Overview of The iPhone Duo should have been an iPad
This Vergecast episode centers on the review fallout around Apple’s first foldable iPhone, which Nilay Patel argues should be understood less as an iPhone and more as an iPad-like device. The hosts also dig into a major AI-assisted PHP vulnerability discovery that exposed how quickly offensive and defensive automation is changing web security, before moving to Brendan Carr’s alarming attempt to pressure broadcasters over election polls. In the lightning round, they cover Arc browser’s shutdown, Meta’s new terms of service, Netflix’s content reset, and Google’s confusing Gemini rollout across devices.
Apple’s First Foldable: Fun, Promising, but Still Not a Great iPhone
The central argument
- Nilay’s core point: the device is fun and feels like a promising first-gen product, but it is not yet a good iPhone for most people.
- He ultimately lands on a 7/10, but only because he’s grading the product as it exists now, not the future version Apple is likely aiming for.
Why the experience improved during the review
- Early on, the phone felt “very Android-y” and awkward: lots of app weirdness, black bars, and unoptimized interfaces.
- Over the course of the week, app support improved as developers updated for the new form factor.
- Even so, many apps remain broken or awkward, especially on the cover screen.
App support is still a mess
The hosts describe three app states:
- Unoptimized iPhone apps: tiny, floating, and awkward.
- Partially updated apps: wider, but still with black bars and layout issues.
- Duo-optimized apps: better, but many are still buggy or incomplete.
The bigger philosophical argument
- The hosts repeatedly say the device feels more like an iPad that can also be a phone than a phone that unfolds into a tablet.
- They argue Apple may have been better off marketing it as a new category or as an “iPad Duo,” rather than selling it as an iPhone.
- Price is a major issue: at $2,000, it’s too expensive for something that still feels unfinished.
Bottom line
- The hardware is compelling, and the long-term trajectory looks obvious.
- But as shipped, it’s compromised: hot, expensive, battery-hungry, and not yet polished enough to be a mainstream phone.
The Deeper Idea: This Feels Like a Future iPad, Not a Future iPhone
Why the iPad comparison matters
- Both Nilay and John Gruber effectively land on the same idea: this is less a worse iPhone than a better iPad.
- The inside display is described as excellent for media, multitasking, and general tablet-style use.
- The outside display, by contrast, is too small and awkward to serve as a great phone experience.
The strategic tension
- Apple has spent years avoiding the idea that the iPad and Mac should converge too much.
- This device, and the broader direction of Apple’s hardware, suggests those product lines may be moving toward each other anyway.
- The hosts also note that if Apple eventually ships a touchscreen Mac, the iPad/Mac/foldable overlap becomes even more obvious.
Key takeaway
- This product is interesting because it points to a real future.
- But it’s not that future yet.
AI Turns a PHP Vulnerability Into an Internet-Scale Security Problem
What happened
- A security research team used AI with a very simple prompt: “find bad vulnerabilities in PHP.”
- PHP is foundational infrastructure for a huge percentage of websites, including major CMS platforms like WordPress.
- The discovered flaw(s) could allow attackers to gain deep access to servers running PHP.
Why this matters now
- The hosts emphasize that this exploit may have existed for a while, but AI changes the game:
- What once required lots of human labor can now be automated.
- Attackers can use agents to scan the web at scale and keep trying until they hit the vulnerability.
- That turns a niche exploit into a broad, fast-moving threat.
Real-world consequences
- PHP maintainers recommended immediate patching, or taking affected systems offline.
- The impact reportedly reached government websites, Wikipedia, and the Internet Archive.
- The story becomes a real example of an AI-driven arms race between offensive and defensive security.
The big picture
- The hosts argue this is the future of cybersecurity:
- Offensive AI will hunt vulnerabilities.
- Defensive AI will need to patch and monitor at machine speed.
- Human-in-the-loop processes may be too slow to keep up.
Brendan Carr and Poll Censorship: A Straightforward First Amendment Problem
What Brendan Carr said
- Carr, head of the FCC, threatened broadcasters over polls he claimed were fake or “suppression polls.”
- He suggested broadcasters might violate their obligations by airing them without enough context.
Why the hosts say this is dangerous
- The FCC does not have authority to punish media outlets for publishing polls it doesn’t like.
- This is framed as a chilling attack on speech before an election.
- Even if a poll is bad or misleading, reporting it is still protected speech.
The broader concern
- The hosts argue Carr is trying to expand the government’s power from broadcast regulation into broader speech control.
- The point is not just the specific poll issue—it’s about normalizing government threats toward any media outlet that says something the administration dislikes.
Their conclusion
- They call it a direct attack on the First Amendment.
- They stress that this is especially dangerous in the lead-up to an election.
Lightning Round
Arc Browser Is Shutting Down
- The Browser Company is officially ending Arc in December.
- The hosts call it a shame because Arc influenced the entire browser market:
- vertical tabs,
- browser-as-operating-system thinking,
- and a more opinionated browser UX.
- They note that many newer AI browsers are effectively borrowing from Arc’s ideas.
Meta’s Terms of Service Controversy
- Meta updated its business terms in a way that appears to allow broad scraping of websites associated with business accounts.
- The company later said it wasn’t trying to train models on that content, but the hosts and lawyers quoted in the reporting argue the language still effectively allows it.
- The concern:
- publishers,
- creators,
- and businesses may be unintentionally granting Meta permission to use their sites and content for AI and ad-related purposes.
Netflix’s Content Reset
- The hosts discuss reporting that Netflix is rethinking its content strategy after disappointing performance and engagement trends.
- Main points:
- Too many weak movies and shows.
- Too much reliance on cheap reality content.
- Binge-dropping all episodes may no longer be the smartest strategy.
- They argue Netflix needs to produce more shows people actually care about, not just more content.
Google Gemini Is Still a Confusing Product
- The hosts poke fun at how Google seems to treat Gemini as different things on different devices:
- phone,
- car,
- smart home,
- and app.
- In the car, Google says Gemini can access Gmail, Calendar, Maps, Photos, etc., but only if users enable “personal intelligence.”
- The problem is not the feature itself—it’s that Google’s product story around Gemini still feels inconsistent and fragmented.
Overall Takeaways
From the iPhone review debate
- The device is innovative, but it’s still a compromised first-generation product.
- It makes more sense as a future iPad than as a current iPhone.
From the security story
- AI is rapidly lowering the barrier to large-scale vulnerability discovery and exploitation.
- The internet’s security model is entering a much more dangerous phase.
From the policy segment
- Brendan Carr’s polling threat is presented as a serious and direct abuse of regulatory power.
From lightning round
- The browser wars, AI platform wars, and content wars are all shifting fast.
- The common theme across all of them: the internet is being reorganized around AI, and no one has fully figured out the rules yet.
