452: ‘A Scam Held Together With Patriotism and Golden Paint’, With Quinn Nelson

Quinn, I’m half‑tempted to skip Cupertino this fall because the “new phone” circus already peaked with that Trump T1, and your thirty‑minute teardown somehow left me wanting more.

The punchline: it’s basically an HTC U24 Pro in a yellow suit, not gold, sold for about a fifty‑dollar premium over a phone that wasn’t a deal to begin with.

The lead‑up was chaos—AI-ish renders, a Photoshopped iPhone, even a Galaxy with a case left on—then a real device that’s serviceable but silly as a value play.

Behind the curtain, it looks like Trump Mobile piggybacks Liberty Mobile, an MVNO riding T‑Mobile, with family names on stage and a handful of people doing the actual work.

A breached preorder list showed about thirty thousand depositors, but conversions appear tiny, and any custom backplate and anodizing likely meant minimums they’ll never clear.

Worse, the service is uncompetitive: bring your own phone, or buy a refurb or the T1, then pay around forty seven dollars a month for middling priority when better MVNO deals are everywhere.

This follows the familiar “this can’t be that hard” playbook: slap a brand on a hot market, hope press and loyalty do the rest, and make the money on the plan, not the handset.

Crypto proved that formula can pay obscene sums, at least for the issuer; phones won’t, so this will likely limp along as a Liberty Mobile side door until it quietly fades.

To their credit, choosing HTC probably dodged spyware risks; packet sniffing looked normal, which suggests a late fixer chose competence over a sketchy white‑label special.

Only Apple and maybe Samsung profit on hardware anyway—the carrier is where the margins live—so the shiny phone was the prop to launch a network.

Quick palate cleanser: someone bought the blingy Trump watch and a letter fell off the dial, leaving a new brand name rattling under the crystal—pure Canal Street energy.

Then we swung to OpenAI’s “super app,” which buried chat, botched the Mac rollout, and told users to double‑click an app bundle that did not actually install the right thing unless both old apps were already on the machine.

Even after cleanup, it’s still Electron‑heavy and memory‑hungry, while their data‑center binge is already causing RAM shortages—driving up prices and then asking our laptops to burn more of it.

If AI can write software, where’s the native Mac app built by the AI that proves it; the truth is great apps come from people who care, and right now the vibe is ship fast, shrug at quality.

Claude’s desktop is no better—confusing “plugins, connectors, extensions” soup—and when you carry a valuation near a trillion dollars, the “we’re just a startup” excuse doesn’t fly.

That haste is a red flag when people lean on these tools for real decisions; if you cannot rename a Mac app without chaos, maybe slow down before you reinvent work.

On the brighter side, Siri AI in iOS and macOS 27 is a genuine step forward: on‑screen awareness means you can ask it to add concert details in Safari to a shared calendar or pull a door code from a note, and it just does it.

It still stumbles—missing a social security note until nudged, misreading the “Family” calendar name, or happily claiming it added items when it did not—and it once sent my grocery list to Reminders instead of Todoist.

Most of that feels fixable, but the mountain is trust; years of useless Siri habits trained everyone to stop trying, and Apple now has to re‑teach us what’s possible.

Think of it like 3D Touch without visual cues: you only discover the magic by asking, so Apple needs crisp onboarding and real‑world examples when 27 ships, or the best parts will stay hidden.

The new Siri can do so much more, but if people are not shown how to use it, they will never think to try; Apple needs a bold, funny campaign that admits the old Siri was a miss and invites everyone back, the way Domino’s, Carlsberg, and even the Get a Mac ads did.

The dedicated Siri app on the Home Screen helps signal the change, but the UI feels like a bare transcript log; it should spark ideas and guide prompts, not just file past chats.

Bake LLM search into Spotlight and Messages so any text field accepts real questions; Mail search is finally usable in iOS 27, but Messages still does blunt keyword matching.

Let simple strings stay fast, then escalate to a richer semantic search when queries get sentence-like or fail; people already type that way.

Good indexing changes habits; Spark AI shows how quick, grounded answers with links become the default, even with battery and on-device limits to juggle.

The iOS 27 betas have been fine overall, but longer Siri sessions can overheat the iPhone Air; the 17 Pro’s thermals keep up much better.

The Air feels pro in capability but not in cooling without a vapor chamber, which likely tempers how broadly these features light up right now.

Siri is not the flashiest model, yet it is clear, useful, and on hand; the real battle is erasing the muscle memory that old Siri was a joke.

It nailed a croque monsieur description, then a one-word cheese follow-up flipped the reply into French; a charming stumble, plus the occasional needless safety disclaimer.

The new voices sound far more natural, though they sometimes clip the final word or stress oddly; dictation now learns names and respects spelled-out terms, which is a real upgrade.

Trust lives or dies on sources; it fed me wrong opening hours via bad listings, then found the right time only after I pushed it to search the web, and Apple Maps business data is still wobbly.

Blow an errand or a group lunch twice and people stop using it; reliability on everyday facts matters.

On Mac gaming, the Game Porting Toolkit took a big leap this year; games that crawled now run at genuinely playable frame rates.

Developers still avoid Mac because demand is tiny and support is a tax, and Apple often highlights titles that ship on Steam anyway, so Valve gets the cut.

There is a path if Apple built a real game box or an Apple TV Ultra with M‑series power and proper cooling; a handheld is culturally a nonstarter, and Apple’s aversion to fans borders on religious.

The Apple TV price hike leaves it stuck between costly streamer and underpowered console; if you are going that high, make it a computer.

Winning gamers would require day‑one Steam libraries, cross‑save, and a modern store; Apple’s thirty percent mindset and services comfort make that a long shot.

Gamers flock to what is fast, affordable, and low‑friction, and Apple silicon plus translation could own that lane if the organization truly committed.

Thanks to Quinn for the great back‑and‑forth and to you for listening; links to his videos are in the notes, and we will do more of this soon.