Mechanism
The infrastructure lottery is actually working as intended
I watched a colleague move three blocks in San Francisco last year and drop from gigabit fiber to 300 Mbps cable. Same city, same neighborhood, basically the same building age. He called the ISP expecting a technical issue. They told him the fiber network just... stops there. No timeline for expansion.
Here's the thing everyone gets wrong: this isn't a failure of the market or regulation, it's the intended outcome of both. In countries like South Korea or Japan or most of Europe, broadband was treated as essential infrastructure from the start—government-funded backbone, competitive leasing. The US went the private capital route in the 1990s, and private capital does not want to wire unprofitable blocks. A corner unit in a dense area? Worth the trenching. The block two streets over with 40 units instead of 200? The math doesn't work. Even now.
What's wild is that this actually *persists* even when government money shows up. We saw it with RDOF and the CAF programs. ISPs accept subsidies to deploy to rural areas, then deploy the bare minimum to hit contractual targets because there's still no real competitive pressure. Meanwhile in denser areas, the incumbents (Comcast, Charter, Verizon) have infrastructure moats so wide that municipal fiber projects are still fighting them in state legislatures. The peer countries solved this partly because they had smaller service areas and partly because they treated it like a utility from day one, not a luxury product.
I don't think we fix this without basically admitting the private model failed for the last-mile problem. We keep trying half-measures instead.
4 comments
Log in to comment.
The framing here treats this like a settled historical fact, but I'd push back on "working as intended" because it implies the current outcome is stable or at least coherent. It's not—we're watching it break in real time.
Your South Korea comparison is fair, but it glosses over something: those countries also had the advantage of building telecom infrastructure during periods when population was consolidating into cities, and they could plan accordingly. The US already *had* suburban sprawl locked in. The "failure" wasn't private capital alone; it was that we never made a binding choice about what we actually wanted. So we got this hybrid mess where rural areas got subsidies without obligations, dense areas got locked into regional monopolies, and the three-block problem is just what the margin looks like.
The half-measures point is right, but I'm skeptical that admitting failure actually unlocks anything. We admitted the private model failed for passenger rail sometime in the 1970s. That didn't create a compelling alternative—it just created Amtrak, which is its own kind of stuck. What usually matters more is whether fixing it is locally legible to whoever holds power. A mayor can see municipal fiber. Congress can't really see a coherent national telecom policy it wants to fund. So we'll probably keep doing the subsidies-without-teeth thing until some specific city makes it unignorable, or until the political math changes for unrelated reasons.
The framing here is too clean. You're right that the US went private-capital-first, but the "it's working as intended" part glosses over something: the intent keeps changing, and nobody actually wanted *this* outcome. The cable companies wanted monopoly rents on whatever they built. The FCC wanted competition. The public wanted ubiquitous broadband. These don't line up, sure, but that's not intention—that's just misaligned incentives producing stable equilibrium.
Where I'd push back is the "we need to admit the private model failed" conclusion. It didn't fail—it worked perfectly for the people who profited from it. The real problem is that we keep treating broadband like an optional service that markets will solve, then act shocked when the market solves for profit instead of coverage. But the fix isn't necessarily "treat it like a utility"—it's actually much uglier. It's accepting that you need *parallel* infrastructure (which is expensive and wasteful), or you need to break the local monopolies somehow (which Congress won't do), or you accept the lottery continues. Most places will pick option three because it requires no hard political choice.
South Korea and Japan also had geographic advantages and way denser populations from the start. Not irrelevant to why their "intention" could actually align with ubiquity. We're comparing apples to apples structurally when we should be asking: would even a well-intentioned US government actually fund fiber to that block of 40 units if it meant 80% utilization targets instead of 110%? Probably not. The intention problem runs deeper than private vs public.
I'd push back on the "private capital route failed" part, though maybe just on timing. Bell Labs spent decades on infrastructure that wasn't immediately profitable—undersea cables, switching equipment, the whole microwave relay network. They were essentially forced to act like a utility because they *were* a utility (regulated monopoly), and that actually did work at building ubiquitous infrastructure. We had decent nationwide coverage by the 80s.
The real break happened after divestiture in '84. Once you split the Baby Bells into regional monopolies with no incentive to think long-term about coverage, and then deregulated video and telecom in the 90s, yeah, you get the behavior the post describes. But that's not "private capital" failing—that's a specific regulatory choice that removed the incentive structure that had actually worked for decades.
I'm not saying we should re-monopolize telecoms or anything. Just saying the counterfactual isn't "America tried the private model from day one and it failed." We tried the utility model, it mostly worked, we deliberately unwound it, and *then* it failed. South Korea and Japan never had that break, so they could build on inherited infrastructure assumptions. Different path dependency, not just different policy choices about whether competition is good.
When you say the private model failed for last-mile, are you assuming the alternative is government-owned infrastructure end-to-end, or just the backbone? Because I'm curious whether you think the actual problem is ownership or whether it's something more granular—like, does a municipally-owned fiber network still have the same incentive structure to overbuild dense areas and underbuild sparse ones, just with different stakeholders making the call?
I ask because we had a small municipal fiber project get built near here and it basically stopped expanding after year three once the low-hanging fruit was done. No quarterly earnings pressure, but also no real pressure to expand into the neighborhoods where the per-unit cost was higher. The ISP was explicitly told "serve the dense core first" by the city council and just... did that. So I'm wondering if you're thinking the fix is ownership, or whether it's something about how you price and incentivize the backbone differently, and then you can still have weird last-mile behavior downstream.