Thank you for reading this post, don't forget to subscribe!
If WWDC’s opening video — which solid Apple executives as characters within the upcoming F1 film, with Senior Vice President of Software program Engineering Craig Federighi within the starring function — was a little bit of a fever dream, then the opening of Federighi’s presentation of Apple’s annual software program updates had the air of a regretful admission the morning after that errors had been made.
To Federighi and Apple’s credit score, there was no try to bop round the truth that final yr’s WWDC ended up being a fever dream in its personal proper: Apple promised a litany of AI-enabled options, significantly for Siri, that haven’t and should not ever ship. Federighi, after itemizing the fundamental and hardly ground-breaking Apple Intelligence options that did ship, admitted proper up entrance:
As we’ve shared, we’re persevering with our work to ship the options that make Siri much more private. This work wanted extra time to achieve our top quality bar, and we stay up for sharing extra about it within the coming yr.
That’s solely two sentences, in fact, however the admission was notable and mandatory; final yr’s WWDC — which garnered excessive reward, together with from yours really — revealed that One thing Is Rotten within the State of Cupertino. That was the title of John Gruber’s Daring Fireball article excoriating Apple for promising one thing it couldn’t ship:
Even with the whole lot Apple overpromised (if not outright lied about) on the WWDC keynote, the preliminary takeaway from WWDC from the information media was wrongly centered on their partnership with OpenAI. The traditional knowledge popping out of the keynote was that Apple had simply introduced one thing referred to as “Apple Intelligence” nevertheless it was powered by ChatGPT, when in actual fact, the story Apple informed was that they — Apple — had constructed a whole system referred to as Apple Intelligence, totally powered by Apple’s personal AI expertise, and that it spanned from on-device execution all the best way to a brand new Non-public Cloud Compute infrastructure they not solely owned however are powering with their very own custom-designed server {hardware} primarily based on Apple Silicon chips. And that on prime of all that, as a proverbial cherry on prime, Apple additionally was including an optionally available integration layer with ChatGPT.
So, sure, provided that the information media gave credit score for Apple’s personal precise introduced achievements to OpenAI, Apple certainly would have been given even much less credit score had they not introduced the “extra customized Siri” options. It’s straightforward to think about somebody within the government ranks arguing “We have to present one thing that solely Apple can do.” But it surely seems they introduced one thing Apple couldn’t do. And now they give the impression of being so out of their depth, so in over their heads, that not solely are they years behind the state-of-the-art in AI, however they don’t even know what they’ll ship or when. Their headline options from 9 months in the past not solely haven’t shipped however nonetheless haven’t even been demonstrated, which I, for one, now presume means they’ll’t be demonstrated as a result of they don’t work.
Gruber — my co-host on Dithering — has been writing and podcasting about Apple for over 20 years; his podcast is known as The Discuss Present, and for the final ten years Apple executives have appeared for a stay model of his present the week of WWDC. Nonetheless, this yr shall be totally different; Gruber introduced on Daring Fireball:1
Ever since I began doing these stay exhibits from WWDC, I’ve stored the visitor(s) secret, till showtime. I’m nonetheless doing that this yr. However lately the friends have appeared a bit predictable: senior executives from Apple. This yr I once more prolonged my regular invitation to Apple, however, for the primary time since 2015, they declined.
I feel this can make for a captivating present, however I need to set everybody’s expectations accordingly. I’m invigorated by this. See you on the present, I hope.
Neither Gruber nor (clearly) Apple stated why the invitation was declined, nevertheless it was laborious to not draw a line to Gruber’s viral article; Marco Arment stated Apple was Retreating to Security:
Perhaps Apple has good causes. Perhaps not. We’ll see what their WWDC PR technique appears to be like like in a few weeks.
Within the absence of some other info, it’s straightforward to imagine that Apple now not desires its executives to be interviewed in a human, unscripted, unedited context which will comprise laborious questions, and that Apple now not feels it mandatory to indicate their appreciation to our group and builders on this means.
I hope that’s both not the case, or it doesn’t keep the case for lengthy.
This would be the first WWDC I’m not attending since 2009 (excluding the distant 2020 one, in fact). Given my realizations about my relationship with Apple and how they view builders, I’ve determined that it’s greatest for me to take a break this yr, achieve some perspective, and resolve what my future relationship ought to appear to be.
Perhaps Apple’s leaders are doing that, too.
My largest takeaway from WWDC is that Arment obtained it proper: Apple is certainly “Retreating to Security”. Retreating would possibly, nonetheless, be precisely the correct factor to do.
Apple’s Retreat to Its Core Competency
The headline function of WWDC this yr was Liquid Glass, a brand new unified design language that stretches throughout its working methods. I’ll reserve judgment on Liquid Glass’s aesthetics and usefulness — Gruber likes it, however I’m not one to put in developer betas on my units — however will make three meta observations.
First, hand-crafted UI overhauls are the polar reverse of the probabilistic world of generative AI. One is about deep consideration and iteration leading to a completed product; the opposite is about in-the-moment token prediction leading to an output that’s ephemeral and disposable. Each are essential and inventive, however the downsides of that creativity — unfamiliarity and edge instances versus hallucination and false confidence — are themselves diametrically opposed.
Apple’s historic strengths have at all times been rooted in designing for finality; in my first yr of Stratechery I did a SWOT evaluation of the massive tech firms and stated about Apple:
Apple’s product improvement course of is fantastic for creating completed merchandise, however that very same course of doesn’t work practically as properly in relation to constructing cloud companies. Cloud companies are by no means “executed”; they’re greatest developed by beginning with a minimal viable product after which iterating primarily based on utilization. That is exactly reverse of what it takes to design a chunk of {hardware}, and it’s a giant cause why Apple struggles a lot with cloud companies (and why different companies firms wrestle with merchandise). The canonical instance of this, in fact, was the MobileMe launch, which was delivered fully-formed and which, when confronted with actual world utilization, crashed-and-burned. Apple’s newest choices are higher, however nonetheless endure from an excessive amount of inside improvement time per launch. It is a laborious downside to repair, as a result of it touches the core of what makes Apple Apple.
I feel it issues whether or not or not Liquid Glass is nice, as a result of it is going to be a testomony concerning the state of Apple’s strengths; the purpose for this Article, nonetheless, is that WWDC was a retreat to these strengths, away from a expertise that may be very a lot inline with Apple’s historic weaknesses.2
Second, the core premise of the Liquid Glass re-design is leveraging Apple’s integration of {hardware} and software program. That is how Vice President of Human Interface Design Alan Dye launched Apple’s new design language:
Now with the highly effective advances in our {hardware}, silicon, and graphics applied sciences, now we have the chance to put the inspiration for the subsequent chapter of our software program. In the present day, we’re excited to announce our broadest design replace ever. Our objective is a ravishing new design that brings pleasure and delight to each consumer expertise, one which’s extra private and places larger focus in your content material. All whereas nonetheless feeling immediately acquainted. And for the primary time, we’re introducing a common design throughout our platforms. This unified design language creates a extra harmonious expertise as he transfer between merchandise, whereas sustaining the qualities that make every distinctive. Impressed by the physicality and richness of visionOS, we challenged ourselves to make make one thing purely digital, really feel pure and alive, from the way it appears to be like to the way it feels because it dynamically responds to the touch.
To realize this, we started by rethinking the elemental parts that make up our software program, and it begins with a wholly new expressive materials we name Liquid Glass. With the optical qualities of glass, and a fluidity that solely Apple can obtain, it transforms relying in your content material, and even your context, and brings extra readability to navigation and controls. It superbly refracts mild, and dynamically reacts to your motion with specular highlights. This materials brings a brand new stage of vitality to each facet of your expertise, from the smallest parts you work together with, to bigger ones, it responds in actual time to your content material, and your enter, making a extra vigorous expertise that we expect you’ll discover really pleasant. Components as soon as thought of for rectangular shows have been redesigned to suit completely concentric with the rounded corners of the {hardware}. This establishes larger concord between our software program and {hardware}, whereas thoughtfully thought of teams of controls, release helpful area on your content material. Liquid Glass is translucent and behaves similar to glass in the actual world. Its colour is knowledgeable by your content material and intelligently adapts between mild and darkish environments, and as a definite purposeful layer that sits above your app, the fabric dynamically morphs once you want extra choices, or as you progress between views.
We’ve at all times cared deeply about each element of our software program design, and it’s these moments of magnificence, craft, and pleasure that convey our merchandise to life. Our new design blurs the strains between {hardware} and software program to create an expertise that’s extra extra pleasant than ever, whereas nonetheless acquainted and simple to make use of. In the present day marks an thrilling and exquisite new chapter for our design, one which units the stage for our subsequent period of our merchandise and the way you work together with them day-after-day.
Sebastiaan de With, in a remarkably prescient submit predicting the character of this new design, emphasised how solely Apple might make Liquid Glass:
A logical subsequent step might be extending physicality to the whole thing of the interface. We would not have to go overboard in such remedies, however we will now have the interface inhabit a way of tactile realism. Philosophically, if I used to be Apple, I’d describe this as lastly having an interface that matches the attractive materials properties of its units. All of the surfaces of your units have glass screens. This brings an interface of an identical materials, giving the consumer a sense of the glass itself coming alive…
The interfaces of computer systems of the longer term are sometimes surprisingly straightforward to think about. We regularly consider them and have them in fiction forward of their existence: our iPhone resembles a contemporary Star Trek tricorder; many trendy AI purposes resemble the units in sci-fi motion pictures like ‘Her’ and (depressingly) Blade Runner 2049. It’s not shocking, then, that idea interfaces from the likes of Microsoft typically function ‘glass fiction’:
The precise interface is sadly not practically as impressed with such life and behavioral qualities. The reason being easy: not solely is the cool residing glass of the video means excessive in some locations, however few firms can really dedicate important effort in the direction of making a hardware-to-software built-in rendering pipeline to allow such UI improvements…Solely Apple might combine sub pixel antialiasing and never-interrupted animations on a {hardware} stage to allow the Dynamic Island and gestural multi-tasking; solely Apple can combine two working methods on two chips on Imaginative and prescient Professional to allow them to composite the dynamic supplies of the VisionOS UI. And, maybe solely Apple can push the cutting-edge to a brand new interface that brings the glass of your display to life.
de With’s prescience really provides me nice hope for Liquid Glass: the most effective improvements are apparent to those that perceive what’s simply changing into doable, and Apple’s integration has and continues to be a significant benefit for issues like consumer interfaces.
Third, Apple CEO Tim Prepare dinner has for a very long time prolonged his framing of Apple’s differentiation to be the combination of {hardware}, software program, and companies, however whereas that’s definitely true from a monetary perspective, I’ve lengthy had a tough time shopping for that the companies part made for higher merchandise; as I famous above, Apple’s companies have sometimes been one thing to be endured, versus a cause to purchase their units, and the Siri debacle has solely emphasised that time.
What’s way more compelling — and the truth that Liquid Glass is a design language meant to unify Apple’s units speaks to this — is the combination of Apple’s units with one another. Each Apple product you purchase is enhanced by the acquisition of further Apple merchandise; to that finish, one of many coolest elements of the WWDC presentation was about Continuity, Apple’s model for options that join varied Apple merchandise:
Let’s speak concerning the energy of continuity. Continuity helps you’re employed seamlessly throughout Apple units, and we’re excited to introduce two new Continuity options. First, we’re bringing Dwell Actions to Mac. So for those who’ve ordered Uber Eats in your iPhone, the Dwell Exercise additionally seems within the menu bar, and once you click on, the app opens in iPhone-mirroring, so you may take motion straight in your Mac. We’re additionally enhancing the calling expertise by bringing the Telephone app to Mac. You may conveniently entry your acquainted content material, like recents, contacts, and voicemail, synced from iPhone, and simply make a name with only a click on. Incoming calls look stunning on the larger display, that includes beautiful contact posters of your family and friends, and the Telephone app on Mac consists of all the good updates we talked about earlier, like maintain help, name screening, and stay translation. In order that’s what’s new in Continuity.
These kinds of options aren’t going to alter the world; they’re, although, options that I can completely see making my life higher and extra handy on an ongoing foundation. And, to the broader level, they’re options that solely Apple can do.
Extra typically, sure, a concentrate on UI design is a retreat from The Gen AI Bridge to the Future; that future, nonetheless, will begin from the units we nonetheless use all day day-after-day, and Apple specializing in making these units higher is a retreat that I count on could have a much more optimistic affect on my life than the corporate struggling to catch up in AI.
Apple’s Retreat to Empowering Builders and Companions
That’s to not say there weren’t some notable AI bulletins in Apple’s keynote.
First, Apple introduced the Basis Fashions framework:
This yr we’re doing one thing new, and we expect it’s going to be fairly huge. We’re opening up entry for any app to faucet straight into the on-device, massive language mannequin on the core of Apple Intelligence, with a brand new Basis Fashions Framework. This provides builders direct entry to intelligence that’s highly effective, quick, constructed with privateness, and out there even once you’re offline. We expect this can ignite an entire new wave of clever experiences within the apps you employ day-after-day.
For instance, for those who’re preparing for an examination, an app like Kahoot can create a customized quiz out of your notes to make finding out extra partaking, and since it makes use of on-device fashions, this occurs with out Cloud API prices. Or maybe you’re tenting off-grid, poring over the hikes you downloaded to AllTrails. Simply describe what you’re within the temper for, and AllTrails can use our on-device fashions to counsel the most suitable choice. We couldn’t be extra enthusiastic about how builders can construct on Apple Intelligence to convey you new experiences which can be good, out there once you’re offline, and that defend your privateness.
It’s essential to not oversell the capabilities of Apple’s on-device AI fashions: in fact builders who need to create one thing that’s aggressive with the output of one thing like ChatGPT might want to use cloud-based AI APIs. That actuality, nonetheless, applies to Apple as properly! A part of the folly of the preliminary Apple Intelligence method is that Apple was promising to ship past state-of-the-art capabilities on a budget, utilizing its customers’ processors and energy.
What’s compelling concerning the Basis Fashions Framework is the way it empowers small builders to experiment with on-device AI at no cost: an app that wouldn’t have AI in any respect for value causes now can, and if that output isn’t aggressive with cloud AI then that’s the developer’s downside, not Apple’s; on the similar time, by enabling builders to experiment Apple is the massive beneficiary of people who uncover the right way to do one thing that’s solely doable when you have an Apple gadget.
Second, Apple deepened its reliance on OpenAI, incorporating ChatGPT’s picture era capabilities into Picture Playground and including ChatGPT evaluation to Visible Intelligence. There’s nonetheless no signal of the long-rumored Gemini integration or the power to modify out ChatGPT for the AI supplier of your selection, however the common pattern in the direction of counting on companions who’re really good at constructing AI is a great transfer.
Third, Apple can be incorporating ChatGPT way more deeply into Xcode, its Built-in Improvement Surroundings (IDE) for constructing apps for Apple platforms; builders can even plugin different fashions utilizing API keys. Xcode nonetheless has an extended methods to go to catch as much as AI-first IDEs like Cursor, however once more, partnering with foundational mannequin makers is a far smarter technique than Apple attempting to do the whole lot itself.
These are, to make sure, apparentstrikes, however that doesn’t make them any much less essential, each when it comes to Apple’s future, and in addition with regard to the theme of this Article: Apple’s preliminary success with the Apple II was due to Third-party builders, and builders have been vital to creating the iPhone a sustainable success. Trusting builders and counting on companions could also be a retreat from Apple’s growing insistence on doing the whole lot itself, however it is vitally a lot a welcome one.
Apple’s [Forced] Retreat to App Retailer Sanity
Apple didn’t say a lot concerning the App Retailer within the keynote, however they did announce a brand new Video games app; M.G. Siegler theorized late final month that this can be laying the groundwork for splitting up Video games from the remainder of the App Retailer:
What if this new gaming-focused app – let’s simply name it ‘Sport Retailer’ – shouldn’t be solely meant to unify Apple’s gaming-focused efforts, but additionally to separate them from the App Retailer itself? Why would possibly Apple do that? As a result of this may permit them to extra simply differentiate between the 2 and, importantly, give the 2 unbiased insurance policies.
That signifies that Apple might, say, drop the speed builders must pay in relation to income share within the App Retailer, whereas conserving it the identical as it’s now within the ‘Sport Retailer’. And that issues as a result of really, gaming makes up some two-thirds of Apple’s App Retailer income in the mean time (between paid downloads and in-app purchases – nevertheless it’s predominantly the latter). It’s the precise key to Apple’s mannequin for this section of the Companies enterprise.
And guess what else is true? In gaming, a 70/30 break up is a well-established norm. The truth is, it’s the place Apple’s personal App Retailer break up originates from (by the use of iTunes, which additionally copied the mannequin again within the day)! Sure, there are others who’ve tried to disrupt this break up, notably Epic, however Apple has a a lot stronger case for a 70/30 break up in relation to gaming than it now does for the general app ecosystem.
So hear me out: the ‘Sport Retailer’ retains the 70/30 mannequin and the ‘App Retailer’ strikes to one thing extra like 85/15 as the usual (matching Apple’s at present convoluted system for small builders with varied arbitrary thresholds). Maybe for smaller builders, Apple even drops it to 90/10.
Apple didn’t announce such a shift yesterday, however the infrastructure is now in place to do precisely what I’ve advocated for years: deal with video games in a different way from different apps. Gaming income is sort of totally primarily based on zero marginal value content material, and video games are extra vulnerable to abuse and extra doubtless for use by youngsters; I don’t thoughts Apple’s extra heavy-handed method in that case and, as Siegler notes, this stage of management is the trade customary for different gaming platforms like consoles. In different phrases, Apple ought to retreat from attempting to take a reduce of the whole lot digital, and as a substitute act like a console maker the place applicable, and a extra impartial pc platform for the whole lot else.
Sadly for Apple, conserving console-level management of video games could now not be doable, significantly after the Ninth Circuit Court docket of Appeals denied Apple’s request for a keep of Choose Yvonne Gonzalez Roger’s order lifting anti-steering restrictions on all apps, together with video games. The purposeful consequence of Gonzalez Rogers’ order is a retreat by Apple from its overbearing management of app monetization, albeit not one Apple is engaged in willingly.
As soon as once more, nonetheless, a retreat is strictly what Apple wants. The corporate has gone too far with the App Retailer, not solely embittering builders and shedding court docket instances, but additionally has put its elementary differentiation in danger. I warned the corporate of precisely this in 2021’s Built-in Apple and App Retailer Danger:
That is the place the nuance I mentioned in App Retailer Arguments turns into way more black-and-white. Sure, Apple created the iPhone and the App Retailer and, beneath present U.S. antitrust doctrine, virtually definitely has the correct to impose no matter taxes it needs on third events, together with 30% on purchases and the primary yr of subscriptions, and fully chopping off builders from their clients. Antitrust regulation, although, whereas ruled by Supreme Court docket precedent, shouldn’t be a matter of constitutionality: it stems from legal guidelines handed by Congress, and it may be modified by new legal guidelines handed by Congress.
One of many central planks of a lot of these pushing for brand new legal guidelines on this space are important limitations on the power of platforms to supply apps and companies, or combine them in any means that benefits their choices. On this potential world it’s not merely problematic that Apple prices Spotify 30%, or else forces the music streaming service to hope that customers determine the right way to subscribe on the net, whilst Apple Music has a totally built-in sign-up movement and no 30% tax; it’s also unlawful to include Apple Music into SharePlay or Shared-with-you or Pictures, or in essentially the most excessive variations of those proposed legal guidelines, even have Apple Music in any respect. This limitation would apply to principally each WWDC announcement: say goodbye to Fast Observe or SharePlay-as-an-exclusive-service, or any variety of Apple’s built-in choices.
I feel these kinds of limitations could be disappointing as a consumer — integration actually does typically result in higher outcomes sooner — and could be a catastrophe for Apple. The whole firm’s differentiation is based on integration, together with its capability to abuse its App Retailer place, and it might be an enormous misstep if the lack to withstand the latter imperiled the previous.
This, greater than something, is why Apple ought to rethink its method to the App Retailer. The deeper the corporate integrates, the extra unfair its arbitrary limits on competing companies shall be. Isn’t it sufficient that Spotify won’t ever be as built-in as Apple Music, or that 1Password won’t be built-in like Keychain, or that SimpleNote will solely ever be in its sandbox whereas Apple Notes is omnipresent? Apple, by advantage of constructing the underlying platform, has each benefit on the planet in relation to providing further apps and companies, and the corporate at its greatest leverages that benefit to create experiences that customers love; on this view demanding 30% and complete management of the customers of its already diminished competitors isn’t merely anticompetitive, it’s risking what makes the corporate distinctive.
That is precisely what is going on in Europe: the DMA requires Apple to open up an entire host of capabilities that undergird its integration each on particular person units and particularly between units, and Apple is signalling that it’ll merely take away these options within the E.U.. That’s one approach to clear up the corporate’s DMA points, however the fee is extreme: in one in every of Apple’s largest markets it will probably’t really ship on the differentiation that, earlier on this Article, I used to be celebrating its retreat to.
Retreat and Reset
The largest a part of WWDC — and the largest a part of this Article — was about Liquid Glass, which has drawn some unflattering comparisons to previous Microsoft working system releases:
The Home windows Vista replace with aero glass was an enormous a part of my childhood, so I am getting critical flashbacks
Once more, I’ll withhold judgment on Liquid Glass till it ships, however there may be one other Microsoft OS comparability I’ve been eager about just lately: to me final yr’s Siri catastrophe is lots like Home windows 8.
Home windows 8 was an try to leverage Microsoft’s massive PC set up base right into a aggressive place in touch-based units, and it didn’t go properly: shoppers — significantly enterprises — hated the brand new UI, and builders weren’t serious about a platform with out customers. Microsoft was pressured to retreat, and finally got here out with Home windows 10, which was way more inline with conventional Home windows releases.
Apple has clearly missed the boat on innovative AI; what I’m open to is the argument that this was a ship the corporate was by no means meant to board, at the least in relation to merchandise like ChatGPT. In the meantime, I’ve lengthy been satisfied that Apple has gone too far in its try to manage the whole lot even tangentially associated to its units; from 2017’s Apple and the Oak Tree:
Apple has had a particular run, due to its particular capability to begin with the consumer expertise and construct from there. It’s why the corporate is dominant and can proceed to be so for a few years. Apple as an entity, although, shouldn’t be particular in relation to the burden of success: there shall be no going again to “Rip. Combine. Burn.” or its trendy equal.
In brief, Apple is now not the little reed they have been when Jobs might fully change the corporate’s technique in mere months; the push in the direction of ever extra lock-in, ever extra centralization, ever extra ongoing monetization of its customers — even at the price of the consumer expertise, if want be — shall be unimaginable to cease, for Tim Prepare dinner or anybody else. In any case, such strikes make Apple sturdy, till in fact they don’t.
To that finish, whereas I perceive why many individuals have been underwhelmed by this WWDC, significantly compared to the AI extravaganza that was Google I/O, I feel it was one of many extra encouraging Apple keynotes in a very long time. Apple is an organization that went too far in too many areas, and wanted to retreat. Specializing in issues solely Apple can do is an efficient factor; empowering builders and relying on companions is an efficient factor; giving even the looks of considerate pondering almost about the App Retailer (it’s a low bar!) is an efficient factor. After all we would like and are excited by tech firms promising the longer term; what’s a prerequisite is delivering within the current, and it’s an indication of progress that Apple retreated to nothing greater than that.
