The last-mile problem didn't get cheaper

You're right that the regulatory structure created perverse incentives, but I'd push back slightly on how inevitable this feels. The boring part isn't actually that boring once you zoom out—it's a genuinely hard optimization problem that got locked in wrong, which is different from "markets can't solve infrastructure." The thing is, cable companies made a rational bet thirty years ago: last-mile coverage is expensive and geographically lumpy, so you build where density justifies the capex. That math still mostly holds. Fiber to a neighborhood of single-family homes three blocks from a major trunk is legitimately expensive per-home-passed, and the ROI gets weird when you're competing against an installed base of coax that already works. It's not *just* that they're rent-seeking—it's also that we never actually created a scenario where someone else could credibly build and operate that last mile cheaper. You can't have competitive last-mile deployment when the incumbent owns the ducts and the pole access. That's the real lock-in. Where I agree with you completely: municipal fiber only works where cities have patient capital and legal authority, which requires treating it like water service, not a competitive market. Stockholm and Tokyo got there because they either never privatized it or they stepped in early enough that the switching costs were bearable. We optimized for consumer choice in the short term and got monopoly control in the long term instead. Starlink's advantage isn't that it's better—it's that it doesn't need the same physical infrastructure, so it actually breaks the lock rather than just competing at the margins. That's the only thing that's moving the needle right now, which tells you something depressing about how stuck the problem actually is.

The comparison to Japan and Europe keeps showing up, but those countries didn't exactly "decide" their way out of this—they had different initial conditions that made public ownership look inevitable rather than like a deliberate policy choice. Japan's NTT was already a state monopoly. Stockholm's fiber came through existing municipal utility corridors. It's easier to rationalize as "they treated it like a utility" than to admit we'd have needed that same path dependency thirty years ago, before the private buildout had already calcified. The real problem isn't that market forces can't solve last-mile connectivity. It's that once you've let a private actor own the infrastructure for two decades, switching costs become astronomical. Municipalities could theoretically overbuild Comcast today, sure, but they're competing against an incumbent that has already recovered its deployment costs and just needs to maintain the network. A new entrant has to make the capital case for deployment in a market where the price floor is whatever the monopolist decides. That's not a market problem anymore—that's an arithmetic problem. You can't regulate or municipalize your way out of sunk costs that are already sunk. Starlink might actually break this, but not because it's elegant. It breaks it because satellite forces the incumbent to compete on terms where their owned infrastructure is suddenly a disadvantage. That's not a solution to the incentive problem—it's just introducing a different technology whose incentive structure happens to point elsewhere. We'll probably end up with okay connectivity everywhere eventually, but let's not mistake that for having solved why the market didn't do it the first time.

You're describing the symptom correctly but I think you're being too gentle on the timeline. The "market will solve it" crowd has had their shot since the late '90s. We're not waiting anymore—we're just watching the same decision get made over and over in different cities, each one thinking they're the first to discover that incumbent carriers won't voluntarily upgrade their own moat. The thing that gets me is the regulatory capture part is almost incidental to the real problem: once you own physical infrastructure, you've basically won forever unless someone with deeper pockets or legal authority decides to compete. Starlink's not actually solving last-mile economics—it's solving "what if we could make it someone else's problem to build ground stations." And sure, that breaks the incumbent's monopoly power in specific places, but it's not cheaper, it's not more reliable, it's just available. The OP's neighbors' fiber gaps could exist indefinitely because Comcast will drag its feet on upgrades as long as the cable business generates cash, and Verizon will do the same with copper. The depressing part—and maybe why this keeps coming up—is that municipal fiber keeps winning when cities actually try it, but it's slow and political and requires treating infrastructure like infrastructure instead of a venture. So we get this weird half-state where everyone agrees the market failed, but admitting that requires actually doing the work instead of waiting for disruption. Easier to complain about it on forums.

The thing that clicks for me here is that you're describing two different failure modes getting confused into one problem. The regulatory/monopoly story is real, but it's not actually about why the gap persists—it's about why it formed in the first place. What's keeping it alive in 2024 is something quieter: the last-mile problem genuinely *is* expensive, and no amount of regulation retroactively makes physics cheaper. Fiber to a single house costs real money. So does maintaining it. A municipal ISP in a dense neighborhood can amortize that over enough subscribers to make the math work. Same for Japan and the Nordic countries—high density, high taxation tolerance, state capital willing to take decades to ROI. But the "three blocks apart" scenario is the actual hard case: you need to serve both the gigabit house and the 12 Mbps house with the same economics, and geography will fight you. The incumbent's mistake wasn't laying fiber where it was profitable; it was never seriously trying the unprofitable parts. Starlink breaks the stalemate not because it's cheaper, but because it sidesteps the problem entirely—you don't need *our* last mile. The regulatory angle matters, sure. But I'd flip it: we didn't need to treat this like a utility in the abstract. We needed to make it obvious that *some* parts of the network were infrastructure and *some* parts weren't, and price accordingly. Instead we let the profitable parts subsidize executive bonuses while the hard parts stayed unsolved.