarely do I link to something just because I want you to go read it, but this piece by Alex Ross in the New Yorker is just such an occasion. It is a wonderful piece about how we sometimes embrace noise and sometimes reject it, and what “noise” even means. (Via Matt.)

Tom Wheeler, former FCC chairman, writing for the Brookings Institution in October, following a vote to begin the process of reclassifying broadband as a “Title II” telecommunications service, regarding efforts to paint net neutrality regulations as no big deal:

It is the conduct of the ISPs that is in question here. Because telephone companies were Title II common carriers, their behavior had to be just and reasonable. Those companies prospered under such responsibilities; as they have morphed into wired and wireless ISPs, there is no reasonable argument why they, as well as their new competitors from the cable companies, should not continue to have public interest obligations.

Jon Brodkin, Ars Technica:

The Federal Communications Commission voted 3–2 to impose net neutrality rules today, restoring the common-carrier regulatory framework enforced during the Obama era and then abandoned while Trump was president.

[…]

ISPs insist the rules aren’t necessary because they already follow net neutrality principles yet also claim the rules are so burdensome that they will be prevented from investing more in their networks. Lobby group USTelecom today said the “relentless regulation” comes at the cost of “failing to achieve Internet for all.”

Karl Bode, Techdirt:

While broadband providers have already started whining about the rules and threatened to sue, privately (just like last time) broadband industry executives doubt the rules will have any meaningful impact on their businesses. The rules aren’t onerous, won’t likely be enforced with any consistency, and big companies like AT&T and Comcast have never, ever really had to worry about serious FCC penalties for any of their various predatory, anti-competitive, or illegal behaviors.

Bode has, for years, covered the effort to paint the reversal of net neutrality rules as inconsequential. Contrary to popular belief, the reclassification to a Title I service produced plenty of ill effects. Part of the problem was in mainstream coverage of what the rules meant and, similarly, in what their 2018 undoing would entail. Given the U.S.’ pivotal role in internet products worldwide, this protective measure to reduce the power of ISPs is a welcome one.