Episode Summary
Executive Summary: The episode is a deep first-look discussion of Apple's 2025 OS betas, centered on the new Liquid Glass design language and its real-world usability problems. The hosts compare beta installs and day-to-day testing across Mac, iPhone, iPad, and Vision Pro, highlighting both promising developer tools (on-device transcription, LLM APIs, Swift 6.2, improved personas) and major interface regressions: contrast, legibility, hidden controls, and excessive transparency.
Main Topics: Liquid Glass design language: aesthetics vs usability: The hosts debate Apple’s new glass-heavy UI, arguing it looks fresh and differentiates Apple platforms but often harms legibility, distracts from content, and creates flashing/inversion issues on dynamic backgrounds. Beta testing across Apple platforms: They compare what they’ve installed and used: macOS Tahoe on Macs/VMs, iOS beta on a primary iPhone, iPadOS beta, and Vision Pro. Experiences range from unstable phone behavior to strong enthusiasm for the iPad update. macOS Tahoe UI changes and regressions: A long section covers Finder, menu bar, title bars, icons, tabs, cursor changes, rounded corners, dialog alignment, and toolbar behavior. The hosts repeatedly criticize hidden controls and floating UI that sacrifices function for appearance. Developer-facing gains: transcription, on-device LLMs, Swift 6.2: Marco and John are excited about Apple’s new transcription API, on-device language model capabilities, and Swift 6.2 improvements to strict concurrency, all of which reduce cloud cost and simplify development. iPadOS 26 as a major step toward desktop-like workflows: Casey praises iPadOS 26, especially windowing and Stage Manager, while noting limits like the disappearance of Slide Over. The group sees it as a meaningful maturation of iPad multitasking. Vision Pro and personas: The improved Vision Pro personas are described as much more realistic and less uncanny, representing one of the standout successes of the new OS cycle. Platform strategy, compatibility, and future constraints: The episode also covers version-number compatibility, Rosetta’s future retirement path, ASIF disk images, and Apple’s platform differentiation strategy through visual redesign and hardware/software integration.
Key Arguments: Apple’s new visual language is meant to look cool and signal that the platform has changed, but many of its implementation choices reduce readability and bury frequently used controls. Liquid Glass can work in constrained contexts, but it fails badly when arbitrary content scrolls behind translucent controls, causing flashing text colors and distracting transitions. Apple has historically solved translucency problems with heavier frosted materials; the new design is much clearer and therefore much more fragile on real-world backgrounds. Many of Apple’s recent UI decisions on iOS and macOS hide common actions behind ellipsis menus without providing enough benefit or configurability. The new iPadOS windowing system is a substantial improvement and makes the iPad feel more like a desktop-class device, though Slide Over’s removal is a loss for some workflows. On-device transcription and LLM APIs are potentially transformative because they are fast, private, and free at point of use, enabling new app features without cloud costs. Swift 6.2 and Apple’s improving concurrency annotations remove some of the friction that previously made adopting strict concurrency painful. Rosetta’s planned narrowing after macOS 27 is seen as unnecessarily aggressive, especially for older Intel-only apps and tools that still run well today.
Data Points: macOS beta install method: 2 Macs with Tahoe plus one VM - John describes booting multiple Macs into Tahoe and experimenting with a VM approach iPhone reboots: 4 reboots in about a week and a half - Marco reports modem and AirPods connectivity failures on his main iPhone beta Users on beta: ~2% of user base - Marco estimates early Tahoe adoption among his users Clipboard history retention: 8 hours - Apple’s built-in clipboard history is limited to an eight-hour cycle Window limit on iPadOS 26: 12 windows - Front-loaded report from Steve Troughton-Smith on an M4 iPad macOS 10.0 memory reference: 128 MB RAM - John cites the original Mac OS X windowing environment as precedent for not limiting windows Fallback version numbers: 19.0 vs 26.0 - Apps built before the iOS 26 SDK report 19.0; apps built with the 26 SDK report 26.0 macOS compatibility version: 16 vs 26 - Tahoe returns 16 for apps built with macOS 15 SDK or earlier, and 26 for the 26 SDK Finder icon size: 64×64 - The new default app icon size mentioned in the squirkle discussion Squirkle radius: 16.5 - Reported radius for the new default app icon shape Toolbar/title bar radii: 16 pt and 26 pt - John notes multiple corner radii depending on toolbar configuration Vision Pro persona improvement: 1 major redesign in VisionOS 26 - The hosts praise the new persona generation as significantly better than prior versions Rosetta support horizon: macOS 27 cutoff; security updates until fall 2028 - Ars Technica’s report summarized in the episode
Pivotal Quotes: "Apple likes its design because it looks cool. That’s it. Just be honest." — Marco: Used to argue that the new Liquid Glass direction is primarily about aesthetics and differentiation, not content visibility "I freaking love it." — Casey: His reaction to iPadOS 26 and its new windowing/desktop-like behavior "If that ships, I will switch to Chrome. It’s that bad." — Marco: His strongest criticism of Safari’s dynamic glass toolbar reacting to webpage colors
Implications: Apple’s 2025 betas suggest a major visual reset with real developer upside, but also serious usability risk. App makers will likely need redesigns, while Apple may need to dial back transparency and hidden controls before release.