iOS 26 Liquid Glass Pixel Audit

Grant: Toot, toot, toot, Maya: toot. Speaker 3: Hey, welcome back to UX Autopsy. I'm Grant. Speaker 4: And I'm Maya. We are doing a full autopsy today. And honestly, Grant, this one's been a long time coming. Speaker 3: Yeah, iOS 18, Liquid Glass, the redesign that basically broke the internet. Speaker 4: And apparently also broke adoption rates. NotebookCheck reported barely 15% of iPhone users have actually upgraded four months post-release, when previous versions were clearing... Hearing 50% by now. Speaker 3: That's wild, right? Like that number alone tells you something went wrong. Or did it? Because here's the thing. The data has some real problems we need to talk through first. Speaker 4: Right, right. StatCounter methodology is messy. We get into it. Speaker 3: So we unpack that, then we dig into what Liquid Glass actually is as a design system, Speaker 4: Mm Speaker 3: where it came Speaker 4: -hmm. Speaker 3: from, and NNg's specific failure modes, screen by screen. Speaker 4: Mail inbox, Settings toggles, Control Center over a photo lock screen. Grant plays attacker, I'll try to defend and we'll see who's left standing. Speaker 3: She says try. Speaker 4: I make no promises. We also run a quick Red Flag or Green Flag on three specific Liquid Glass choices. And then we close on the thing nobody's really talking about. Speaker 3: Alan Dye, Apple's UI design chief since 2015, left for Meta in December, the guy who personally unveiled Liquid Glass at WWDC. See. Speaker 4: Right, after shipping it. And we end the whole thing with one rule you can actually use in your own work tomorrow. Speaker 3: So let's get into it. Segment one, the numbers, the data drama, and what Apple actually shipped. All right, let's start with a number. Four months after launch, roughly 15% of iPhone users had updated to iOS 18, according to NotebookCheck and StatCounter. And before I say anything else, I learned a long time ago that first numbers are often wrong. Speaker 4: 15%? That's not great. Speaker 3: Compare that to iOS 18 at the same point the year before. MacRumors reported 63%. iOS 17 was 54%. iOS 16... 16 with 62 percent. iOS 18 running at less than a quarter of the pace. That's not noise. That's a pattern. Speaker 4: Okay, that's actually wild. Speaker 3: Right? And MacRumors looked at their own readers, which skew heavily towards early adopters. First week of January 2025, 89.3 percent were on iOS 18. Same window a year later, only 25.7 percent on iOS 18. Even the people who would upgrade didn't. Speaker 4: So even the people who would upgrade didn't? Speaker 3: Exactly. That's the gut punch. Speaker 4: Okay, but Grant, I gotta push back here. Those initial StatCounter numbers were broken. Like fundamentally broken. Speaker 3: Fair. Speaker 4: Syphon's anti-fingerprinting was misidentifying iOS 26 as iOS 18. StatCounter fixed it January 19th and boom. Boom, the numbers jumped to over 50%. Speaker 3: Right. And TelemetryDeck put it around 55%. So the 15% was fundamentally broken data. Speaker 4: Mostly noise. Speaker 3: Yeah, okay. But here's what I do with cars when the numbers get messy. I dig deeper anyway. Even after the corrections, TelemetryDeck showed 55% for iOS 26 versus 78% the prior year. That gap's real. Speaker 4: Yeah, I hear you—there's definitely a gap, but worst upgrade cycle in Apple history is doing a lot of heavy lifting when your data had such an obvious flaw. Speaker 3: Totally fair. And when I get access to first-party data, which is rare and valuable, I trust it. Apple's own App Store numbers from February showed 74% on iOS 17 versus 76% for iOS 18, so by the source that actually knows- It's basically flat. Speaker 4: Wait, really? So by Apple's numbers, it's almost a non-story. Speaker 3: By Apple's numbers, yeah. But automatic updates were slow rolled. So you've got a data problem and a distribution problem. That's exactly the kind of tangle where something real is hiding underneath. Speaker 4: Which means the headlines were way cleaner than the actual story. Speaker 3: No, but something still happened, Maya. The social media response, the NN group critique. The pieces about Apple losing quality? These things don't generate smoke without fire. There's real hesitation in the signal. Speaker 4: The score might be disputed, but you're saying the game was still ugly. Speaker 3: Yeah, exactly. So the real question isn't whether adoption crashed. The question is, what did Apple actually ship that made people hesitant in the first place? Alright, so with that hanging in the air, let's actually open this up. What is Liquid Glass? Speaker 4: Right, so it's not just cosmetic. It's a complete material system rebuild. Translucency, blur, refraction, real-time depth rendering. Apple basically repainted the entire infrastructure. And I'm skeptical, because every time someone rebuilds the infrastructure for beauty, I ask what's actually broken underneath. Speaker 3: And by chrome, I mean the navigation layer. Tab bars, navigation bars, Control Center, the lock screen clock, all of it. Speaker 5: Mm-hmm. Speaker 3: Even app icons got the treatment. Speaker 4: Every surface that used to be solid and opaque is now Glass floating over your content, Speaker 3: Which, on paper, sounds kind of beautiful. Alan Dye at WWDC framed it as the foundation for the next chapter of our software. Big language. Speaker 4: Skeptically big language. Press release called it their broadest software design update ever across the whole stack, which is the kind of framing I've learned to read carefully because when everything is revolutionary, usually something is being oversold. Speaker 3: And here's what's actually interesting. The design philosophy came straight out of visionOS and the Vision Pro where translucent floating layers actually make sense because there's real depth in that space. Speaker 4: Okay, that's the crucial thing though, right? In a headset, you have literal depth. The Glass sits in actual three-dimensional space. On a flat phone screen, you're faking something that doesn't exist. That's not subtle. That's the entire premise. Speaker 3: Which Apple says is fine because the hardware can render it convincingly (Silicon advances, graphics advances); the pitch is that the phone is powerful enough now to fake the physics. Speaker 4: Fake the physics. I spent 15 years in markets reading how people can... will confuse elegant theory with working reality. This is that exact move. Grant: Little of both, maybe. But look, the theory IS sound: controls recede visually, content comes forward, Glass belongs on the navigation layer, not the content itself. Speaker 3: In theory, tab bars float above your content, they collapse on scroll, they pick up color underneath. It's designed to feel alive, which is marketing language, and I'm trained to hear that as a warning bell. Grant: And the WWDC presentation leaned hard on words like "joy" and "delight"; which, you know, those are real design values. I don't want to be cynical about them. No, but look at what's missing from that entire framing: zero functional rationale, nothing about legibility, predictability, learnability, accessibility—just aesthetic language. And in my experience, that gap between "looks beautiful" and "actually works" is where the real problems hide. Speaker 3: I mean— To be fair, that's WWDC. It's a keynote, not a usability report. Grant: Sure, but you ship that system to a billion phones and adoption stalls and suddenly the keynote pitch matters. You sold joy, people opened their Mail app and found chaos. That disconnect is the story. Speaker 3: And that's exactly where it gets uncomfortable, right? Because on a photo background, on a noisy wallpaper, Glass controls over that kind of visual chaos? That's not joy." NN group put it bluntly: the visual language is working against the content instead of getting out of its way. Grant: Which is the opposite of what Apple explicitly committed to at that same presentation. That's not a detail, that's a failure of execution against stated intent. Speaker 3: Right, so the gap between what they meant to do and what actually shipped, that's where the autopsy really starts. And that's where we're headed next because NNg's Raluca Budiu went through screen by screen and documented exactly what fell apart. Grant: Specific failures on specific services. I've seen deals blow up over smaller gaps than that. This is going to sting. So we've laid out what Liquid Glass is, now let's put it on the autopsy table. Speaker 3: Yeah, scalpel time. Grant: Raluca Budiu at NNg documented specific failure modes, and they are not subtle. The Mail inbox, for instance, has a semi-transparent header that literally renders on top of message text behind it, text on top of text. That's not a contrast issue, that's a legibility collapse. Speaker 3: Okay, so I want to push back a little. Some of that is... That is beta-era behavior. Apple shipped point releases pretty fast after launch. Grant: Sure, but the toggle animation in Settings, that one is not a bug. They redesigned it. Toggle a switch and it jumps, slows down, bounces. What was a half second ephemeral tap is now this little performance. Speaker 3: I mean, it's animated feedback. There's an argument for that. Grant: There's an argument, sure, but NNg's verdict was that the whole interface feels, quote, restless, needy, less predictable, less legible. That is not about one animation. That's a system-level indictment. Speaker 3: Right, right, and that came from Budiu specifically, not just a blogger. Grant: Exactly. And Control Center over a photo-heavy lock screen? The Glass surface is refracting whatever image is behind it in real time. Beautiful in a demo, genuinely hard to locate buttons in daily use. Speaker 3: Okay, I'll grant you that one. Location of controls matters. Grant: And then there's the battery story, which surprised me. A YouTube channel called In Depth Tech Reviews ran a side by side, same iPhone 16 Pro Max, same 150 actions, notifications, app switching, photo scrolling, unlocking, iOS 18. burned thirteen percent battery. iOS 18 burned one percent. Speaker 3: Wait, really? Thirteen versus one? Grant: Thirteen versus one. And the device ran hotter throughout. The GPU is computing real-time refraction and blur on every surface transition, every swipe. It's basically rendering a 3D effect on a 2D screen constantly. Speaker 3: Okay, but here's where I'll defend Apple. That test wasn't- an early release, and if you look at 26.1, they added a proper transparency dial, not just the old reduced transparency accessibility toggle, but an actual design setting, so users can dial the glass back without losing the whole visual language. Grant: That's a real concession, I'll take it. But it also confirms they knew it was a problem. Speaker 3: Or they heard feedback and responded. That's not nothing. Grant: Fair point, though the battery hit lands hardest on phones two to four years old. Gold, which is most of the installed base, a twenty six point one toggle helps current hardware; it doesn't fix the experience on an iPhone 12. Speaker 3: Hmm. That I don't have a great answer for. Grant: So we're leaving this one unresolved? Speaker 3: Feels right. Some of it is fixable, some of it is baked in. Grant: All right, let's shift gears and go faster. We've got a red flag or green flag round, and I want to hear where you actually land on a few of these. Okay, shifting gears. Red flag or green flag time? Fast grounds, no speeches. Speaker 3: Let's go, hit me. Grant: Floating translucent tab bar over a full bleed photo feed, red or green? Speaker 3: Red, full stop. You're putting a see-through element on top of literal photos. The content you're trying to see is being eaten by the navigation meant to help you see it. Grant: Same page, red for me too. NNg flagged exactly this pattern. Pattern, controls competing with content instead of framing it. That's not a trade-off. That's a fumble. Speaker 3: Agreed. Okay, next one. Grant: System-wide refraction and blur. One unified visual language spanning iPhone, iPad, Mac, Watch and Vision Pro for the first time ever. Red or green? Speaker 3: Hmm. Okay, I'm going cautious green? Like, the ambition is real. Getting five platforms to speak the same visual grammar is genuinely hard. Grant: I want to say green, but I can't quite get there. Speaker 4: Hmm. Grant: One language that works beautifully on Vision Pro and 3D depth might just not translate to a flat watch face on a 45 millimeter display Speaker 4: Right, right. So the concept is green, but the execution on the smallest screens is red. Grant: That's where I land split decision this Speaker 4: Okay, Grant: Speaker 4: contested. Moving on. Last one. Grant: one's going to get us iOS 26.1 adds a dedicated Liquid Glass toggle you go to Settings Display and Brightness Pick clear or tinted. Tinted bumps opacity, improves contrast, keeps the design language intact, red or green. Speaker 4: Wait, so it's not just the old accessibility reduced transparency, it's a proper first-class setting now? Grant: Yep. According to MacRumors, the 26.1 toggle sits in Display and Brightness, not buried in Accessibility. You keep the glass, just dial. Dial back the Speaker 3: Wow. Grant: C3. The nuclear option was always in Accessibility. This is the civilian version. Speaker 3: Okay, Green. If you've listened to feedback and given people a middle path without nuking the whole design, that's responsive. That's a team paying attention. Grant: I'm half green. The thing that bugs me is that it doesn't touch Control Center. MacRumors can find. The toggle has, quote, little to no change on Control Center, which is arguably the messiest surface in the whole OS. Speaker 3: Ah, of course it doesn't. Grant: So it's a green for effort and a yellow for coverage. A toggle that misses the worst offender is still a partial fix. Speaker 4: I'll take partial. Better than nothing. Grant: Fair. And honestly, the fact that they shifted it all raises a bigger question about who's steering this thing going forward. Speaker 4: Curious. What do you mean? Grant: That's the part of the story we haven't touched yet. The people behind the design, where they went, and what Apple's actually building toward. Speaker 3: Oh, yeah, that's a whole different layer. Grant: And it reframes everything we've been picking apart. So who was actually steering Liquid Glass? Because that answer got messy fast. Speaker 4: Alan Dye, head of human interface at Apple for a decade, he personally unveiled it at WWDC, and then December 2025, he's gone. Grant: Gone to Meta of all places. Speaker 4: According to TechCrunch, Dye left to lead a new design studio inside Reality Labs. His deputy, Billy Sorrentino, followed him over. And Zuckerberg framed it as treating, quote, intelligence as a new design material. Grant: Which, I mean, that phrase is doing a lot of work. Speaker 4: It really is. But here's what's interesting. His replacement at Apple is Stephen Lemay, a 26-year Apple veteran, and internal sources per 9to5Mac reportedly describe the reaction inside the company as relief. Grant: Relief. That's not a small word. Speaker 4: No, it's not. Grant: Okay, so here's where I get stuck playing skeptic. If Liquid Glass was really this grand spatial computing aerosol, why did the architect leave right after shipping it? Speaker 3: I don't think that's actually the defender you think it is. The visionOS design lineage goes back years before Dye left. The floating layers, the depth cues, the material responsive controls. That grammar is already in Vision Pro. iOS 26 was always training users on that grammar on a flat screen first. Speaker 5: So it's a long game. Speaker 3: That's the steel man, yeah. People who get fluent with Glass layers on their iPhone won't be- Speaker 6: Won't be confused when they strap on AR glasses; the 2D version is the tutorial. Speaker 5: I don't fully hate that argument, but Vista shipped with a thing called Aero Glass, remember that? Translucent chrome everywhere. Microsoft also called it a preview of future interfaces. Speaker 6: And then they killed it in Windows 8. Speaker 5: Exactly. The long game defense only works if you actually ship the future you promised. Apple has Vision Pro. Microsoft had... not much. Speaker 6: Fair. And that's probably why the Dye departure matters as a signal, not a cause. If the people who built the vision are now at Meta building the actual next-gen hardware layer, Speaker 5: Yeah. Speaker 6: that's worth watching. Speaker 5: Okay, so what do listeners walk away with? Because we've torn this apart pretty thoroughly. Speaker 6: Here's the thing I keep coming back to. NN/g flagged it, and it holds for anyone building layered or translucent interfaces: build the accessibility- Speaker 3: The off state first. Speaker 5: Say more. Speaker 3: Reduced transparency mode is not a fallback. If your interface only looks good with effects on, the design isn't finished. The no effects version is your proof of concept. Start there. If it works without the Glass, the Glass is a bonus. If it only works with the Glass, you have a problem. Speaker 5: And iOS 18 by that standard. Speaker 3: Has a problem in some places, yeah. Speaker 5: That's a clean rule. Build the stripped version first. Steal that. Okay, so that was a lot that you won. Speaker 6: Right? Like, we started with a number that looked alarming, found out it was messier than it first appeared, and still ended up with a genuine story about user resistance. Speaker 5: Yeah, and I think the thing I keep coming back to is the WWDC pitch, you know? Almost zero functional rationale, just vibes, and users noticed. Speaker 6: That's the Autopsy result. Speaker 3: You can ship beautiful and still ship confusing. Speaker 5: The takeaway for anyone building anything, design the accessibility off-state first. If that version still holds up, you're probably on solid ground. Speaker 3: Steal that. Seriously. Speaker 5: All right, if this episode made you side-eye your own UI a little, we did our job. Screenshot something that bugs you and tag us. Speaker 3: Subscribe so you don't miss next week's audit. And hey, if you learned something today? Faith, a review goes a long way. Grant: Thanks for being here, Maya. Speaker 3: Always, Grant. See you next time, everyone.