Skeptic

Britain is moving toward an under-16 social-media ban, and the headline understates the institutional shift behind it. A rule framed as “keep children off social media” does not stay confined to children, because enforcement requires platforms to decide who counts as a child and who does not. Once that happens, age assurance stops being a niche safety tool and starts becoming a general access condition for the wider population. Reports say Starmer is preparing to announce the move after strong parental support in the government consultation, and ministers are also considering extending the logic beyond classic social apps toward gaming and chat-heavy platforms. Access by underage users should be controlled. The problem begins with the enforcement architecture chosen to do it. The story starts with politics. Starmer had been reluctant to back a blanket ban, then shifted after a consultation that drew heavy support from parents and after Australia’s under-16 ban became the international reference point. The likely scope is broad enough to matter: not only Instagram, TikTok, Facebook, X, and YouTube, but potentially Discord, Roblox, Fortnite, Minecraft, and other spaces where chat, discovery, and contact with strangers are central features. Once the category expands that far, the policy stops looking like a narrow fix for one corner of the internet and starts looking like a gateway rule for ordinary online participation. The technical problem follows immediately. A platform cannot reliably exclude under-16s unless it can identify, classify, or verify everyone else as old enough. That is the hidden arithmetic of the policy. The public hears “ban children.” The platforms hear “verify the age of your whole user base or accept legal risk.” This is why age-gating for minors tends to become de facto age-checking for adults. Once a service must prove that minors are being kept out, self-declared ages and weak honor systems stop being enough. Britain already has part of that infrastructure in motion. Under the Online Safety Act, platforms dealing with pornography and certain high-risk content have had to introduce highly effective age assurance, and real services have responded with face scans, ID uploads, and payment-card checks. Bluesky rolled out age verification in the UK using face scans, ID checks, or payment-card methods, and users who did not verify lost access to certain functions. Reddit was later fined by the ICO over children’s-data failures, with the regulator criticizing the company’s earlier reliance on weak age checks. The pattern is already visible: once legal liability rises, platforms move toward harder forms of age assurance. That is why the under-16 ban points toward a much broader identity layer. The government can say this is not national digital ID because the checks are outsourced to private vendors and handled platform by platform. The user experience can still converge on the same reality: repeated proof-of-age requests, face scans, document uploads, third-party brokers, and growing pressure to standardize credentials across services. The distinction between age assurance and identity verification narrows in practice even when it remains neat on paper. The strongest version of the concern does not depend on every platform suddenly demanding a passport. The concern is structural. Once platforms face fines for getting age wrong, they will prefer methods that are defensible, scalable, and hard to game. That pushes them toward biometric age estimation, ID-based checks, payment verification, carrier signals, or some combination of them. A government that wants a child-safety headline can therefore create the commercial and regulatory incentives for a general verification regime. Australia’s law already works in that direction, and the UK debate is openly borrowing from that model. The privacy question comes later in the political sequence, but it is embedded from the beginning. A fragmented ecosystem of age-assurance vendors can still become a national identity layer by accumulation. More checks produce more pressure for interoperability, fewer repeated verifications, and eventually some standard credential or wallet. That is how a child-protection policy can become a de facto digital identity architecture without Parliament ever passing a law called digital ID. The formal target remains under-16s. The operational consequence falls on everyone old enough to stay. The point to keep in view is simple. The story begins as online child safety. The implementation logic points toward universal age classification. The likely endpoint reaches far beyond children and pushes adults toward proving who or what they are before ordinary participation in large parts of digital life.

Alphabet’s private investment strategy works less like a conventional venture portfolio and more like a growth-capture architecture built around future control points in technology. The objective goes beyond markups or eventual IPO gains. Alphabet is placing capital and control around AI, cloud infrastructure, enterprise data, payments, autonomy, and long-duration platform shifts so that future growth can flow back toward Google through ownership, demand, or strategic dependence. The most important position is Anthropic. Alphabet’s planned investment of up to $40 billion, combined with Anthropic’s reported $200 billion cloud and chip commitment to Google over five years, makes this far more than a financial stake in a frontier-model company. It gives Alphabet exposure to upside in one of the strongest independent AI labs while also anchoring massive demand for Google Cloud and the TPU stack. Alphabet is therefore positioned both inside the AI race through Gemini and around it through ownership in a major external model builder that also depends on Google infrastructure. That arrangement gives Alphabet unusual resilience, but it also introduces circularity. Google is helping finance a company whose spending then helps fill Google’s own backlog. The strategic logic is clear, but part of the value depends on Anthropic’s projected growth actually materializing. If it does, Alphabet benefits from both the equity upside and the infrastructure demand. If it does not, part of that demand story becomes less secure than the headline commitment suggests. The next layer sits in enterprise software and infrastructure. CapitalG’s positions in Databricks and Stripe place Alphabet near two of the most valuable adjacencies to AI and cloud expansion. Databricks gives it exposure to the enterprise data and AI workflow layer. Stripe gives it exposure to digital transaction infrastructure. These are positions in the software and payment rails that should expand as cloud and AI adoption deepen. A separate bucket sits between minority stakes and ordinary commercial relationships: businesses Alphabet still controls or substantially shapes while using outside capital to scale them. Waymo is the clearest example. It remains an Alphabet-controlled company even after raising outside money, so the strategic value lies in retained control over one of the most important autonomous-mobility platforms in the world rather than in a simple portfolio stake. Isomorphic Labs belongs in the same category. Alphabet founded it, still participates in its funding rounds, and is using it as a long-horizon platform bet on AI-driven drug discovery. These companies are not external wagers in the usual sense. They are controlled option bets on large future markets. Broadcom belongs nearby, but under a different logic. There is no verified material Alphabet ownership stake. The strategic importance comes from deep dependence rather than equity ownership. Broadcom is tied into Google’s custom AI chip roadmap through a long-term agreement to co-develop and supply TPUs through 2031. That makes Broadcom part of Alphabet’s future architecture even without a meaningful equity position. SpaceX remains a long-duration optionality case. Google’s original investment alongside Fidelity was one of the most important private technology stakes of the last decade, even though the exact current Alphabet holding is no longer publicly clear and should not be overstated. The relevance now lies less in the old percentage and more in the kind of platform exposure the investment still represents: launch, satellite connectivity, defense relevance, and strategic infrastructure. Taken together, Alphabet’s private bucket shows a layered future-growth architecture rather than a simple portfolio of bets. Anthropic ties Alphabet directly to frontier AI while reinforcing its cloud and chip business. Databricks and Stripe place it near enterprise data and transaction infrastructure. Waymo and Isomorphic Labs keep Alphabet in control of major long-duration bets in autonomy and AI-enabled science. Broadcom secures a critical non-equity layer of the TPU stack. SpaceX adds longer-range communications and infrastructure optionality. Alphabet is not only positioning itself near the growth engines, bottlenecks, and infrastructure layers likely to define the next decade. In several cases it is helping build those layers or retaining control over the companies operating them.

Every time you remember something, you’re not replaying the memory. You’re rewriting it. Every single time. That childhood story you love? It’s already a little bit fiction.