 |  |
By John Gruber
Mux — Video API for developers. Build in one sprint or less.
Hannah Knowles, writing for The Washington Post (via Taegan Goddard):
Mamdani won two-thirds of voters under 45 in preliminary exit polls, while Cuomo led him by 10 points with voters 45 and older. The polls also showed an education divide: College graduates backed Mamdani by 55 percent, while voters without college degrees narrowly favored Cuomo.
“By 55 percent” is horrendously unclear writing. It could be misread to suggest that Mamdani won amongst college grads by a 55-point margin. He did not. CNN’s exit poll — the link cited by Knowles above — show Mamdani garnering 57 percent of the vote from college graduates, with Cuomo at 38, and Sliwa 5. Amongst voters without a college degree, it was Cuomo 47, Mamdani 42, and Sliwa 11.
Mandani cruised to an easy win while losing amongst voters without a degree because in New York City, 59 percent of voters yesterday had college degrees.
That level of education in the electorate is not representative of the United States as a whole. In last year’s presidential election (for consistency’s sake, I’m citing exit poll data from CNN), only 43 percent of voters nationwide had college degrees. Kamala Harris beat Trump 56–42 amongst those voters. Amongst the 57 percent of voters without a college degree, Trump won by almost the exact reverse split, 56–43.
Democrats, nationwide, don’t need to make gains with college-educated voters. They need to make gains amongst voters without college degrees. There’s no other demographic gap that is more crucial for Democrats to address. Education trumps race, gender, income, and age. In 2020, Biden won college grads 55–43, and Trump won non-college-grads by a mere 50–48.
WhatsApp:
In addition to reading and responding to messages, for the first time WhatsApp on Apple Watch will now support many requested features:
- Call notifications: You can see who’s calling without needing to look at your iPhone.
- Full messages: You can read full WhatsApp messages on Apple Watch — even long messages are visible directly from your wrist.
- Voice messages: You can now record and send voice messages.
- React to messages: We’ve added the ability to send quick emoji reactions to messages you receive.
- A great media experience: You’ll see clear images and stickers on your Apple Watch.
- Chat history: You can see more of your chat history on screen when reading messages.
All of these features have long been available on the Apple Watch apps for Apple’s Messages and Phone apps. But it’s an interesting sign that Meta sees Apple Watch as an important platform for personal communication. Not just for notifications that you need to act upon using your phone, but for actually using on your watch itself. And I think it speaks to how hard Meta is pushing to make WhatsApp the new universal baseline for texting and calling. By keeping iMessage and FaceTime to its own devices, Apple has ceded this opportunity to WhatsApp, and Meta is trying to capitalize on it.
I know there are many people who spend time wearing their Apple Watch while away from their iPhone — often while working out — who want or even feel they need these features. For me though, one of the things I like least about wearing an Apple Watch is getting badgered on my wrist with notifications. I feel not so much like I need less screen time, but rather that I need less notifications time. I feel good when I have time where I’m unreachable by texts, calls, and news alerts. I spent my recent month-plus semi-hiatus wearing only a mechanical watch, and I didn’t miss the lack of notifications-on-my-wrist at all.
News from Apple’s Podcasts for Creators site, regarding new features in the iOS 26.2 beta releases:
When you supply chapters in your episode description or in your RSS feed, they display in Apple Podcasts. If you submit chapters through your hosting provider, you can include images. For shows in English, when chapters aren’t provided, Apple Podcasts generates them for you and an “Automatically created“ label appears in the chapter list. If you prefer not to use automatically created chapters, you can disable this feature in Apple Podcasts Connect. Learn more about chapters.
It’s unclear to me whether this feature is actually exclusive to iOS/iPhone, or will be available across Apple’s 26.2 OS releases. This strikes me as a great use of AI, but I also think most multi-topic podcasts should include human-created chapters.
Sean Hollister, reporting for The Verge:
The details of how, when, and where Google would charge its fees are complicated, and they seem to be somewhat tailored to the needs of a game developer like Epic Games. Google can charge 20 percent for an in-app purchase that provides “more than a de minimis gameplay advantage,” for example, or 9 percent if the purchase does not. And while 9 percent sounds like it’s also the cap for apps and in-app subscriptions sold through Google Play, period, the proposal notes that that amount doesn’t include Google’s cut for Play Billing if you buy it through that payment system.
That cut will be 5 percent, Google spokesperson Dan Jackson tells The Verge, confirming that “This new proposed model introduces a new, lower fee structure for developers in the US and separates the service fee from fees for using Google Play Billing.” (For reference, Google currently charges 15 percent for subscriptions, 15 percent of the first $1M of developer revenue each year and 30 percent after that, though it also cuts special deals with some big developers.)
If you use an alternative payment system, Google might still get a cut: “the Google Play store is free to assess service fees on transactions, including when developers elect to use alternative billing mechanisms,” the proposal reads. But it sounds like that may not happen in practice: “If the user chooses to pay through an alternative billing system, the developer pays no billing fee to Google,” Jackson tells The Verge.
According to the document, Google would theoretically even be able to get its cut when you click out to an app developer’s website and pay for the app there, as long as it happens within 24 hours.
This seems as clear as mud, other than being music to Epic Games’s ears.
John Dvorak back in 2014, two months before Apple Watch was announced:
I got a lecture from a potential buyer, who will only purchase an iTime as a replacement for the iPhone rather than an accessory. But all evidence leads me to believe this device will be an accessory.
Doing that limits the appeal to people who were promised a sleeker gadget profile, which they desperately need, because they never manage to pare down anything. It’s tablet computing all over again.
If he’d meant that Apple Watch would be like the iPad, in terms of being a durable long-term many-billion-dollars-in-sales-per-quarter platform, he’d have been correct. But he meant that both were duds.
Dvorak is still writing, but alas, only occasionally.
Chance Miller, 9to5Mac:
Apple has launched a dramatic new web interface for the App Store. You can now get the full App Store experience right in your browser, with dedicated pages for the iPhone, iPad, Mac, Vision, Watch, and TV app libraries.
Previously, Apple’s “apps.apple.com” domain simply redirected you to a generic page about the App Store on Apple’s website. Now, it takes you to a full-fledged version of the App Store you can browse on your computer.
This new website is nice, but it’s not the “full” App Store experience, insofar as you can’t buy or download apps from it. It’s more like a full website mirror of the App Store than a web version of the App Store.
John Voorhees, writing at MacStories:
With iOS 26, Apple placed two big buttons onscreen when an alarm went off. One was for stop and the other snooze. That wasn’t a big deal for many of the alarms you set throughout the day, but when you’re waking up in the morning blurry-eyed, two big buttons stacked on top of each other weren’t ideal. For a lot of users, it was a toss-up whether stabbing at their iPhone through a morning haze would stop their alarm or snooze it.
With iOS and iPadOS 26.1, the “Stop” button for an alarm set in the system Clock app now requires a slide to stop gesture, which echoes the Slide to Unlock gesture of the original iPhone. The more deliberate gesture is a good move on Apple’s part. I can’t imagine someone tapping and sliding their finger to stop an alarm by accident.
This is a clever little change. I enjoy that it harks back to the original iPhone’s slide-to-unlock.
Update: If, for whatever reason, you don’t like this slide-to-stop feature, you can turn it off by toggling this option in Settings: Accessibility → Touch → Prefer Single-Touch Actions.
Mark Gurman, in his weekly Power On column for Bloomberg (paywalled, alas):
Even with the rosy sales forecast, the road ahead won’t be easy. Apple is betting heavily on the new Siri, which will lean on Google’s Gemini model and introduce features like AI-powered web search. But there’s no guarantee users will embrace it, that it will work seamlessly or that it can undo years of damage to the Siri brand.
And then, down below in his “Post Game Q&A”:
Q: Is Apple still planning to use Google Gemini to power the new Siri?
A: As I’ve reported a few times now, Apple is paying Google to create a custom Gemini-based model that can run on its private cloud servers and help power Siri. Apple held a bake-off this year between Anthropic and Google, ultimately determining that the former offered a better model but that Google made more sense financially (partly due to the tech giants’ preexisting search relationship). I don’t expect either company to ever discuss this partnership publicly, and you shouldn’t expect this to mean Siri will be flooded with Google services or Gemini features already found on Android devices. It just means Siri will be powered by a model that can actually provide the AI features that users expect — all with an Apple user interface.
This is quite the aside to tuck into a one-paragraph Q&A item. First, I love the idea that Apple is pursuing technical excellence as a top priority for the next-gen LLM-powered Siri. If Apple winds up using its own models, it should be because those models are truly competitive with the best models on the market. And if they can work out a deal to use models from Google because those models are technically superior to Apple’s own, they should.
It’s kind of wild though to think that, if this comes to pass, neither company will publicly acknowledge the arrangement. I believe it’s possible — but it would be odd. Right now Apple has a public partner for Apple Intelligence: optional integration with ChatGPT. Apple labels that integration as an “extension”, and has repeatedly stated — including as recently as last week — that they’re looking at other partners to add. The most obvious partner Apple could add — one that Craig Federighi mentioned by name on the day that Apple Intelligence was announced at WWDC 2024 — would be Google Gemini.
If what Gurman is reporting comes to pass, and Apple’s own cloud-based LLM technology is a white-label version of Google Gemini, it’d be pretty weird if that ships and Google Gemini still is not a named extension partner for Apple Intelligence. But it would also be a little weird if Google Gemini does become a named partner for Apple Intelligence alongside ChatGPT, while Apple’s own default cloud-based Apple Intelligence is powered by Gemini’s models.
Anthony Castrovince, writing for MLB (News+ link):
Behind a stunning, game-tying swat from Miguel Rojas in the top of the ninth, a first-of-its-kind, go-ahead blast from Will Smith in the top of the 11th and the absurd extra work World Series MVP Yoshinobu Yamamoto provided on zero days’ rest, the Dodgers broke Toronto hearts with their comeback 5-4 victory in a game that merited its own month on the MLB calendar.
The Dodgers are MLB’s first repeat champs since the 1998-2000 Yankees, and the four-hour, seven-minute, extra-innings affair it took to decide that was a fitting end to a true Fall Classic in which these two clubs exhausted each other — not just in the 18-inning epic at Dodger Stadium in Game 3 but throughout a Series in which they both had to empty the tank.
When a series goes to game 7, every fan hopes it’s a good game. But this was a great game — as good as baseball gets. Great pitching, clutch hitting, and some amazing fielding plays. Simply riveting to watch. The Blue Jays came within an inch or two of winning the Series on this play in the bottom of the 9th (and I think they would’ve won on that play if Kiner-Falefa had run through home plate rather than inexplicably sliding).
My thanks to Jaho Coffee Roaster for sponsoring this week at DF. Jaho has been family-owned since 2005, and they’re guided by their slogan: “Live Slow”. Jaho knows that great coffee takes care. From sourcing small-lot single origins to blending coffees for balance, they small-batch roast their award-winning coffees in Salem and Tokyo.
For the at-home coffee drinker, they roast to order and pack the same coffees brewed and served in all of their cafés. For the office worker, Jaho is proud to be a wholesaler with select partners across the nation and in Japan. Jaho was kind enough to send me a few bags of their beans, and I can vouch that they roast excellent coffee — the kind of tasty beans where, when I finish my last morning cup, I’m tempted to brew a little more even though I know I’m fully caffeinated.
For the month of October, every year, Jaho donates all online coffee bean sale profits to Susan G. Komen for Breast Cancer Awareness. (The pink in their brand colors originated from this partnership.) For this DF sponsorship, they’re carrying that promotion for all online sales through the end of day on Monday, November 3.
Jaho ships their fresh beans nationwide, and they’re offering a special deal for DF readers: take 20% off with code DF. Give up bad coffee for good, and support a great cause at the same time.
The president of the United States, on his blog:
Seth Meyers of NBC may be the least talented person to “perform” live in the history of television. In fact, he may be the WORST to perform, live or otherwise. I watched his show the other night for the first time in years. In it he talked endlessly about electric catapults on aircraft carriers which I complain about as not being as good as much less expensive steam catapults. On and on he went, a truly deranged lunatic. Why does NBC waste its time and money on a guy like this??? - NO TALENT, NO RATINGS, 100% ANTI TRUMP, WHICH IS PROBABLY ILLEGAL!!!
The funny part about Trump wildly flailing that Late Night With Seth Meyers is somehow “probably illegal” is that the very sentence of the segment that so upset Trump begins with this: “Donald Trump called criticism of his trip to Asia ‘almost treasonous’ and threatened to send active duty military into US cities. For more on this, it’s time for ‘A Closer Look’.”
Paul Lukas, founder of Uni Watch:
Due to a perfect storm of negative developments, I have reluctantly come to the unfortunate conclusion that continuing to publish Uni Watch is no longer viable. This will be the site’s final post.
Yes, I’m serious. And no, this isn’t a Halloween-related prank. Uni Watch is shutting down, for real.
I realize this news probably comes as a shock and that you no doubt have lots of questions, so let’s shift into Q&A mode. [...]
Will the site’s archive remain on the web?
No, unfortunately. Most of the archive — everything but the past few days’ worth of content — has already been taken down. The rest of the site, including this post, will be taken offline soon, probably around next Wednesday.
26 years is a hell of a run (dating back to 1999, a few years before Uni Watch became a standalone site), but I don’t understand why sites don’t leave their archives standing when they close down. It shouldn’t cost much to keep the domain name registered and a static version of the site’s archive online.
Uni Watch, to me, epitomized a certain mindset from the early web. To wit, that there ought to be a blog (or two or three) dedicated to every esoteric interest under the sun. You want to obsess about sports team uniform designs? Uni Watch was there. For a good long stretch, there seemingly was a blog (or two or three) dedicated to just about everything. That’s starting to wane. New sites aren’t rising to take the place of retiring ones.
Special guest Brian Mueller, developer of Carrot Weather, joins the show to commemorate the 10th anniversary of his utterly ridiculous but totally serious weather app.
Sponsored by:
Tim Bray:
Last night I had a very strange experience: About two thirds of the way through reading a Web page about myself, Tim Bray, I succumbed to boredom and killed the tab. Thus my introduction to Grokipedia. Here are early impressions.
My Grokipedia entry has over seven thousand words, compared to a mere 1,300 in my Wikipedia article. It’s pretty clear how it was generated; an LLM, trained on who-knows-what but definitely including that Wikipedia article and this blog, was told to go nuts.
Putting aside the political slant of Grokipedia, a 1,300-word article being better than a 7,000-word one exemplifies the current shortcomings of LLMs as creative engines (as opposed to serving as mere tools in the arsenal of human creators).
The French philosopher and mathematician Blaise Pascal famously quipped: “I would have written a shorter letter, but I did not have the time.” No encyclopedia in history has been written with less time or effort than Musk’s LLM-generated vanity project. Verbosity is not the worst of Grokipedia’s deficiencies, but it’s one of them. The more its entries stray from simply regurgitating the equivalent entry in Wikipedia, the more they suffer from verbal diarrhea.
(My own Grokipedia entry is just a clone of my Wikipedia entry, with a few mistakes added, including one in the first sentence regarding the creation of Markdown.)
Jason Koebler, writing at 404 Media:
Wednesday, as part of his ongoing war against Wikipedia because he does not like his page, Elon Musk launched Grokipedia, a fully AI-generated “encyclopedia” that serves no one and nothing other than the ego of the world’s richest man. As others have already pointed out, Grokipedia seeks to be a right wing, anti-woke Wikipedia competitor. But to even call it a Wikipedia competitor is to give the half-assed project too much credit. It is not a Wikipedia “competitor” at all. It is a fully robotic, heartless regurgitation machine that cynically and indiscriminately sucks up the work of humanity to serve the interests, protect the ego, amplify the viewpoints, and further enrich the world’s wealthiest man. It is a totem of what Wikipedia could and would become if you were to strip all the humans out and hand it over to a robot; in that sense, Grokipedia is a useful warning because of the constant pressure and attacks by AI slop purveyors to push AI-generated content into Wikipedia. And it is only getting attention, of course, because Elon Musk does represent an actual threat to Wikipedia through his political power, wealth, and obsession with the website, as well as the fact that he owns a huge social media platform.
In season 10 of Curb Your Enthusiasm, Larry David gets into an argument with Mocha Joe, the owner of an eponymous coffee shop. David leases the space next door and opens Latte Larry’s, a copycat “spite store” cafe. Grokipedia reminds me of this, except that Larry David is genuinely funny and (in real life, as opposed to his Curb alter ego) at least somewhat self-aware.
Claudie Moreau, reporting for Euractiv:
Earlier in their presidency, Denmark had revived a controversial provision in the draft law that would mean online platforms — such as messaging apps — could be served with mandatory CSAM detection orders, including services protected by end-to-end encryption. However opposition from several other EU countries derailed any agreement in the Council.
Today, Danish Justice Minister Peter Hummelgaard told local press that the Council presidency would move away from mandatory detection orders — and instead support CSAM detections remaining voluntary.
Sanity prevails.
Apple Support:
This update includes enhancements to the Shortcuts app across all platforms, including new intelligent actions and an improved editing experience. Shortcuts on macOS now supports personal automations that can be triggered based on events such as time of day or when you take actions like saving a file to a folder, as well as new integrations with Control Center and Spotlight.
TechRadar, summarizing this German-language report from Handelsblatt:
The International Criminal Court (ICC) is looking to replace its internal work environments to move away from US-made software in fear of retaliation from the US administration.
The Microsoft software currently used in the Hague-based ICC is likely to be replaced with Open Desk, a German collaboration software alternative which is open source, meaning developers have chosen to release the source code — opening it up to scrutiny and often meaning that bugs and vulnerabilities are picked up quickly by the community. [...]
Early in 2025, Chief Prosecutor for the ICC Kamrin Khan, after being hit with sanctions from the Trump administration, was disconnected from his email service. This action was thought to be from Microsoft supporting US sanctions — although the firm denied this, with a spokesperson stating; “at no point did Microsoft cease or suspend its services to the ICC.”
This sparked fears that US tech firms could flip a ‘kill switch’ and cut digital services on orders of Trump — outlining the need to become less dependent on US technology, with firms like Google, Microsoft, and Amazon dominating Europe’s digital services and cloud markets.
This is what makes US technology firms’ support for Trump so confounding. It’s easy to see the short-term benefits (e.g. tariff exemptions), but just as easy to see the long-term reputational harm. The US was long seen as the most trustworthy powerful nation in the world. Now it’s one of the least trustworthy. Why would companies like Microsoft, Apple, Google, and Amazon tie their own reputations to Trump’s? Trump’s reign of abject corruption, ignorance, and personality-driven retribution — and these companies’ support for all of it — will be remembered long after Trump himself is gone.
I’m not calling on these companies to outright oppose the Trump administration. But there’s a lot of space between outright opposition and helping to fund Trump’s illegal vanity ballroom on the White House grounds.
Kif Leswing at CNBC interviewed Tim Cook ahead of yesterday’s Apple earnings report:
Cook said that the company still plans to release an updated version of Siri next year, and said that there were more forthcoming partnerships like the company’s agreement to integrate OpenAI’s ChatGPT into Apple Intelligence.
“Our intention is to integrate with more people over time,” Cook said.
And from Cook’s prepared remarks at the start of yesterday’s analyst call:
“We’re also excited for a more personalized Siri. We’re making good progress on it, and as we’ve shared, we expect to release it next year.”
No news here, but worth noting that Cook claims both the next-gen “more personalized” Siri and deals with AI partners other than OpenAI are still on track. But Craig Federighi hinted at adding Google Gemini as an option alongside ChatGPT for Siri all the way back at WWDC 2024, within a few hours of Apple Intelligence being announced. Still nothing. 16 months later and ChatGPT remains the one and only Apple Intelligence partner.
Travis M. Andrews, Jeremy B. Merrill, and Shelly Tan, reporting for The Washington Post (News+ link:)
“We had spent way too much on programming that doesn’t bring in any revenue,” Richard Grenell, a Trump ally and former ambassador to Germany, told the Washington Reporter, a conservative media outlet, in late March. According to Grenell, the center hadn’t been making money. It was too woke and niche. The new team was, in Trump’s words, going to make it “hot” again.
Nearly nine months after Trump became chair of the center and more than a month into its main season, ticket sales for the Kennedy Center’s three largest performance venues are the worst they’ve been in years, according to a Washington Post analysis of ticketing data from dozens of recent shows as well as past seasons. Tens of thousands of seats have been left empty.
Since early September, 43 percent of tickets remained unsold for the typical production. That means that, at most, 57 percent of tickets were sold for the typical production — and some tickets may have been “comps,” which are given away, often to staff members or the press. That compares with 93 percent sold or comped in fall 2024 and 80 percent in fall 2023.
Crickets chirping.
Speaking of vaporware, Dominic Preston at The Verge on the T1 Trump phone, which was announced back in June:
In fairness, for months now, the store page has only promised an arrival “later this year,” a change made at the same time Trump Mobile stopped claiming the T1 would be made in America. That gives the company two more months to release it and still pretend it’s on time.
Trump Mobile never responded to my request last month for an update on the phone’s release date, and it hasn’t replied to my latest email either. People of lesser faith might worry that this phone is no more than vaporware, but I refuse to give up. Place your bets now on whether I’ll be back here in another month’s time, still asking: where is the Trump phone?
Joanna Stern, writing for The Wall Street Journal (gift link):
It was wild to watch. Sure, Neo nearly toppled over while closing the dishwasher, took two minutes to fold the shirt and twisted its arm attempting to dance the Macarena. But shhh. Remember the rule. Oh, did I mention Neo had a human puppet master, controlling it with a VR headset?
Neo’s creator, 1X Technologies, is making the Rosie-the-Robot dream: some of the first humanoid housekeepers. Starting Tuesday, you can apply to its early adopter program and preorder one for $20,000, with delivery expected in 2026. The company will also offer a $499 monthly rental plan with a six-month minimum commitment.
Just one hidden cost: your privacy. For now, you’ll need to be cool with a company representative potentially peering through the robot’s camera eyes to get chores done. There are guardrails, including controls over when and what the operator can do.
As usual, Stern made a delightful short film to accompany her article, which is also available on YouTube.
The argument from CEO Bernt Børnich is that they’re using the videos from the current state of Neo, where its actions are entirely remote-controlled by employees of 1X Technologies, to train its autonomy. I call bullshit. This looks to me like nothing but a scam. It’s not autonomous at all, I don’t believe this company is going to achieve any practical degree of autonomy with this product, and even while it’s remote-controlled by human operators, it’s slow and clumsy.
See also: Marques Brownlee, who smells vaporware as clearly as I do: “There seems to be a bit of a lost art in waiting for a tech product to be actually finished before announcing and unveiling it.”
Jason Snell, writing at Six Colors:
In the post-results call with financial analysts, Wamsi Mohan of Bank of America asked Cook for a little more detail about Apple’s search revenue, given its lucrative deal with Google, and whether that revenue growth might decelerate if Google’s search traffic were to be impacted by the growth of AI. Cook’s response was, if I do say so myself, an all-timer for these calls:
Cook: This is Tim. The advertising category, which is a combination of third-party and first-party, did set a record during the quarter.
Mohan: Okay, and sorry, just to be clear, both Apple’s own internal advertising and within the licensing individually set records?
Cook: I actually I’m not saying that. I’m just saying that the combination of the two set a record. We don’t divulge — I’m dodging the question intentionally because we don’t split it at that level.
Look, these calls are almost entirely Apple execs dodging the questions of fiscal analysts. At least Tim Cook admitted it this time. You want to know how much Google is paying us and if that’s growing or shrinking? Well, I’m not gonna tell you!
If Apple’s quarterly analyst calls were a podcast, “Dodging the Question Intentionally” would be a great episode title for this one.
Matt Rosoff, writing for The Register:
If Microsoft owns 27 percent of OpenAI, it stands to reason under equity accounting that it bears 27 percent of OpenAI’s losses. Microsoft’s admission that it shaved $3.1 billion off its net income to account for its share of OpenAI losses therefore suggests OpenAI lost about $11.5 billion during the quarter. Microsoft declined to comment beyond confirming that the $3.1 billion loss “this year” referred to Microsoft’s current fiscal year, which started July 1, not the calendar year. So that’s a quarterly loss, not a nine-month loss.
That’s a humongous number for OpenAI given it reportedly generated only $4.3 billion in revenue for the first half of the year, but a sum that won’t hurt Big Daddy Redmond too much given it earned $27.7 billion in net income in the last quarter alone.
A pre-IPO startup is a different animal from an established publicly-held corporation, but an $11.5 billion quarterly loss is quite different from the $20–30-ish billion quarterly profits booked by the big six.
Apple Newsroom:
Apple today announced financial results for its fiscal 2025 fourth quarter ended September 27, 2025. The Company posted quarterly revenue of $102.5 billion, up 8 percent year over year. Diluted earnings per share was $1.85, up 13 percent year over year on an adjusted basis.
“Today, Apple is very proud to report a September quarter revenue record of $102.5 billion, including a September quarter revenue record for iPhone and an all-time revenue record for Services,” said Tim Cook, Apple’s CEO.
Looking at Apple’s Consolidated Statement (PDF), the numbers look great across the board year-over-year: iPhone up 6%, Mac up 13%, iPad even, Wearables/Home even, and Services up 15%. Services now generates more revenue ($28.8 billion) than Mac, iPad, and Wearables/Home combined ($24.7 billion).
Six Colors, as usual, has Apple’s quarter illustrated in charts.
Here’s a comparison of net income (profit) from Apple’s peers for their most recent quarters:
Joe Rosensteel:
I have no plan to purchase a GM vehicle, but I do rent cars. GM makes up a sizable portion of rental car fleets. At some point in the future those cars will no longer support CarPlay. I’m not going to sign up for a GM federated ID that stores my login credentials in their cloud. I’m not going to individually sign into apps in the car like Google Maps with my Google ID that I use for way more than just navigation. There’s no chain of trust with me and this random car from GM. No convenience that is achieved in exchange for increased exposure risk for storing my sensitive data in a car I don’t own.
If GM goes through with this abandonment of CarPlay, I don’t see how they’ll continue to sell any vehicles to rental agencies. I would never rent a car without CarPlay, and I would never consider signing up for a GM cloud service just to drive a rental car. Complete dealbreakers.
Tom Ellison, at McSweeney’s:
How are my competitors doing, the ones you all insisted students use instead of me? That’s right, they were supposed to go to the American Journal of Social Sciences, Powered by OpenAI. Or museums, like the Smithsonian’s Charlie Kirk Shrine to American Greatness. I guess they can still count on credible journalism, once they get past the paywall for Palantir Presents: The Washington Post, so they read the Pulitzer-Bezos Prize–winning work of coeditors-in-chief Bari Weiss and Grok.
Jay Peters, writing for The Verge:
However, despite Elon Musk promising that Grokipedia would be a “massive improvement” over Wikipedia, some articles appear to be cribbing information from Wikipedia. At the bottom of the page for the MacBook Air, for example, you can see this message: “The content is adapted from Wikipedia, licensed under Creative Commons Attribution-ShareAlike 4.0 License.” In some cases, the cribbing goes farther than a rewrite: I’ve also seen that message on pages for the PlayStation 5 and the Lincoln Mark VIII, and both of those pages are almost identical — word-for-word, line-for-line — to their Wikipedia counterparts.
“Even Grokipedia needs Wikipedia to exist,” Lauren Dickinson, a spokesperson for the Wikimedia Foundation, the non-profit that operates Wikipedia tells The Verge. You can read Dickinson’s full statement in full at the end of this article.
At launch, Grokipedia is to Wikipedia as a chewed piece of gum is to a fresh piece of gum still in its wrapper. And imagine that the gum was chewed by someone with a dipping tobacco habit.
Nick Vadala, reporting for the Philadelphia Inquirer:
Longtime WMMR-FM host Pierre Robert was found dead in his home Wednesday. He was 70.
Robert’s surname, I must point out, rhymes with Pierre (and with Colbert).
A native of Northern California, Robert joined WMMR as an on-air host in 1981. He arrived in the city after his previous station, San Francisco’s KSAN, switched to an Urban Cowboy format, prompting him to make the cross-country drive to Philadelphia in a Volkswagen van. “I came because of a relationship,” he told The Inquirer last year. “I was in love. The love part didn’t work out, but the job part did.”
As a newly minted Philadelphian, Robert began working at a local health food store as he interviewed for radio jobs around town, but found little luck initially. One day, while dining at Astral Plane, a long-closed restaurant formerly on Lombard Street, he introduced himself to WMMR program director Joe Bonnadonna and announcer Charlie Kendall, and despite getting on well with the pair, he learned there were no openings at the station.
But weeks later, he received a letter from Bonnadonna, and interviewed for a job at the station during a concert from Philly rock band The Hooters at the Chestnut Cabaret. He soon started working in the station’s music library and office making $3.50 an hour, and later began appearing on the air.
There’s no more Philadelphia a Philadelphia origin story than a radio host interviewing for his job during a Hooters concert at the Chestnut Cabaret — and then going on to stay at the same station for 44 years. Impossible for me to overstate just how much Robert’s voice was the voice of music for me and my entire friend group growing up and even through college. You tuned the dial to 93.3 FM and left it there.
My favorite bit of his was an obscure one, a character named Reginald the Butler. Robert always had Reginald on during the holidays, while spinning Christmas rock songs. But here’s a classic segment from 1988 with Robert and Reginald interviewing David Lee Roth, who was then on a solo tour and about to play the Spectrum.
Rest in peace, my fellow citizen.
James Thomson:
I’ve released a small PCalc 4.11.1 update that’s out now for the Mac.
There was a bug with the theme getting reset, which I could have fixed in five minutes, but I ended up doing what I should have done over three decades ago, and added a dedicated section to the settings that puts all the visual customisation in one place.
No more having to search for all this stuff in a submenu somewhere!
After the glum news this week regarding Nisus Writer, it feels good to link to a similarly-aged Mac app that’s still thriving. If you’ve never tried PCalc, you’re missing out.
Tim Hardwick, writing for MacRumors:
The Apple Maps EV Routing option will allow Toyota BEV users to plan travel routes that include stops for charging. Without it, drivers would have had to exit out of CarPlay in order to create a route that included charging stops.
Apple Maps’ EV Routing feature uses real-time data from the vehicle to guide drivers to their destinations more efficiently, automatically suggesting charging stops when needed. The system takes into account elevation changes and other driving conditions to decide when a recharge is necessary. If the vehicle’s battery level becomes too low, Apple Maps will automatically direct the driver to the nearest compatible charging station.
Meanwhile GM CEO Mary Barra is spending her lunch hour eating another jar of paste.
The Onion, in February 2004: “Fuck Everything, We’re Doing Five Blades”.
The Economist:
Presidents’ popularity tends to wane. In his second term Donald Trump’s has fallen faster than that of his recent predecessors.
Since modern polling began most presidents have started their terms with positive net approval ratings (the share of voters who approve of their job performance minus the share who disapprove). Both of Mr Trump’s terms began with public opinion split nearly evenly. In both cases his net approval rating quickly turned negative. Now it is -18, the lowest it has been since his inauguration — and three percentage-points lower than at any point in his first term.
State-by-state, Trump is only above water in nine states: Idaho, Wyoming, West Virginia, Montana, North Dakota, Utah, Kentucky, Tennessee, and Alabama.
Donald Trump’s presidential approval rating fell in recent days, tying the lowest level of his term, as more Americans frowned on his handling of the cost of living, according to a new Reuters/Ipsos poll.
The three-day poll, which concluded on Sunday, showed 40% of Americans approve of the Republican leader’s job performance, compared to 42% in an October 15-20 Reuters/Ipsos poll. Trump’s popularity has been within a percentage point or two of its current level in every Reuters/Ipsos poll since mid-May. The share of people who say they disapprove of his performance has grown, from 52% in a May 16-18 poll to 57% in the latest survey.
So the Economist has him at -18, Reuters at -17.
Other Mac-related news from OpenAI last week:
Sky is a powerful natural language interface for the Mac. With Sky, AI works alongside you, whether you’re writing, planning, coding, or managing your day. Sky understands what’s on your screen and can take action using your apps.
We will bring Sky’s deep macOS integration and product craft into ChatGPT, and all members of the team will join OpenAI.
Two of the founders of Software Applications Incorporated, the company behind Sky, are Ari Weinstein and Conrad Kramer, who a decade ago co-created Workflow, which Apple acquired in 2017 and turned into Shortcuts.
Federico Viticci got an advanced look at Sky and wrote a glowing preview back in May.
OpenAI, one week ago:
Today we’re introducing ChatGPT Atlas, a new web browser built with ChatGPT at its core.
AI gives us a rare moment to rethink what it means to use the web. Last year, we added search in ChatGPT so you could instantly find timely information from across the internet — and it quickly became one of our most-used features. But your browser is where all of your work, tools, and context come together. A browser built with ChatGPT takes us closer to a true super-assistant that understands your world and helps you achieve your goals.
A few minutes into the 22-minute introduction video, Ben Goodger,1 engineering lead for Atlas, says:
“We wanted to make sure that Atlas didn’t feel like your old browser, just with a chat button that was bolted on. But instead, we made ChatGPT the beating heart of Atlas.”
After giving it a try over the last week, to me Atlas feels like … Chrome with a chat button bolted on. I do not see the appeal, at all, despite being a daily user of ChatGPT. Atlas offers nothing to me that’s better than using Safari as a standalone browser and ChatGPT’s excellent native Mac app as a standalone AI chatbot. But, for me, my browser is not “where all of [my] work, tools, and context come together”. I use an email app for email, a notes app for notes, a text editor and blog editor for writing and programming, a photos app for my photo library, a native feed reader app for feed reading, etc. My web browser is for browsing pages on the web. Perhaps this sort of browser/chat hybrid appeals better to people who live the majority of their desktop-computing lives in browser tabs.
The main interface isn’t a combo search/location field, but rather a chat/location field. Instead of getting search results for a query, you get a chat response. If I wanted this I’d just ask my prompt in ChatGPT. Oftentimes — usually, even — I really do want a list of search results, and I want them fast. ChatGPT responses in Atlas are not a list of web pages, and are — compared to Google Search or my preferred search engine, Kagi — very slow. ChatGPT is many things but a good search engine replacement it is not. But that seems to be the entire premise of Atlas.
Atlas offers an agent mode where it actually surfs the web for you. One of the demos from their launch video involved getting a list of ingredients from a recipe on a web page, and then allowing Atlas to buy all those ingredients for you. That seems crazy to me. Do not want.
Atlas is a Chromium browser, supports Chrome extensions, and but currently is only available for the Mac. It’s not particularly Mac-like though, as Michael Tsai notes:
Alas, it doesn’t support AppleScript and has System Settings–style preferences.
System Settings-style preferences are certainly better than Chrome-style “settings in a web page tab”, though. Also, in my testing, Atlas doesn’t make good use of Apple Passwords for autofill.
ChatGPT is running a promotion that offers users increased rate limits if they make — and keep — Atlas their default web browser. I’ve never before seen a web browser offer any sort of incentive like this for making it your default. This promotion strikes me as simultaneously clever and icky.
Simon Willison’s initial thoughts echo my own:
I continue to find this entire category of browser agents deeply confusing.
The security and privacy risks involved here still feel insurmountably high to me — I certainly won’t be trusting any of these products until a bunch of security researchers have given them a very thorough beating. [...]
I also find these products pretty unexciting to use. I tried out agent mode and it was like watching a first-time computer user painstakingly learn to use a mouse for the first time. I have yet to find my own use-cases for when this kind of interaction feels useful to me, though I’m not ruling that out.
Lastly, Anil Dash’s assessment is rather scathing, “The Browser That’s Anti-Web”:
In the demo for Atlas, the OpenAI team shows a user trying to find a Google Doc from their browser history. A normal user would type keywords like “atlas design” and see their browser show a list of recent pages. They would recognize the phrase “Google Docs” or the icon, and click on it to get back to where they were.
But in the OpenAI demo, the team member types out:
search web history for a doc about atlas core design
This is worse in every conceivable way. It’s slower, more prone to error, and redundant. But it also highlights one of the biggest invisible problems: you’re switching “modes”. Normally, an LLM’s default mode is to create plausible extrapolations based on its training data. Basically, it’s supposed to make things up. But this demo has to explicitly walk you through “now it’s time to go search my browser history” because it’s coercing the AI to look through local content.
Chat is a great interface for, well, chatting. People love texting. And it turns out that chat conversations are a very good user interface for interacting with LLMs. We humans enjoy texting with other humans, and we enjoy texting with LLMs. But typed-out text commands are not a good user interface at all for browsing the web. We had an entirely text-based Internet before the World Wide Web, and the point-and-click visual metaphor of the Web won out.
Dash, later on:
It’s no coincidence that hundreds of people who work at OpenAI, including many of the most powerful executives, are alumni of Facebook/Meta, especially during the era of many of that company’s most egregious abuses of people’s privacy. In the marketing materials and demonstrations of Atlas, OpenAI’s team describes the browser as being able to be your “agent”, performing tasks on your behalf.
But in reality, you are the agent for ChatGPT.
During setup, Atlas pushes very aggressively for you to turn on “memories” (where it tracks and stores everything you do and uses it to train an AI model about you) and to enable “Ask ChatGPT” on any website, where it’s following along with you as you browse the web. By keeping the ChatGPT sidebar open while you browse, and giving it permission to look over your shoulder, OpenAI can suddenly access all kinds of things on the internet that they could never get to on their own.
This jibes with my impression after giving Atlas a try. The point of it doesn’t seem to be to provide a better web browser for me to use, but rather, to provide ChatGPT with the personal context of my digital life that it otherwise couldn’t get.
That last point raises the question of just how stable we should consider the Apple-OpenAI partnership for ChatGPT-backed Apple Intelligence features. Apple’s goal for a “more personalized Siri” — the whole thing Apple promised at WWDC 2024 but had to postpone for a full year early this year — is for the ecosystem of native apps on Apple platforms, particularly iOS and MacOS, to serve as the personal knowledge context for personalized AI features through App Intents. That’s the basis for the “When is my mom’s flight arriving?” type of interaction that Apple has promised, but still has not delivered. The premise of Atlas (and its brethren AI-integrated browsers like The Browser Company’s Dia and Perplexity’s Comet) is that you should live your entire desktop computing life inside your browser, which in turn will give the AI agent that is integrated with your browser the contextual knowledge for your entire life.
OpenAI’s ambitions are clearly at odds with Apple’s.
OpenAI’s advantage here is that ChatGPT is the most popular LLM chatbot in the world, by far. Apple doesn’t even have an LLM chatbot of its own, let alone a good or popular one. But Apple’s advantage is a big one: most people don’t live their digital lives on desktop computers, where it’s an option to do most things in a web browser. Most people’s primary computing devices are their phones — and even for people whose primary devices are desktop computers, their phones are much-used satellite devices. And on both iOS and Android alike, people live their mobile digital lives through native apps, not websites. ★
Goodger is a titanic figure in the web browser world, having helped create Mozilla Firefox in the early 2000s, and then joining Google in 2005 to help create Chrome. I noted last year that Goodger leaving Google for OpenAI was a pretty clear sign that OpenAI was creating its own web browser. ↩︎
Joe Kissell, writing at TidBITS:
For more than a year, we’ve heard scattered complaints: problems with Nisus Software’s website, particularly the user discussion forum; slow or absent responses to support requests; assorted bugs; and other issues. But earlier this week, on 22 October 2025, the reports changed to: “Did you know the Nisus website is completely down, and that Nisus Writer is no longer in the Mac App Store? Does this mean Nisus is out of business?”
On the one hand: The site is back online as I write this. The app still works. I’m writing the first draft of this article in Nisus Writer Pro on a Mac running macOS 26 Tahoe, and it’s fine. You can still download it and buy a license. At least one person is actively involved in the company, to some extent. It’s (mostly) alive!
On the other hand: All available evidence suggests that development and support for Nisus Writer have ceased, and barring some new information, its future is doubtful. It’s (mostly) dead!
I’m going to tell you what I know. (Well, most of what I know.) I’m also going to speculate a bit, because despite my best efforts, I have been unable to obtain verifiable information about certain topics, though I have a pretty good idea of what’s likely the case.
Seems like an ignominious demise for a once-great app. Nisus Writer has been an acclaimed Mac-only (and Mac-assed) word processor since 1989. I never got into it, but I could always see the appeal. Nisus had a macro language for automation and regex-style advanced search and replace. But when I wanted features like those, I wanted them in a plain text editor, not a word processor, so I got into BBEdit.
Sam Tobin, reporting last week for Reuters:
Apple abused its dominant position by charging app developers unfair commissions, a London tribunal ruled on Thursday, in a blow which could leave the U.S. tech company on the hook for hundreds of millions of pounds in damages. The Competition Appeal Tribunal (CAT) ruled against Apple after a trial of the lawsuit, which was brought on behalf of millions of iPhone and iPad users in the United Kingdom.
The CAT ruled that Apple had abused its dominant position from October 2015 until the end of 2020 by shutting out competition in the app distribution market and by “charging excessive and unfair prices” as commission to developers. [...] The case had been valued at around 1.5 billion pounds ($2 billion) by those who brought it. A hearing next month will decide how damages are calculated and Apple’s application for permission to appeal.
Dan Moren and I discussed this at some length in the new episode of The Talk Show that dropped over the weekend. What makes this ruling interesting isn’t that it’s particularly significant or different from other regulatory/antitrust investigations around the world. It’s the fact that it’s completely in line with other regulatory/antitrust investigations regarding the App Store (and Play Store) from around the world.
When is the last time an investigation regarding the legality of the App Store’s dominant market position went in Apple’s favor, in any country? I can’t recall one. Apple is clearly fighting a losing battle here. Whether Apple ought to be losing all these legal and regulatory battles regarding the App Store is, from a strategic standpoint, almost irrelevant. The obvious fact is, they are losing them.
Apple has approached all this regulatory conflict from a perspective that they’re right, and the regulators are wrong. That the App Store, as Apple wants it, is (a) good for users, (b) fair to developers, and (c) competitive, not anti-competitive, legally. But even if Apple is correct about that, at some point, after being handed loss after loss in rulings from courts and regulatory bodies around the globe, shouldn’t they change their strategy and start trying to offer their own concessions, rather than wait for bureaucrat-designed concessions to be forced upon them?
However Apple thinks all of this should work out is not the way it is working out. The best time to adjust the rules of the App Store — its exclusivity on app distribution for the entire iOS platform, the exclusivity of Apple’s IAP for purchasing digital content, the commission percentage splits on IAP — was over a decade ago. The next best time to make those adjustments is now. ★
Joe Rossignol, MacRumors:
Apple might be preparing iPad apps for Pixelmator Pro, Compressor, Motion, and MainStage, according to new App Store IDs uncovered by MacRumors contributor Aaron Perris. All four of the apps are currently available on the Mac only. A quick overview of each app:
- Pixelmator Pro: Professional image editing app acquired by Apple earlier this year
- Compressor: Final Cut Pro companion app for compressing audio and video files
- Motion: Final Cut Pro companion app for creating 2D/3D titles, transitions, and effects
- MainStage: Logic Pro companion app for live performances
There is already a less-capable Pixelmator app available for the iPad and iPhone.
Interesting though that — just like Final Cut and Logic — these new pro apps are reportedly iPad-only, with no support for iPhone.
Also: still no Xcode, even for iPad.
Mark Gurman, in his weekly paywalled Power On column for Bloomberg:
I reported a few years ago that Apple was working to bring more advertising to iOS. Well, now that effort is gaining traction — with a plan to start the ads as early as next year. The company is focusing on Apple Maps, which will allow restaurants and other businesses to pay to have their details featured more prominently within the app’s searches.
The concept is quite similar to Search Ads inside of the App Store, where developers can pay for their software to appear in a promoted slot based on user queries. I’m told the Maps version will have a better interface than what Google and other companies offer inside of mapping services. The Apple approach also will leverage AI to ensure that results are relevant and useful.
The big risk Apple faces here is a potential consumer backlash.
I don’t love the ads in the App Store, but I don’t hate them. They’re restrained, and clearly labeled. I do, however, despise the ads in Apple News. They’re low-quality, distracting, highly repetitive, and appear far too frequently within articles.
Joe Rosensteel, writing at Six Colors, regarding the demise of Apple’s Clips app:
It’s not that it was completely inept, but it was an aimless showcase to demonstrate what Apple could do. It withered over the course of eight years before it was quietly killed.
At no point did it supplant iMovie for iOS as the fun, easy-breezy video editor, which is also in a similarly stagnant state. The only updates iMovie has received in the past year were onboarding screens for permissions settings.
Why is it that Apple can make what is widely regarded as the best video recording experience on any smartphone, but it can’t make a good video editor for a smartphone? Is it partly because these apps don’t have direct payments, so they can only ever be demos for hardware and services that do earn money?
Rosensteel is concerned about the radio silence from Apple regarding Pixelmator and Photomator, the apps (and team) that Apple acquired a year ago:
Of course, Apple may be assembling its own mirror of the Adobe Creative Cloud suite so that it can charge one bundle price for access to a suite of pro apps, and maybe that’s why pricing for everything is frozen in place, and the iPad Pro apps aren’t in step with the Mac ones.
That’s what I hope: that Apple is somewhere near the cusp of announcing some sort of “pro apps” subscription.
Dian Zhang and Ignacio Calderon, reporting for USA Today:
Even before Terry Rozier dropped out of the 2023 NBA game in which he’s accused of rigging his statistics, computers at an “integrity monitor” firm flagged a flood of bets that did not match a mathematical model of how this game should go. The company, now called IC360, alerted the NBA and sportsbooks about the unusual bets coming in on Rozier’s performance.
The investigation that led to the arrest of the Miami Heat point guard and dozens of others for illegal gambling started with math. It ended Oct. 23 with Rozier charged with manipulating his performance in that 2023 game so that gamblers in the know could win tens of thousands of dollars.
Beep. Boop. Busted.
Federal authorities allege more than $200,000 poured in betting that Rozier would turn in a below-average performance in that game after Rozier told another defendant he would drop out of the game early with an injury. Rozier played 9 minutes, 34 seconds for the Charlotte Hornets in the game against the New Orleans Pelicans before leaving with an injury and finished under his usual totals for points, assists and 3-pointers.
A lot of these stories about cheating on sports betting involve characters who aren’t exactly the sharpest tools in the shed. Makes me wonder how many inside-info cheaters are getting away with it, because they’re not doing anything conspicuous like placing very large wagers on very obscure games or prop bets.
Adobe Design profiles Adobe’s new Premiere app for iOS, and interviews Christopher Azar, group design manager for Adobe Video, regarding the thinking behind the app and its design:
What was the primary goal when you set out to design Premiere on iOS?
Christopher Azar: Our goal was to design a professional-grade product that carried the powerful, precise spirit of Premiere while feeling modern, approachable, and even fun. We call our vision “intuitive precision”: a high-performance, intelligent tool powered by cutting-edge AI that enables creators to work how and where they want — in the field, experimenting, and honing their storytelling craft.
That meant making this editing power available to a broader creative community. Desktop software has traditionally been built for professionals with large budgets. Our goal was not only to make a professional tool easier to use, but to make it available to more people than ever before. I would have wanted to use this app when I was coming up as a creative, so I’m excited we’re providing high-quality software for everyone who wants it — without a big investment in time or money.
It really does seem like a breakthrough app for the platform. An Android version is in the works, Adobe says, but for now, Premiere is an iOS exclusive. Kind of weird that Apple itself makes Final Cut Pro for both the Mac and iPad, but still hasn’t made a serious video editing app for the iPhone.
I took the above photo on Monday, June 9, this year at WWDC. Keynote day, around 1:30pm PT. I captured it using my iPhone 16 Pro and Not Boring’s !Camera app, using the built-in Mono Tokyo LUT. Like the other apps in Not Boring’s growing suite, !Camera can be mistaken by the too-cynical as a toy. It is fun and colorful, and some of its features exist for the sake of fun alone. But, just like Not Boring’s other offerings (my favorites: !Weather, !Calculator, and !Habits), it’s a genuinely serious tool. And of the bunch, I think !Camera is the most innovative. The fact that it’s fun makes me want to use it — a vastly underestimated attribute of tool design. From Not Boring’s website:
Go from snap to sharing without any editing. !Camera is the first camera app to enable professional-level color grading with 3D LUTs (“lookup tables”) used in high-end workflows by pro photographers to achieve realistic film simulations and unique cinematic looks. Use !Camera’s designed presets, add LUTs from your favorite creators, or make and import your own! New Styles and collaborations released every season.
!Camera looks gimmicky but I assure you it’s not — and what might strike you as gimmicky is really just plain fun and whimsical. My affection for it, and my use of it, has grown, not shrunk, as the months have gone by. While my hardware Camera Control buttons (plural, as I’m currently testing multiple iPhones) remain set to open Apple’s own Camera app, which I continue to use by default, I keep !Camera’s simple widget on my iPhones’ Lock Screens to launch it quickly after unpocketing my iPhone.
!Camera’s use of LUTs for filter-like effects opens the app to a wide world of non-proprietary looks. The best source I’ve found for new LUTs to import is the Panasonic LUMIX Lab app — Panasonic’s built-in LUTs are boring, but the app has a whole community of user-submitted LUTs and I’ve found several of them that are lovely. !Camera’s custom “SuperRAW” format, is, in my opinion, key to the appeal of the app:
No more flat lifeless photos, no AI processing, no weird artifacts. Our SuperRaw™ photo processing has been crafted to showcase more film-like tones and preserve a photo’s beautiful natural grain.
Rather than fighting the nature of the small (and thus, noisy) sensors in the iPhone camera systems, SuperRAW processing embraces the noise, imbuing images with natural-looking grain. The results, to my eyes, are genuinely film-like. If you want, you can configure !Camera to save a raw DNG file alongside each capture, for post-processing in an app like Darkroom, Lightroom, or Photoshop. I’m glad that option is there, but I just shoot in SuperRAW, which saves ready-to-share HEIC files with the LUT applied in my camera roll, so what I see is what I get.
Each of Not Boring’s apps is available for a $15/year subscription, but the way to go is Not Boring’s $50/year “Super !Boring” subscription, which grants you a license to their entire suite of apps. I was already a Super !Boring subscriber when !Camera launched, so, effectively, I got it for free. $50/year isn’t nothing, but it’s not much, and subscriptions have proven to be the best monetization strategy for indie developers in today’s world.
Marc Levoy, Adobe fellow, and Florian Kainz, principal scientist, on the Adobe Research blog back in June:
Second, people often complain about the “smartphone look” — overly bright, low contrast, high color saturation, strong smoothing, and strong sharpening. To some extent this look is driven by consumer preference. It also makes photos easier to read on the small screen and in bad lighting. But to the discerning photographer, or anybody who views these photos on a larger screen than a phone, they may look unrealistic. [...]
What’s different about computational photography using Indigo? First, we under-expose more strongly than most cameras. Second, we capture, align, and combine more frames when producing each photo — up to 32 frames as in the example above. This means that our photos have fewer blown-out highlights and less noise in the shadows. Taking a photo with our app may require slightly more patience after pressing the shutter button than you’re used to, but after a few seconds you’ll be rewarded with a better picture.
As a side benefit of these two strategies, we need less spatial denoising (i.e. smoothing) than most camera apps. This means we preserve more natural textures. In fact, we bias our processing towards minimal smoothing, even if this means leaving a bit of noise in the photo. You can see these effects in the example photos later in this article.
One more thing. Many of our users prefer to shoot raw, not JPEGs, and they want these raw images to benefit from computational photography. (Some big cameras offer the ability to capture bursts of images and combine them in-camera, but they output a JPEG, not a raw file.) Indigo can output JPEG or raw files that benefit equally from the computational photography strategy outlined here. [...]
In reaction to the prevailing smartphone look, some camera apps advertise “zero-process” photography. In fact, the pixels read from a digital sensor must be processed to create a recognizable image. This processing includes at a minimum white balancing, color correction to account for the different light sensitivity of the red, green and blue pixels, and demosaicing to create a full-color image. Based on our conversations with photographers, what they really want is not zero-process but a more natural look — more like what an SLR might produce. To accomplish this, our photos employ only mild tone mapping, boosting of color saturation, and sharpening. We do perform semantically-aware mask-based adjustments, but only subtle ones.
You may recognize Levoy’s name. After a distinguished career at Stanford teaching computer science, Levoy spent 2014 to 2020 leading the computational photography team at Google for their highly-regarded-as-cameras Pixel phones. In 2020 Levoy left Google for Adobe, and Indigo is one of the first fruits of his time there.
Allison Johnson of The Verge — notably, she came to The Verge by way of DPReview — wrote a splendid piece on Indigo shortly after the app debuted, under the headline “Adobe’s New Camera App Is Making Me Rethink Phone Photography”:
If you hate the overly aggressive HDR look, or you’re tired of your iPhone sharpening the ever-living crap out of your photos, Project Indigo might be for you. It’s available in beta on iOS, though it is not — and I stress this — for the faint of heart. It’s slow, it’s prone to heating up my iPhone, and it drains the battery. But it’s the most thoughtfully designed camera experience I’ve ever used on a phone, and it gave me a renewed sense of curiosity about the camera I use every day.
You’ll know this isn’t your garden-variety camera app right from the onboarding screens. One section details the difference between two histograms available to use with the live preview image (one is based on Indigo’s own processing and one is based on Apple’s image pipeline). Another line describes the way the app handles processing of subjects and skies as “special (but gentle).” This is a camera nerd’s love language.
Slow and battery-draining is exactly why Apple hasn’t pursued these sorts of advanced computational photography techniques in the built-in Camera app. Apple’s Camera app is super-fast and takes extraordinary effort to go easy on the battery. Apple is making entirely different trade-offs — correctly — for the default Camera app. Pro and prosumer photographers may want to make completely different trade-offs when it comes to image processing time and energy.1 (For the last few years, Apple has shot its keynote events using iPhone cameras exclusively, but they use apps like Blackmagic Camera, not the built-in Camera app, to shoot them.)
I’m deeply intrigued by Indigo, and I have a few friends who’ve shown me some extraordinary photographs taken with the app. If they hadn’t told me, I’d have wagered their photos were taken with dedicated large-sensor digital cameras, not phones. Johnson described Indigo as “not for the faint of heart”, and I’m just faint-hearted — or perhaps lazy — enough that, when venturing to a third-party camera app during the past few months, I’ve reached for !Camera, not Indigo, mainly because I don’t want to bother with any sort of manual post-processing for any but my very favorite of favorite images. But Indigo — available free of charge from the App Store — is well worth your attention.2 I hope it’s an app that Adobe is serious about maintaining and developing into the future. ★
Johnson also interviewed Levoy last month on The Vergecast. The interview starts at 30m:22s. ↩︎
Indigo is currently iOS-only, but in their introductory blog post, Levoy and Kainz write: “What’s next for Project Indigo? An Android version for sure. We’d also like to add alternative ‘looks’, maybe even personalized ones. We also plan to add a portrait mode, but with more control and higher image quality than existing camera apps, as well as panorama and video recording, including some cool computational video features we’re cooking up in the lab.” Also worth noting: Indigo’s computational photography is so tied to specific hardware that it doesn’t yet support any of the iPhones 17 nor the iPhone Air. ↩︎︎