By John Gruber

• Archive
• The Talk Show
• Dithering
• Projects
• Contact
• Colophon
• Feeds / Social
• Sponsorship

Lex.Games: Free daily word games from Lex Friedman. Not the weird Elon stan;
the real Lex Friedman.

Release Day for Apple’s .4 OS Versions ★

All out today:

• iOS 18.4 (priority notifications, lots of new languages and EU support for Apple Intelligence).
• MacOS 15.4 (AI message categorization in Apple Mail, Quick Start for setting up a new machine).
• VisionOS 2.4 (Apple Intelligence, new Spatial Gallery app, new Vision Pro sibling app for iPhone).
• WatchOS 11.4 (new Sleep Wake Up alarm option that will emit sound even when the watch is in silent mode).
• tvOS 18.4 (UI tweaks to TV app).

The White House Correspondents Association Speaks Cowardice to Power ★

Brian Steinberg and Pat Saperstein, reporting for Variety over the weekend:

The White House Correspondents’ Association has canceled plans to have comedian Amber Ruffin perform at its annual dinner on April 26, a new sign of the pressures being brought to bear on news organizations during President Donald Trump’s second term.

The journalism group, which has seen its control over interactions with Trump eroded in recent weeks, made the decision after Taylor Budowich, a White House deputy chief of staff, raised comments Ruffin has made in the past that are critical of Trump. Earlier this week, Ruffin told a podcast backed by The Daily Beast that she would not try to make sure her jokes targeted all sides of the political spectrum as the WHCA had requested, and likened the Trump administration to “kind of a bunch of murderers.” Playing to both sides “makes them feel like human beings,” she said, “cause they’re not.” [...]

“The WHCA board has unanimously decided we are no longer featuring a comedic performance this year. At this consequential moment for journalism, I want to ensure the focus is not on the politics of division but entirely on awarding our colleagues for their outstanding work and providing scholarship and mentorship to the next generation of journalists,” WHCA president Eugene Daniels wrote to members in a statement.

“For the past couple of weeks, I have been planning a re-envisioning of our dinner tradition for this year,” he added. “As the date nears, I will share more details of the plans in place to honor journalistic excellence and a robust, independent media covering the most powerful office in the world.”

What an enormous mountain of obvious horseshit this explanation is. The WHCA only announced that Ruffin would be hosting this year’s show on February 4, at which point this lickspittle clown Eugene Daniels was quoted thus by The Hollywood Reporter:

“When I began to think about what entertainer would be a perfect fit for the dinner this year, Amber was immediately at the top of my list,” Eugene Daniels of Politico, president of the association, said in a statement on Tuesday. “She has the ability to walk the line between blistering commentary and humor all while provoking her audience to think about the important issues of the day. I’m thrilled and honored she said yes.”

So eight weeks ago this obsequious bootlicker Daniels thought Amber Ruffin “would be a perfect fit for the dinner this year” but now, four weeks before the show, he’s trying to claim with a straight face that “For the past couple of weeks, I have been planning a re-envisioning of our dinner tradition for this year”?

The kids magazine Highlights for Children has a long-running comic strip called “Goofus and Gallant”, the premise of which is that Goofus is a kid who always does the wrong thing, and Gallant always does the right thing. Goofus, especially in the older strips, is an absurd parody. This Eugene Daniels toady is the Goofus of journalism. The entire point of the WHCA is to assert the group’s collective independence as journalists — and the independence they assert is specifically from the White House. So of course they shouldn’t have responded to White House pressure to fire Ruffin as this year’s host. But it’s even worse for an ostensible journalist — the president of the WHCA for chrissakes — to try to get even a single person in the world to believe that this is anything other than caving to White House pressure, and that in fact (“Yeah, that’s the ticket!”) he’d been planning to cancel the entire concept of having a comedian host at all “for the past couple of weeks” when just eight weeks ago he described Ruffin as “a perfect fit for the dinner this year”.

This is Baghdad Bob level nonsense. I’m not one for performative resignations, but how does any news outlet or journalist agree to remain a member of the WHCA after this?

Sunday, 30 March 2025

Lex.Games ★

My thanks to Lex Friedman for sponsoring this past week at DF to promote Lex.Games, a collection of eight daily word games. Quoting from Friedman’s own description in the sponsored RSS entry at the start of the week:

I paid Gruber many thousands of dollars to run this ad for free games which themselves have no ads. Please keep reading.

The games:

Conlextions: Inspired by NYT’s Connections
Lexicogs: Solve crossword-style clues by assembling letter “cogs”
By a Vowel: A word jumble game with missing vowels
Six Appeal: Wordle with six-letter words

There’s also a daily Mini Crossword; a Full-Size Crossword; and Mind Control, which is a whole lot like Mastermind and not actually a word game at all; don’t sue me.

Oh, and if you only counted seven games here, the eighth is iOS-only. It’s called Letter Opener, and it’s my favorite.

I actually hate Letter Opener, because I’m terrible at games like that. Looking at the leaderboard, though, obviously some of you are really good at it. Six Appeal is more my speed (which is to say, like Wordle, it has no clock). But go ahead and download the iOS app and try Letter Opener. Maybe you’re a fast enough thinker for it.

So the basic pitch is that Lex.Games really is just a bunch of fun daily games that are free to play, without ads (let alone without annoying ads). But you can — and should! — pay a modest $20/year to subscribe to get access to extra games, leaderboards, and just to support a very fun and satisfying endeavor.

Vertu Is Still Selling Phones, Which Suggests Someone Is Still Buying Them ★

Calfskin for $1,500, flip-foldables for $5,000, and whatever these are for a lot more. Who needs any sense (or a spelling checker) when you’ve got “elesant charisma / heroic essence”?

Or as I cited Andy Warhol back in 2012:

A Coke is a Coke and no amount of money can get you a better Coke than the one the bum on the corner is drinking. All the Cokes are the same and all the Cokes are good. Liz Taylor knows it, the President knows it, the bum knows it, and you know it.

See Also: This 2023 investigation by Andrew Williams for Wired, that more or less uncovers that today’s Vertu is just a brand snapped onto white-label phones made by ZTE: “Never before have I used a phone where I felt so unsafe, one that feels like it could be used to scam me — though, to be clear, I have no evidence that it is.”

Keach Hagey Reports on the Backstory Behind Sam Altman’s Firing and Quick Rehiring at OpenAI in 2023 ★

Interesting excerpt at the WSJ from Keach Hagey’s upcoming book The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future. (Main link is a gift link, but also here’s a News+ link.)

Sunday, 30 March 2025

MacOS 15 Sequoia’s Annoying-as-Hell ‘Turn On Reactions’ Menu Bar Prompt ★

Matt Birchler:

I really thought that the screen recording notifications in macOS Sequoia would be the bane of my existence, but thankfully those have been changed quite a bit from the early betas last summer and they’re totally a non-issue in my book today. However, these god damned “turn on reactions” alerts have got to die in a fire, and they need to have done it yesterday.

I understand why Apple decided to show this once. Why though, is it seemingly designed to reappear every time I start a video call? Who is not annoyed by this?

Friday, 28 March 2025

ChatGPT 4o Adds Image Generation and It’s Fun as Hell ★

MG Siegler:

I’m not even talking about Apple and AI here. We’ve done that, a lot. Probably enough — for now (famous last words). I’m talking about Apple in general. Watching this OpenAI video — again, not an event, just a product walk-through with various team members (though this one happened to be “MC’d” by Sam Altman) — I had this old, familiar feeling as they walked through the new features: joy.

As ridiculous as it may sound, I was almost giddy around what I was seeing. It’s a feeling that I recall well from many an Apple event back in the day.

“Yes, this is exactly what I wanted! They did it!” That kind of thing.

This will sound unfair or harsh to Apple, but I really don’t think that it is. I can really only speak for myself here, and perhaps I’m alone — but I suspect that I’m not — it has been a while since I’ve gotten that feeling from an Apple announcement. That loving feeling.

I gave the updated ChatGPT the instruction “Create an image of the main characters from ‘Severance’ as Lego figures” and this is the first response it gave me:

I gave Apple’s Image Playground, running on MacOS 15.3.2 Sequoia, the equivalent prompt — “The main characters from ‘Severance’ as Lego figures” — and it gave me this as its first response:

One of the above images qualifies as “Hey, that could be better but it’s pretty good for the first crack from a simple prompt”, and the other qualifies as “This bears zero resemblance to anyone or anything from Severance”.

Guess which one of the two Apple is actively promoting to users as something they should try?

Thursday, 27 March 2025

Yours Truly on ‘Geared Up’ ★

You’ll never guess what cohosts Andru Edwards and Jon Rettinger talked to me about on their Geared Up podcast this week. OK, fine, you guessed it. But I bet you didn’t guess that Flash on iPhone came up. Very fun show — I think you’ll enjoy it.

You’ll Never Guess Which Recent DF Article Was ‘Flagged’ at Hacker News ★

“Flagging” isn’t what I was writing about yesterday, with regard to Daring Fireball’s unexplained ghosting at Hacker News in recent years. If you look at the list of recent DF articles at HN, only one is explicitly “flagged”. Whoops, now it’s two, but they’re both the same article.

Tim Bray:

I’m really glad @gruber posted this, because me too. Same timeframe, same behavior.

Oliver Reichenstein (from iA):

Same for content from iA. Again and again our posts get to the top page and then someone ghosts them. Ghosting started around the same time as DF ghosting, I think. My take: It looks like someone that has enough access just doesn’t like us. ¯\(ツ)/¯

I’ve heard privately from a few other bloggers that they’ve seen similar ghosting at HN in recent years.

Near the top of Hackers News’s FAQ:

How are stories ranked?

The basic algorithm divides points by a power of the time since a story was submitted. Comments in threads are ranked the same way.

Other factors affecting rank include user flags, anti-abuse software, software which demotes overheated discussions, account or site weighting, and moderator action.

Hacker News presents itself as a forum that is primarily driven by the community, where ranking and moderation are mostly, perhaps even overwhelmingly, transparent. And that “moderator action” is only a secondary or even tertiary factor. The core HN audience buys into this — the HN audience is comprised of people who view themselves as independent thinkers. Part of why they like and trust HN as an aggregator is that they believe they’re driving it, and that they know how the whole thing works.

My thesis is that the above might once have been an accurate summary of how Hacker News functioned, but hasn’t been for years, and that there now exists a cabal of moderator/admins with their thumbs on the scale, and their personal predilections are the primary steering force behind what’s permitted to surface and what gets ghosted. This moderation cabal operates more or less in secret. Their actions, and thus even their usernames, are invisible — lest the HN community discover that it’s steering things about as much as Maggie Simpson is.

Read between the lines of the “Hey, why is this post flagged?” wonderment from genuinely openminded HN users in the comments on my now-flagged submission, and you’ll catch a strong whiff of “Pay no attention to the man behind the curtain!”

Wednesday, 26 March 2025

The Atlantic Has an Owner Committed to the Cause: Laurene Powell Jobs ★

Oliver Darcy, writing at Status (paywalled — great content, terrible CMS experience that keeps logging me out on all my devices and requires a stupid email magic link to get back in), in a brief interview with Atlantic editor-in-chief and man of the moment Jeffrey Goldberg:

It goes without saying that there are many ironies associated with this particular story. One of them is that Goldberg, a journalist who Donald Trump loathes for his past reporting (remember the “suckers and losers” piece), somehow became the unintended recipient of high-level, real-time military intelligence from inside his own inner circle. One wonders whether any heads will roll as a result of the whole matter. On Monday, Trump again made his disdain for Goldberg known, blasting him at a press conference and absurdly claiming The Atlantic is “a magazine that is going out of business.”

If anything, of course, publishing a story like the one Goldberg did on Monday proves how strong the outlet currently is. That type of muscular journalism requires skill, strong leadership, and the backing of a courageous publisher. I asked Goldberg about owner Laurene Powell Jobs’ role in the matter. He declined to comment specifically on this particular story, but offered words of praise: “Laurene Powell Jobs is a stalwart and brave publisher at a time when cowardice rules the day.”

If it had been a Washington Post reporter who was inadvertently included on the Trump national security team’s Signal group chat, would they have run the story? No fucking way with that abject lickspittle coward Jeff Bezos running the show.

Apple Now Selling PCs From Other Companies at Apple.com ★

The work remains mysterious and important.

2026 Porsches Still Won’t Have Next-Gen CarPlay, Which Was Announced in 2022 ★

Hartley Charlton, MacRumors:

Apple’s next-generation CarPlay experience is still nowhere to be seen following Porsche’s announcement of a major upgrade of its infotainment system for 2026.

The upcoming 2026 model year Porsche Taycan, 911, Panamera, and Cayenne feature an upgraded version of the Porsche Communication Management (PCM) system, making it more responsive, adding Dolby Atmos support, and integrating Amazon’s Alexa voice assistant. The new system brings the Porsche App Center, a kind of app store for the vehicle, to all of the new models.

It continues to support the standard version of Apple CarPlay and Android Auto. Support for Apple’s next-generation CarPlay experience is again conspicuously missing from Porsche’s new lineup, and the automaker did not mention it at all during its latest announcement — another bleak sign for the delayed feature.

I’d crack a joke about it looking less and less likely that next-gen CarPlay was going to appear in 2024, but I already did that in January, when Apple itself took the date off its CarPlay page. That announcement came at WWDC 2022.

In this case (unlike the advanced personalized features of Apple Intelligence) it was actually sensible for Apple to pre-announce the existence of next-gen CarPlay, given the reliance on third parties. But it also should have been clear just how incredibly hard it would be to get third party carmakers up to snuff on being able to ship it, so Apple giving a date, any date, was always odd. Apple doesn’t make a car, and you can’t promise what you can’t control. They should have just said “soon”.

Calling the White House’s Bluff, The Atlantic Releases the ‘Houthi PC Small Group’ Signal Text Thread ★

Jeffrey Goldberg and Shane Harris, reporting once again for The Atlantic:

Ratcliffe said much the same: “My communications, to be clear, in the Signal message group were entirely permissible and lawful and did not include classified information.”

President Donald Trump, asked yesterday afternoon about the same matter, said, “It wasn’t classified information.”

These statements presented us with a dilemma. In The Atlantic’s initial story about the Signal chat — the “Houthi PC small group,” as it was named by Waltz — we withheld specific information related to weapons and to the timing of attacks that we found in certain texts. As a general rule, we do not publish information about military operations if that information could possibly jeopardize the lives of U.S. personnel. That is why we chose to characterize the nature of the information being shared, not specific details about the attacks.

The statements by Hegseth, Gabbard, Ratcliffe, and Trump — combined with the assertions made by numerous administration officials that we are lying about the content of the Signal texts — have led us to believe that people should see the texts in order to reach their own conclusions. There is a clear public interest in disclosing the sort of information that Trump advisers included in nonsecure communications channels, especially because senior administration figures are attempting to downplay the significance of the messages that were shared.

I linked yesterday to a quote from Hannah Arendt, whom Wikipedia aptly describes as “one of the most influential political theorists of the twentieth century”. The quote I linked to was her observation that the ranks of authoritarian governments inevitably wind up being filled with “crackpots and fools” because they’re the people whose loyalty is most assured. In some sense the Jedi mind trick is real — it works on the weak-minded. Regardless of one’s political beliefs, no intelligent person of integrity (as opposed to, say, a foreign mole) would participate in a discussion of obviously classified and highly sensitive war plans in a Signal chat. It’s jarring to see it so clearly but U.S. national security is now led entirely by morons.

Most of the quotes on the Goodreads page I linked to, culled from Arendt’s seminal The Origins of Totalitarianism, are related to truth, not idiocy. Here’s one:

The outstanding negative quality of the totalitarian elite is that it never stops to think about the world as it really is and never compares the lies with reality.

And:

Before mass leaders seize the power to fit reality to their lies, their propaganda is marked by its extreme contempt for facts as such, for in their opinion fact depends entirely on the power of man who can fabricate it.

And:

The ideal subject of totalitarian rule is not the convinced Nazi or the convinced Communist, but people for whom the distinction between fact and fiction (i.e., the reality of experience) and the distinction between true and false (i.e., the standards of thought) no longer exist.

When The Atlantic’s initial story hit, everyone responsible in the Trump administration, right up to the president himself, just immediately began telling bald-faced lies about what happened, despite the fact that they knew Jeffrey Goldberg literally had the receipts to prove otherwise. That works, until it doesn’t.

Tuesday, 25 March 2025

Of Course Trump Has Surrounded Himself With Idiots This Time Around ★

Hannah Arendt, The Origins of Totalitarianism:

Totalitarianism in power invariably replaces all first-rate talents, regardless of their sympathies, with those crackpots and fools whose lack of intelligence and creativity is still the best guarantee of their loyalty.

Eight years of Trump was going to be eight years too many no matter how it worked out, but the four year Biden term between Trump terms makes the difference clear. Trump corrected what he perceived as a lack of loyalty/fealty in his first term by surrounding himself with nothing but morons this time.

Notification Summary Miscues ★

Paul Kafasis:

Since they were first enabled last year, I have frequently found Apple Intelligence’s notification summaries for emails to be something less than helpful. Here are some I spotted in just the past few days.

The first one of these is particularly interesting because it highlights a key area where LLMs are frustratingly stupid. Kafasis got a notification summary from Apple Intelligence claiming “Package shipped for $427 order” for a used book he’d purchased. The email from Amazon, from which Apple Intelligence gleaned the information, had the price formatted thus: $4²⁷ — omitting the decimal and putting the cents in superscript. That’s a centuries-old formatting idiom for prices that remains common — e.g. at Walmart — to this day. But Apple Intelligence just sees dollar-sign, four, two, seven, and thus $427.

That’s just stupid.

But where it really gets frustrating is that everyone has to learn this at some point. If you were at Walmart with a kid, and the kid asked why, say, dog food is so expensive, pointing to a sign that says it cost $9⁸⁷ per bag, you’d explain it, once, and the kid would never forget it. “Oh, that’s just another way of writing nine dollars and eighty-seven cents — they do it that way to emphasize the dollar amount and de-emphasize the cents, which really don’t matter.” This would make intuitive sense to the child as well, because they know dog food probably costs about $10 per bag, not $1,000 per bag.

There is no way to properly explain something like this to an LLM (yet?). You can’t teach it like we do with children. Or at least you can’t do it in a way that jibes with our human sense of “learning” — it’s more like how the Guy Pearce protagonist “learns” in Christopher Nolan’s Memento. Here, tattoo another thing to remember on your arm. But at least ChatGPT is trying to learn about us, albeit in its crude Memento-like way. With Apple Intelligence in particular, you can’t teach it at all. There’s no place in the system where you can correct the very simple, easily-explained mistake it made upon seeing $4²⁷ in an email. The next time an email from Amazon comes with a price formatted like that, Apple Intelligence is likely to summarize it the exact same wrong way — off by a factor of 100 — again. And there’s nothing we can do about it.

OPSEC Isn’t Even the Worst Part of ‘SignalGate’ ★

Josh Marshal, writing at Talking Points Memo:

Especially in the national security domain, many things the government does have to remain secret. Sometimes those things remain secret for years or decades. But they’re not secrets from the U.S. government. The U.S. government owns all those communications, all those facts of its own history. Using a Signal app like this is hiding what’s happening from the government itself. And that is almost certainly not an unintended byproduct but the very reason for the use. These are disappearing communications. They won’t be in the National Archives. Future administrations won’t know what happened. There also won’t be any records to determine whether crimes were committed.

This all goes to the fundamental point Trump has never been able to accept: that the U.S. government is the property of the American people and it persists over time with individual officeholders merely temporary occupants charged with administering an entity they don’t own or possess.

Think this is hyperbole? Remember that when Trump held his notorious meeting with Vladimir Putin in Helsinki in 2019 he confiscated his translator’s notes and ordered him not to divulge anything that had been discussed. Remember that Trump got impeached over an extortion plot recorded in the government record of his phone call with President Zelensky. An intelligence analyst discovered what had happened and decided he needed to report the conduct. These aren’t hypotheticals. They’ve already happened. And he’s even been caught. Which is probably one reason there’s so much use of Signal.

The Problem Is Far More Than Just Whether Signal Is ‘Secure’ ★

Maggie Miller and Dana Nickel, reporting for Politico:

The app’s security is viewed as fairly strong due to its robust privacy features and minimal data collection, as well as default end-to-end encryption of all messages and voice calls. The app also includes a function that deletes all messages from a conversation within a set time frame, adding an additional layer of data protection. But experts agree that it shouldn’t be used by government officials as an alternative to communicating through more secure, sanctioned government communications — which Signal is not.

“It’s so unbelievable,” a former White House official, granted anonymity to discuss The Atlantic’s report candidly, said Monday. “These guys all have traveling security details to set up secure comms for them, wherever they are.”

Signal’s encryption is more than just “fairly strong”. It’s very strong, arguably the gold standard in consumer-available communications. But that’s not the point. The point is it’s a consumer application. This whole fiasco happened because you can just mistakenly add the wrong person to a group conversation, which wouldn’t be possible if the Trump national security team were using appropriate channels.

And the disappearing messages thing doesn’t add security. It adds some level of privacy, but it’s an additional factor that makes all of this completely illegal. But avoiding any future scrutiny is almost certainly one reason Trump’s kakistocratic cabinet is using Signal in the first place.

The former White House official pointed out that members of Trump’s Cabinet — including the vice president, Defense Secretary Pete Hegseth, and Director of National Intelligence Tulsi Gabbard, among others — were likely using personal devices, since in most cases, Signal cannot be downloaded onto official federal devices. This alone creates a host of cybersecurity issues.

Wrote one DF reader (who has professional experience in this area) to me today, “There is no legal way whatsoever that classified information can be communicated over the public Internet — private device, personally owned device, Chromebook, anything. It is all wildly illegal.”

Days After the Trump National Security Team’s Signal Leak, the Pentagon Warned That Russian Hackers Are Using Phishing Attacks to Abuse Signal’s ‘Linked Devices’ Feature ★

NPR:

Several days after top national security officials accidentally included a reporter in a Signal chat about bombing Houthi sites in Yemen, a Pentagon-wide advisory warned against using the messaging app, even for unclassified information.

“A vulnerability has been identified in the Signal messenger application,” begins the department-wide email, dated March 18, obtained by NPR. The memo continues, “Russian professional hacking groups are employing the ‘linked devices’ features to spy on encrypted conversations.” It notes that Google has identified Russian hacking groups who are “targeting Signal Messenger to spy on persons of interest.”

It’s not a weakness in Signal’s cryptography, it’s a hole in their device-mirroring setup. From that Google Threat Intelligence post, published last month:

The most novel and widely used technique underpinning Russian-aligned attempts to compromise Signal accounts is the abuse of the app’s legitimate “linked devices” feature that enables Signal to be used on multiple devices concurrently. Because linking an additional device typically requires scanning a quick-response (QR) code, threat actors have resorted to crafting malicious QR codes that, when scanned, will link a victim’s account to an actor-controlled Signal instance. If successful, future messages will be delivered synchronously to both the victim and the threat actor in real-time, providing a persistent means to eavesdrop on the victim’s secure conversations without the need for full-device compromise.

You’d have to be a bit of a doofus to fall for such a phishing attack if you were in a national security leadership position, but, well, our national security leadership positions are currently occupied by what the Russians call “useful idiots”.

New York Post: ‘European Union to Fine Meta Up to $1B or More for Breaching DMA’ ★

Thomas Barrabi, reporting for The New York Post:

The European Union is set to slap Mark Zuckerberg’s Meta with a fine that could stretch to $1 billion or more for allegedly violating its strict antitrust rules, The Post has learned — setting up a possible showdown with President Trump, who has compared the EU’s penalties to “overseas extortion.”

The European Commission, the EU’s antitrust watchdog, is expected to conclude that Meta is not in compliance with the Digital Markets Act, sources close to the situation told The Post on Monday. [...] The fine is expected to be hundreds of millions of dollars and potentially more than $1 billion, the sources said. [...]

Apple is also in the EU’s crosshairs and a fine against the iPhone maker could be announced this week or next week, the sources said. Earlier this month, Reuters reported that Apple and Meta were likely to face “modest fines” for DMA breaches. EU antitrust chief Theresa Ribera previously said a decision on enforcement actions for both companies was coming in March.

Reuters Reports European Commission Will Decline to Fine Apple Over Browser Choice Screen, But Hints It Will Over Anti-Steering Provisions ★

Foo Yun Chee, reporting for Reuters under the headline “Exclusive: Apple Set to Stave Off EU Fine Into Browser Options, Sources Say”:

Apple is set to stave off a possible fine and an EU order over its browser options on iPhones after it made changes to comply with landmark EU rules aimed at reining in Big Tech, people with direct knowledge of the matter said on Tuesday. The European Commission, which launched an investigation in March last year under the Digital Markets Act (DMA), is expected to close its investigation early next week, the people said.

A win’s a win and a closed investigation’s a closed investigation, but the browser choice screen never seemed like a problem for Apple. I follow this stuff closely, and have even written (at times extensively) about how dumb and ineffective these mandatory browser choice screens are, and I didn’t realize this investigation was still open, because it seemed so clear Apple had done what they needed to for compliance.

So, more interesting to me is this bit buried lower in the article, suggesting the EC is going to fine Apple next week over non-compliance with the DMA’s anti-steering provisions:

The Commission’s decision to close the investigation early next week will come at the same time as it hands out fines to Apple and Meta Platforms for DMA violations and orders to comply with the legislation, the people said.

In this second Apple case, the issue is whether the company imposes restrictions that hinder app developers from informing users about offers outside its App Store free of charge.

WWDC 2025 Dates: June 9–13 ★

Apple Newsroom:

To celebrate the start of WWDC, Apple will also host an in-person experience on June 9 that will provide developers with the opportunity to watch the Keynote and Platforms State of the Union at Apple Park, meet with Apple experts one-on-one and in group labs, and take part in special activities. Space will be limited; details on how to apply to attend can be found on the WWDC25 website.

Right on time: in recent years, WWDC dates have been announced on:

• Tuesday 30 March 2021
• Tuesday 5 April 2022
• Wednesday 29 March 2023
• Tuesday 26 March 2024

and now today, Tuesday 25 March 2025. Last Tuesday in March next year is March 31 — that’s my guess for next year’s announcement.

And, yes, the “25” in the logo has a decidedly glassy look and some animation that’s just plain fun.

Tuesday, 25 March 2025

Threads Is Losing to Bluesky ★

Jon Passantino, writing at Status:

Now Threads feels rather lifeless. While users still post there, for many it has become something of a second-tier platform — a place that they dump content out of habit, not because they’re having real conversations or finding meaningful engagement.

Matt Birchler:

I believe Meta that there are hundreds of millions of people signing on every month, but they seem to be doing absolutely nothing there. More interesting stuff is on Bluesky and Mastodon, and better conversation happens on those platforms as well.

I feel the same way. Threads has dropped to a decided #3 for me after Mastodon and Bluesky, and (a) I don’t really have room in my head or time in my day for 3 of these platforms, and (b) I’m more than OK with Meta’s entry falling by the wayside.

Like, if the answer at the moment for Twitter-style social media is Bluesky (general audience) and Mastodon (nerds), that’s ... the best outcome? Even X (chaos and Musk sycophancy) seems to have a better, more defined vision for what it’s supposed to be than Threads.

Monday, 24 March 2025

‘Tim, Don’t Kill My Vibe’ ★

Bryan Irace:

Barring a sharp correction, Apple looks increasingly likely to miss out on a generation of developers conditioned to first reach for tools like Cursor, Replit, or v0 — especially as Apple’s own AI tooling remains notably absent. This goes well beyond enabling new entrants to “vibe code” — experienced mobile developers who, despite history with Xcode and a predilection for building native apps, are begrudgingly swapping out their tools in acknowledgement of the inarguable productivity benefits. [...]

App Review has always long been a major source of developer frustration. Authoritarian yet inconsistent policy enforcement aside, it’s simply too hard to distribute software even to your own Apple devices, let alone someone else’s. This isn’t new by any means, but as the time to build an app shrinks from weeks/months to hours/days, it feels more egregious — and thus like more of a liability — than ever before.

This is a thoughtful and thought-provoking piece. I feel like Irace is onto something here that I haven’t seen anyone put their finger on before.

The App Store, when it debuted, made developers deliriously happy. The UIKit frameworks (a.k.a. CocoaTouch), Objective-C, and Xcode were all way better ways to create apps for mobile devices than anything else at the time. And for distribution, going through Apple and the App Store was way easier and way more democratic, and 70/30 was way more generous to developers, than anything from the various phone carriers around the world. You’d be lucky to get a 30/70 split from the carriers, and they’d only deal with large corporate developers. There were no indie or hobbyist mobile app developers before the App Store. (It’s kind of nutty in hindsight that network carriers were the only distribution channel for apps 17 years ago.)

17 years is a long time, though. And developers long ago stopped seeing the App Store as something that makes them happy, or that reduces friction and hassle from their lives. Instead they view it as a major source of friction and hassle. Apple should have focused on keeping the App Store as a thing that makes developers (mostly) happy all along, not (as things stand today) mostly miserable.

Basically, the threat to Apple that the App Store poses is not regulators coming for it. That’s a distraction. The threat, as I’ve long tried to argue, perhaps unsuccessfully, is that market forces will work against it eventually. Developers have long since grown resentful toward the technical and bureaucratic hassles of publishing through the App Store, and the size of the purchase commissions Apple keeps for itself. Apple’s commission percentages haven’t grown over time, but a 70/30 split that in 2008 seemed remarkably generous (or even the newer 85/15 small-business split) today seems like a platform engaging in usury and abusive rent-seeking.

AI might be the disruption that brings about the “eventually”, because now it’s coming for the developer tooling experience too. If Apple’s native programming frameworks and developer tools aren’t the best, most satisfying, most productive ways to create great apps, what’s left that makes developers happy to be creating for the iOS platform?

Apple should move mountains to refocus itself on making the experience of developing for (and on) Apple platforms the best in the world, including distribution and monetization. Instead, they seem to be resting on the assumption that it’s a privilege, self-evident to all, just to be allowed to develop for Apple platforms.

Software Update for AirPods Max to Enable Lossless Audio and ‘Ultra-Low Latency’ ★

Apple Newsroom:

Next month, a new software update will bring lossless audio and ultra-low latency audio to AirPods Max, delivering the ultimate listening experience and even greater performance for music production. With the included USB-C cable, users can enjoy the highest-quality audio across music, movies, and games, while music creators can experience significant enhancements to songwriting, beat making, production, and mixing.

Apple also started selling a new $40 USB-C to 3.5mm audio cable — male USB-C on the side that goes into your AirPods Max, male headphone jack on the other side to go into the audio-out port on a Mac or, say, an airplane seat.

Getting a Modern LLM Running on a 2005 PowerBook G4 ★

Andrew Rossignol:

I have been diving into the world of large language models (LLMs), and a question began to gnaw at me: could I bring the cutting-edge of AI to the nostalgic glow of my trusty 2005 PowerBook G4? Armed with a 1.5GHz processor, a full gigabyte of RAM, and a limiting 32-bit address space, I embarked on an experiment that actually yielded results. I have successfully managed to achieve LLM inference on this classic piece of Apple history, proving that even yesteryear’s hardware can have a taste of tomorrow’s AI.

A fun project, well-explained. Even a great choice of computer to run it on — the 12-inch PowerBook G4 is one of the best-looking computers ever made. (Via Joe Rossignol.)

‘On Tyranny’ by Timothy Snyder ★

I read Timothy Snyder’s On Tyranny after the election. A collection of 20 essays — each relatively brief, some exceptionally brief — it’s more booklet than book, and can easily be consumed in an afternoon or a few evenings. I finished it with an unsettled feeling. I read it again last week, and my feeling now is both more unsettled and more resolute.

Snyder, a plain-speaking history professor at Yale, has a core message, which he’s been hammering since before Trump’s re-election: Do not obey in advance. Resist. The following passage hit me harder on this second reading, two months into Trump 2.0, than it did in November. From Chapter 19: “Be a Patriot”:

It is not patriotic to admire foreign dictators. It is not patriotic to cultivate a relationship with Kim Jong Un; or to say that Bashar al-Assad and Vladimir Putin are superior leaders. It is not patriotic to call upon foreign leaders to intervene in American presidential elections. It is not patriotic to cite Russian propaganda at rallies. It is not patriotic to share an adviser with Russian oligarchs. It is not patriotic to appoint advisers with financial interests in Russian companies. It is not patriotic to appoint a National Security Advisor who likes to be called “General Misha,” nor to pardon him for his crimes. It is not patriotic when that pardoned official calls for martial law. It is not patriotic to refer to American soldiers as “losers” and “suckers.” It is not patriotic to take health care from families, nor to golf your way through a national epidemic in which half a million Americans die. It is not patriotic to try to sabotage an American election, nor to claim victory after defeat. It is not patriotic to try to end democracy.

A nationalist might do all these things, but a nationalist is not a patriot. A nationalist encourages us to be our worst, and then tells us that we are the best. A nationalist, “although endlessly brooding on power, victory, defeat, revenge,” wrote Orwell, tends to be “uninterested in what happens in the real world.” Nationalism is relativist, since the only truth is the resentment we feel when we contemplate others. As the novelist Danilo Kiš put it, nationalism “has no universal values, aesthetic or ethical.”

A patriot, by contrast, wants the nation to live up to its ideals, which means asking us to be our best selves. A patriot must be concerned with the real world, which is the only place where his country can be loved and sustained. A patriot has universal values, standards by which he judges his nation, always wishing it well — and wishing that it would do better.

Democracy failed in Europe in the 1920s, ’30s, and ’40s, and it is failing not only in much of Europe but in many parts of the world today. It is that history and experience that reveals to us the dark range of our possible futures. A nationalist will say that “it can’t happen here,” which is the first step toward disaster. A patriot says that it could happen here, but that we will stop it.

I highly recommend the book. Get it at Amazon, Bookshop.org, or Apple Books.

The Trump Administration Accidentally Included Jeffrey Goldberg, Editor of The Atlantic, in a Signal Group Chat That Revealed War Plans for Yemen ★

Jeffrey Goldberg, editor in chief of The Atlantic (News+ link):

The world found out shortly before 2 p.m. eastern time on March 15 that the United States was bombing Houthi targets across Yemen.

I, however, knew two hours before the first bombs exploded that the attack might be coming. The reason I knew this is that Pete Hegseth, the secretary of defense, had texted me the war plan at 11:44 a.m. The plan included precise information about weapons packages, targets, and timing.

This is going to require some explaining. [...]

The notion of a journalist being accidentally included in a war-planning group of national security leaders — and the very notion that U.S. national security leaders would use Signal to conduct such a group — is so preposterous that Goldberg had assumed the group was a hoax, with the intention of embarrassing him. But it was real.

Earlier today, I emailed Waltz and sent him a message on his Signal account. I also wrote to Pete Hegseth, John Ratcliffe, Tulsi Gabbard, and other officials. In an email, I outlined some of my questions: Is the “Houthi PC small group” a genuine Signal thread? Did they know that I was included in this group? Was I (on the off chance) included on purpose? If not, who did they think I was? Did anyone realize who I was when I was added, or when I removed myself from the group? Do senior Trump-administration officials use Signal regularly for sensitive discussions? Do the officials believe that the use of such a channel could endanger American personnel?

I’ll add: Do they sniff glue and eat paste?

There’s so much chaos at the moment resulting from the Trump administration’s actions during these first two months that it’s easy to overlook one salient fact: Trump has chosen to surround himself with idiots.

Sunday, 23 March 2025

The Talk Show: ‘Podcasting Technology Cadence’ ★

MG Siegler returns to the show to talk about the drama surrounding Siri and Apple Intelligence.

Sponsored by:

• WorkOS: The modern identity platform for B2B SaaS — free up to 1 million monthly active users. Check out their latest features from Launch Week.
• BetterHelp: Give online therapy a try at BetterHelp and get on your way to being your best self.
• OpenCase: MagSafe perfected that’s thinner, lighter, and more secure. Save 10% with code TALKSHOW.

Friday, 21 March 2025

Weekly Sponsorships Here at Daring Fireball ★

How has your week been? My week was ... busy. That includes a new episode of The Talk Show recorded yesterday, dropping in your favorite podcast app soon. Amidst all the writing (and talking) I’ve been doing, I’m also working on filling up open weeks on the sponsorship schedule for Q2.

After a very full February and March, I’ve got a bunch of openings in the next few months — and openings for the next two weeks, starting with this Monday. Update: The coming week just sold, but the next week, starting March 31, remains open.

Weekly sponsorships have been the top source of revenue for Daring Fireball ever since I started selling them back in 2007. They’ve succeeded, I think, because they make everyone happy. They generate good money. There’s only one sponsor per week and the sponsors are always relevant to at least some sizable portion of the DF audience, so you, the reader, are never annoyed and hopefully often intrigued by them. And, from the sponsors’ perspective, they work. My favorite thing about them is how many sponsors return for subsequent weeks after seeing the results.

If you’ve got a product or service you think would be of interest to DF’s audience of people obsessed with high quality and good design, get in touch.

WorkOS: Launch Week ★

My thanks to WorkOS for sponsoring DF, once again, this last week. This has been WorkOS’s Launch Week, and they’ve got a slew of new features to show. Honestly, though, you should check out their Launch Week page just to look at it — it’s beautiful, fun retro-modern pixel-art goodness. Great typography too. I wish every website looked even half this cool.

New features launched just this week include:

• WorkOS Connect — “Sign in with [Your App]”
• WorkOS Vault — Encryption Key Management (EKM) and Bring-Your-Own-Key (BYOK)
• AuthKit Integrations — Native support for several new identity providers including LinkedIn, Slack, GitLab, BitBucket, Intuit, and more.

Ookla: ‘A First Look at How Apple’s C1 Modem Performs With Early Adopters’ ★

Ookla, the company behind the Speedtest download/upload bandwidth testing app:

Although it’s early in the adoption curve for the iPhone 16e, we analyzed the performance of the new device from March 1st through March 12th, and compared it to the performance of iPhone 16, which has a similar design and the same 6.1” screen. Both devices run on the same Apple-designed A18 SoC.

When we compare Speedtest Intelligence data from the top 90th percentile (those with the highest performance experience) of iPhone 16e and iPhone 16 users from all three of the top U.S. operators, we see the iPhone 16 performing better in download speeds. However, at the opposite end, with the 10th percentile of users (those who experience the lowest performance) we see the iPhone 16e performing better than the iPhone 16.

There are some differences, but overall the 16e’s cellular performance seems great for the frequencies it supports. And given the efficiency claims from Apple, it might be the better overall modem. (I also think the frequencies it doesn’t support don’t really matter all that much in real-world practice. If you know that you really make use of the crazy-high speeds of mmWave from Verizon, then you know the C1 modem is not for you.)

if the System Clipboard has an image
    play a sound
else
    simulate the keystroke ⌘V

Friday, 21 March 2025

Yahoo Sold TechCrunch ★

Emma Roth, The Verge:

TechCrunch has a new owner, again. Yahoo has sold the tech news site to the private equity firm Regent for an undisclosed sum, according to an announcement on Friday.

Regent is the same company that snapped up Foundry, the firm behind outlets like PCWorld, Macworld, and TechAdvisor on Thursday. Founded in 2005, TechCrunch has experienced many shakeups in ownership after AOL acquired the site in 2010.

A lot of shakeups in a lot of media companies’ ownership lately. Steady as she goes here at The Daring Fireball Company, a subsidiary of Fedora World Media Industries.

Matthew Belloni on the ‘Apple TV+ Experiment’ ★

Matthew Belloni has a very good take on Apple TV+ at Puck (that’s a gift link that should get you through their paywall — but which requires you creating a free account, sorry):

All of which fed into the self-centered fears of my lunch date. What, if anything, does the current state of Apple mean for its entertainment business? After all, more than five years into the Apple TV+ experiment, it’s never been entirely clear what C.E.O. Tim Cook and services chief Eddy Cue are up to in Hollywood. Certainly not making money, at least not in the traditional sense. The Information reported today that Apple lost $1 billion on Apple TV+ last year, following a Bloomberg report that more than $20 billion has been shoveled into making original shows and movies since 2019. That’s not nothing, even for a company worth $3 trillion.

The “loss” number is a bit misleading, of course, considering Apple has always said that a key goal is to leverage Leo DiCaprio and Reese Witherspoon to thicken its brand halo and the device “ecosystem,” ultimately boosting its other businesses. But still… for all its billions, Apple TV+ has accumulated only about 45 million subscribers worldwide, according to today’s Information report and other estimates.

That’s far less than Disney+, Max, and Paramount+, all of which launched around the same time. Those rival services are attached to legacy studios with rich libraries, but they’re not attached to a company with $65 billion in cash on hand and a device in the pockets of 1 billion people that also delivers bundle-friendly music, news, and games. Apple declined to confirm or comment on any numbers, but a source there suggested the subscriber number is higher than 45 million and that the global nature of the sub base is being undercounted by U.S.-oriented research firms. Maybe. The company reveals zero performance data beyond B.S. “biggest weekend ever!” press releases that the trades accept without skepticism and producers like Ben Stiller and David Ellison post with “blessed” emojis on their social media. No one outside the company really knows how the Apple TV+ business is performing.

One interesting nugget is this chart, which suggests that subscriptions to TV+ have boomed since Apple and Amazon worked out a deal to sell TV+ subscriptions through Amazon Channels in Prime Video at the end of last year. That deal has, seemingly, moved the needle. Another interesting nugget is that TV+ seems to suffer from a higher churn rate than other streaming services. Said Belloni’s Puck colleague Julia Alexander, “Fewer than 35 percent of all subscribers keep the service for longer than six months.”

That’s kind of crazy. I’d think TV+ would have less churn, not more, than the industry average — that the Apple TV+ audience is small but loyal. Perhaps this is the unsurprising side effect of Apple giving away 3-month trials when you purchase new devices. But I also truly wonder if TV+ subscriptions are the hardest for industry groups to measure, because so many people who do subscribe watch through tvOS (or, on their phones, on iOS) where everything is private. Belloni hints at this, and says little birdies at Apple told him the TV+ subscriber base is larger than they’re getting credit for.

And how do you count Apple One subscribers toward TV+’s subscriber base? My vague theory about Cue and Cook’s thinking about getting into this business has been about making it one leg among several on the stool of reasons to subscribe to Apple One. That Apple will take subscribers who are only subscribed to TV+, or only subscribed to TV+ and Apple Music, but what they really want is to get people to subscribe to Apple One, which, because it includes iCloud storage, almost certainly has very little churn.

Belloni closes thus:

Apple wouldn’t be the first tech powerhouse to dabble in professionally produced content only to retreat. [...] Neither Cook nor Cue has suggested anything like that, and Apple, in just over five years, has become a reliable partner and a high-quality buyer for Hollywood shows and movies. In some ways, it’s remarkable how fast Apple TV+ became part of the entertainment community. Whether that lasts is the question.

Here’s where I will point out that Apple isn’t like other tech companies. Apple isn’t a move fast and break things company. They’re a measure twice, cut once company. When they commit to something, they tend to stay committed. And they’re very, very good at playing long games that require patience, especially when entering new markets. Look at Apple Pay. 10 years ago, it was widely panned as a flop after a slow first year. Now it’s everywhere.

Is Apple’s Spending on TV+ Content a ‘Loss’ or a ‘Cost’? ★

Jill Goldsmith, Deadline:

Apple is losing more than $1 billion a year on streamer Apple TV+, according to a report in the Information that cited two people familiar with the matter. The tech giant has spent over $5 billion a year on content since launching Apple TV+ in 2019 but trimmed that by about $500 million last year, the report said.

The headline on Wayne Ma’s report at The Information set the framework: “Apple Streaming Losses Top $1 Billion a Year” — the story got picked up widely, and almost everyone who did framed it in terms of losing or a loss. But is it a loss when Apple expected the business to be unprofitable for a decade or more? From Scharon Harding’s paraphrasing at Ars Technica of Ma’s paywalled report:

Apple TV+ being Apple’s only service not turning a profit isn’t good, but it’s also expected. Like other streaming services, Apple TV+ wasn’t expected to be profitable until years after its launch. An Apple TV+ employee that The Information said reviewed the streaming service’s business plan said Apple TV+ is expected to lose $15 billion to $20 billion during its first 10 years.

For comparison, Disney’s direct-to-consumer streaming business had operating losses of $11.4 billion between the launch of Disney+ in fall 2020 and April 2024. Disney’s streaming business became profitable for the first time in its fiscal quarter ending on June 29, 2024.

The above two paragraphs of essential context are buried 13 paragraphs down. If Apple expected TV+ to operate in the red, to the tune of $15–20 billion over its first decade, and halfway through that decade (TV+ debuted in November 2019) it operated in the red to the tune of $1 billion for the year — doesn’t that mean costs are exactly in line with their expectations?

The insinuation here is that Apple’s pissing this money away and doesn’t know what they’re doing. Maybe they are! But if so it was exactly Eddy Cue and Tim Cook’s strategy to piss this money away. If Apple had expected TV+ to be profitable or break-even in 2024, then a $1 billion operating loss would be a story. But as it stands it’s just a cost. How much did Apple “lose” on electricity bills last year?

iOS 18 Software Updates Keep Re-Enabling Apple Intelligence for Users Who Had Turned It Off ★

Juli Clover, writing for MacRumors last week:

With new iOS software updates, Apple has been automatically turning Apple Intelligence on again even for users who have disabled it, a decision that has become increasingly frustrating for those that don’t want to use Apple Intelligence .

After installing iOS 18.3.2, iPhone users have noticed that Apple Intelligence is automatically turned on, regardless of whether it was turned off prior to the update being installed. There is an Apple Intelligence splash screen that comes up after updating, and there is no option other than tapping “Continue,” which turns on Apple Intelligence .

If you’ve updated to iOS 18.3.2 and do not want Apple Intelligence enabled, you will need to go the Settings app, tap on Apple Intelligence, and then toggle it off. When Apple Intelligence is enabled, it consumes up to 7GB of storage space for local AI models, which is an inconvenience when storage space is limited.

I’d been seeing complaints about this, including from some friends who are developers and/or had previously worked on iOS as engineers at Apple. A bunch of regular DF readers have written to complain about it too. I wouldn’t call it a deluge, but I’ve gotten an unusual number of complaints about this. (And at CNet, Jeff Carlson reports the same thing happening with MacOS 15.3.2.)

I hadn’t experienced it personally because I have Apple Intelligence enabled on my iPhone. But my year-old iPhone 15 Pro was still running iOS 18.2. So I disabled Apple Intelligence on that phone, then updated it to 18.3.2. When it finished, Apple Intelligence was re-enabled. I also tried this on my iPhone 16e review unit, which was still running iOS 18.3.1 (albeit a version of 18.3.1 with a unique build number for the 16e). I turned Apple Intelligence off, upgraded to 18.3.2, and on that iPhone, Apple Intelligence remained off after the software upgrade completed.

So I don’t know if this is a bug that only affects some iPhones, or a deliberate growth hacking decision from Apple to keep turning this back on for people who have explicitly turned it off. But it’s definitely happening.

And while the 7 GB of storage space required for the model is a legitimate technical reason to turn it off, I think (judging from my email from DF readers) the main reason people disable Apple Intelligence is that they don’t like it, don’t trust it, and to some degree object to it. It could take up no additional storage space at all and they’d still want it disabled on their devices, and they are fucking angry that Apple’s own software updates keep turning it back on. Put aside the quality or utility of Apple Intelligence as it stands today, and there are people who object to the whole thing on principle or, I don’t know, just vibes alone. Feelings are strong about this. Turning it back on automatically, after a user had turned it off manually, leads those users to correctly distrust Apple Intelligence specifically and Apple in general.

If it’s a bug, it’s a bug that makes Apple look like a bunch of gross shysters. If it’s not a bug, it means Apple is a bunch of gross shysters. I’d wager on bug — especially after seeing it not happen on my 16e review unit. I’m thinking it’s something where it’s supposed to be enabled by default, once, for people who’ve never explicitly turned Apple Intelligence on or off previously, but that for some devices where it has been turned off explicitly, somehow the software update is mistaking it for the setting never having been touched. Apple needs to get it together on this one.

Apple Sued for False Advertising Over Apple Intelligence ★

Ina Fried, reporting for Axios:

The suit, filed Wednesday in U.S. District Court in San Jose, seeks class action status and unspecified financial damages on behalf of those who purchased Apple Intelligence-capable iPhones and other devices.

“Apple’s advertisements saturated the internet, television, and other airwaves to cultivate a clear and reasonable consumer expectation that these transformative features would be available upon the iPhone’s release,” the suit reads. “This drove unprecedented excitement in the market, even for Apple, as the company knew it would, and as part of Apple’s ongoing effort to convince consumers to upgrade at a premium price and to distinguish itself from competitors deemed to be winning the AI-arms race.”

Most of these class action lawsuits are bullshit, but it’s hard to argue with the basic premise of this one.

Thursday, 20 March 2025

The Seneca ★

This is beautiful and crazy, and no, I’m not going to buy one, but damn I’m tempted and I’d sure like to try one. I’m glad it exists.

Gurman: Tim Cook Has Put Mike Rockwell in Charge of Siri, Reporting to Craig Federighi ★

Mark Gurman, with a blockbuster scoop for Bloomberg:

Apple Inc. is undergoing a rare shake-up of its executive ranks, aiming to get its artificial intelligence efforts back on track after months of delays and stumbles, according to people familiar with the situation.

Chief Executive Officer Tim Cook has lost confidence in the ability of AI head John Giannandrea to execute on product development, so he’s moving over another top executive to help: Vision Pro creator Mike Rockwell. In a new role, Rockwell will be in charge of the Siri virtual assistant, according to the people, who asked not to be identified because the moves haven’t been announced.

Rockwell will report to software chief Craig Federighi, removing Siri completely from Giannandrea’s command. Apple is poised to announce the changes to employees this week. The iPhone maker’s senior leaders — a group known as the Top 100 — just met at a secretive, annual offsite gathering to discuss the future of the company. Its AI efforts were a key talking point at the summit, Bloomberg News has reported. [...]

My quick take on this is that it’s a turf battle that Craig Federighi just won. It’s not just putting a new executive in charge of Siri, it’s moving Siri under Federighi’s group.

How Gurman got this scoop before Apple had announced the changes — even internally — is rather unbelievable. It’s not “Bloomberg” that got this scoop. It’s Mark Gurman. And trust me, Apple PR did not leak this to him deliberately. I’m sure they’re now accelerating an announcement, at least internally, framing it on their own terms. I can only guess that Gurman hinted at his sourcing in the passage above: Tim Cook must have announced these changes at the Top 100 retreat this week, and at least two of those attendees leaked the news to Gurman. Unprecedented.

Also:

Rockwell is currently the vice president in charge of the Vision Products Group, or VPG, the division that developed Apple’s headset. As part of the changes, he’ll be leaving that team and handing the reins to Paul Meade, an executive who has run hardware engineering for the Vision Pro under Rockwell.

I don’t find it surprising at all that Rockwell was given this task.

Giannandrea will remain at the company, even with Rockwell taking over Siri. An abrupt departure would signal publicly that the AI efforts have been tumultuous — something Apple is reluctant to acknowledge. Giannandrea’s other responsibilities include oversight of research, testing and technologies related to AI. The company also has a team reporting to Giannandrea investigating robotics.

This I find a little surprising. But maybe I shouldn’t. I don’t buy Gurman’s argument that dismissing Giannandrea would “signal publicly that the AI efforts have been tumultuous”. Apple already signaled that publicly when they announced that all of the ambitious features for Siri and Apple Intelligence that were promised for this year’s OS cycle would be postponed until next year’s OS cycle. That’s public tumult. But I mean, you can see for yourself that Apple’s AI efforts have been “tumultuous” by asking Siri on your iPhone, right now, what month it is.

What Apple needs to signal is that they don’t expect to deliver a significantly better Siri without making significant changes to the team behind Siri.

But maybe the answer is as simple as that Giannandrea is good at leading and managing teams doing advanced research that is abstracted from product. So move the products out of his division and into Federighi’s, and put someone who knows how to ship directly in charge of Siri. Leave Giannandrea in charge of a division focused on research and technology. Attention has moved on from “machine learning” to LLMs, but Apple’s machine learning game has gotten very good.

HealthKit as a Model for an Open Semantic Index From Apple ★

Here’s an update I just appended to my post yesterday, after linking to Gus Mueller’s suggestion that Apple open up a semantic index to third-party AI apps:

HealthKit already works a lot like what Mueller is suggesting here (for, say, “SemanticKit”). With explicit user permission — that can be revoked at any time — third party apps can both read from and write to your Health data. Apple does a lot of that itself, both through Apple Watch and from the various activity-related things an iPhone can track, but third-party apps and devices are welcome participants, in a private, easily-understood way.

Nobody is suggesting Apple should give up on AI. Quite the opposite. They really need to go from being a joke to being good at it, fast. But there’s no reason at all they should build out a strategy that relies on Apple doing all of it themselves, and Apple users relying solely on Apple’s own AI. Do it like Health — a model that has proven to be:

• profitable (for Apple itself, selling devices like Watches);
• popular (with users, who actually use it, understand it, and like it);
• private;
• and open to third-party developers, device makers, and medical service providers.

(Thanks to Bill Welense for the suggestion.)

The M1 MacBook Air Lives on at Walmart, Now Just $650 ★

Last March, when Apple introduced the then new M3 MacBook Airs, they moved the base model 13-inch M2 MacBook Air into the magic $999 spot in their own lineup, replacing the M1 MacBook Air. But mid-March it was announced that Walmart would begin selling the M1 MacBook Air — in one tech-spec configuration (8 GB RAM, 256 SSD), but three colors (gold, silver, space gray) for just $700.

This year Apple replaced the entire lineup of MacBook Airs that it sells itself with M4-based models, including the $999 starting-price model. Online, Walmart sells a handful of MacBook models now, at, per Walmart’s brand, slightly lower prices than Apple itself. But the one and only MacBook they seem to stock in their retail stores is the classic wedge-shaped M1 MacBook Air — now down to $650.

It’s over four years old now, and yes, 8 GB RAM and 256 GB of storage are meager, but it’s almost certainly the best new laptop you can buy for that price. Assuming Apple thinks this partnership is a success, eventually they’ll have to replace this with a more recent MacBook Air. But I suspect the main reason it’s still the M1 Air (and hasn’t been replaced by, say, the M2 Air) is not about the specs or performance, per se, but rather simply how it looks. It looks like an older MacBook. Walmart might not get an updated MacBook with a more-recent-than-M1 chip until Apple refreshes the industrial design on its current MacBook Airs.

Display Preferences

Copyright © 2002–2025 The Daring Fireball Company LLC.