The Great Silicon Valley Reckoning: When Tech Ate Its Own Rules

---

A 20-year-old man threw a homemade bomb at Sam Altman’s gate. The authorities found a list of AI leaders in his possession. Meanwhile, Snapchat just fired 1,000 people because it thinks AI will do their jobs better. Allbirds, a sneaker company that couldn’t figure out footwear, is now rebranding as an AI chip company. And the UK government is calling social media executives to Downing Street to explain why they can’t keep children safe.

This is what happens when an entire industry builds on the assumption that disruption is always good, move-fast-and-break-things works at scale, and the market will eventually solve everything.

It’s not working.

The Safety Theater Is Over

The meeting at Downing Street isn’t surprising—it’s overdue. Meta, YouTube, and the rest have been promising to protect kids online for, what, fifteen years now? And yet here we are in 2024 asking the same executives the same questions. The difference this time is that governments are finally treating it like a criminal matter, not a marketing problem.

Photo by Robert So / Pexels

What’s interesting isn’t that they’re being called in. It’s that the companies have been given no real deadline, no specific mandate, no teeth. It’s a conversation. A strongly-worded one, sure, but essentially the tech equivalent of a parent asking their teenager if they’re really sure they’re making good decisions online.

The Palantir situation tells a different story. NHS guidance that hospitals should use Palantir’s data analytics software has sparked actual pushback from MPs. This isn’t abstract. This is the government literally deploying surveillance-adjacent technology into healthcare systems, and British politicians are asking uncomfortable questions about data use and oversight. The company is defending itself, which means the old playbook—announce, implement, apologize later if necessary—isn’t working the same way it used to.

My read: We’re at an inflection point where “we’ll self-regulate” stopped being a credible argument around 2018, and we’re only now seeing the regulatory response catch up. The UK’s approach with Palantir is the template. Specific scrutiny. Named companies. Real friction. This is coming to the US eventually, and it’s going to be messier because the US government moves slower but hits harder when it lands.

The AI Jobs Story Is Backwards

Snapchat’s cutting 1,000 people—16% of staff—because AI will handle repetitive work. Simultaneously, there’s a new theory called “Jagged Intelligence” floating around that reframes how we think about AI. The basic idea: AI isn’t a replacement for human intelligence, it’s a different shape. It’s good at some things, terrible at others, and that mismatch might actually be what saves certain jobs.

Then there’s this: “That Meeting You Hate May Keep AI From Stealing Your Job.” The argument is that as AI handles more tasks, the human work of persuasion, negotiation, and reassurance becomes more valuable.

These pieces are fighting each other in the same ecosystem. Snapchat is betting that AI does everything better. The “Jagged Intelligence” crowd is saying AI has blind spots so large that human judgment becomes premium. And the meeting story is saying that the interpersonal labor nobody wants to do is suddenly the work nobody can automate away.

I think they’re all partially right and the real story is that the job market is about to fragment in weird ways. Some roles get crushed—pure data entry, routine coding, certain types of analysis. Some roles get turbocharged—a designer with AI tools can do five times the work. And some roles become weird hybrids where you need a human to babysit the AI or fix what it breaks or, yes, convince the board to actually use the thing.

Snapchat’s playing a dangerous game. If those 1,000 people were handling messy human stuff—dealing with ad partners getting mad, handling trust and safety escalations, managing creative relationships—then the company just fired its immune system. We’ll know in about eighteen months if that worked.

Photo by nappy / Pexels

The Infrastructure Wars Are Hot

Amazon just bought Globalstar for $10.8 billion. This is satellite internet. This is Elon Musk’s Starlink territory. This is the infrastructure that, in 2030, will determine whether rural areas can compete with urban ones, whether you can get reliable connectivity in a crisis, and whether Bezos or Musk controls another layer of the internet.

The amount of money being spent on satellite internet right now is genuinely wild. SpaceX launched Starlink assuming they’d get maybe one percent market share. Instead, it’s approaching viability for actual consumers. Amazon watched that happen and decided to spend eleven billion dollars to not be left out.

This isn’t even about disruption anymore. This is about infrastructure moats. Once you’ve got tens of thousands of satellites in orbit and the regulatory approval and the ground stations, a competitor can’t just spring up. You’ve won a generation of control. Bezos knows this. He’s not building this to lose money on broadband. He’s building it to own the pipes that everything else runs on.

What concerns me: This is a bet on a world where governments let these companies own orbital infrastructure with minimal oversight. The regulatory environment around satellite internet is weirdly permissive right now. That won’t last. When it tightens, Amazon and SpaceX will have already cemented their position. That feels like an error waiting to happen.

The Booking.com Hack Is a Symptom

Booking.com got hacked. Customer reservation data was compromised. They changed their security system (they called it a PIN, which is hilarious and also depressing). They won’t say how many people were affected.

This is almost boring now, right? Travel company gets hacked, offers vague assurances, moves on. But it’s the template for how these breaches actually happen to consumers: slow detection, delayed disclosure, and never enough detail to actually know if you need to worry.

The fact that they won’t disclose the scale suggests it’s bad enough that saying it out loud would trigger actual response—class actions, regulatory inquiries, credit monitoring demands. So they don’t. And people book hotels slightly nervously for a few weeks until they forget.

Google’s Back Button Crusade

Google is punishing sites that trap users by disabling the back button. This is scheduled for June. This is the company that built its entire empire on the principle of “let the user leave” essentially saying out loud: some of you have forgotten how the web is supposed to work.

The sites that do this—trap users with broken back buttons, hijack navigation—are usually garbage. Scam sites. Low-rent ad farms. Places that know they can’t retain attention through quality so they resort to dark patterns.

What’s darkly funny is that Google has to tell people to stop doing this. It’s not even controversial. The back button is foundational. And yet here we are in 2024 with Google needing to threaten punishment to get sites to respect a basic web convention that’s been around since 1995.

This is what happens when the entire monetization structure is based on engagement metrics and time-on-page. You get perverse incentives. You get companies deciding that trapping the user for an extra thirty seconds matters more than the user’s experience. And you get Google standing up and saying: no, actually, we’re going to make this cost you.

I respect this move, but it also feels like patch-on-patch-on-patch. The real problem is the incentive structure. Fix that and you don’t need to police back buttons.

What I’m Watching

• Palantir’s NHS rollout timeline and pushback escalation. If MPs successfully slow or block hospital adoption, it signals that data privacy concerns are now a real veto point for tech deals in government. Watch for the first NHS trust to publicly refuse Palantir. Expected window: Q2-Q3 2024.

• Snapchat’s engagement metrics over the next two quarters. If the 1,000-person cut leads to slower feature development and user churn accelerates, it proves that some jobs can’t be replaced by AI without losing the company’s entire value prop. This will be the canary in the coal mine for other tech layoffs.

• Amazon Globalstar integration announcements. When Bezos starts talking about launch dates for actual satellite-internet-to-consumer service, that’s when the infrastructure war becomes real and visible. Also watch for any regulatory pushback. Expected: Q3 2024.

• The Allbirds/NewBird AI rebrand’s actual product. They’re buying chips and rebranding. If they launch something with actual customers in six months, this is a real pivot. If they’re still in “acquiring GPUs and making announcements” mode next year, this is just a press release with capital allocation attached.