Schueller uses his own 25 Gbit/s symmetrical connection as a case study, arguing that Swiss municipalities building and owning fiber — then leasing dark fiber to competing ISPs — creates genuine retail competition. He documents how this model delivers 50-100x faster speeds than American broadband at comparable prices, with 5-10 ISPs competing on the same physical infrastructure.
The post's title — 'The Free Market Lie' — frames the core argument: American broadband isn't a market failure but a policy failure. Over $400 billion in taxpayer subsidies since the 1990s created a duopoly rather than competition, with the 1996 Telecommunications Act enabling consolidation instead of the competition it promised. The US model of letting private companies own the physical layer produced the opposite of free-market outcomes.
The editorial synthesis emphasizes that the most substantive discussion threads weren't about politics or speed benchmarks, but about downstream engineering consequences. Symmetrical upload speeds enable fundamentally different application architectures — self-hosting, peer-to-peer, distributed systems — that throttled American upload speeds structurally prevent. The editorial frames this as infrastructure decisions determining tech outcomes decades downstream.
A blog post by Swiss developer Stefan Schueller hit 321 points on Hacker News with a straightforward premise: his home internet connection in Switzerland delivers 25 Gbit/s symmetrical — meaning upload matches download — for roughly the equivalent of $50 USD per month. The median American household, meanwhile, pays a similar amount for a connection roughly 50-100x slower, with upload speeds often throttled to a fraction of download.
The post isn't really about speed benchmarks. It's a case study in how infrastructure ownership models determine tech outcomes decades downstream. Schueller walks through how Swiss municipalities built and own the physical fiber network, then lease dark fiber to any ISP willing to compete on that infrastructure. The result: genuine retail competition on a publicly-owned physical layer. Customers in many Swiss cities can choose from 5-10 ISPs offering service over the same fiber, competing on price, speed tiers, and service quality.
The Hacker News discussion — hundreds of comments deep — split predictably along US-vs-Europe lines, but the most substantive subthreads weren't about politics. They were about engineering consequences.
The US broadband story is well-documented but bears repeating with current numbers. American taxpayers have subsidized telecom infrastructure to the tune of over $400 billion since the 1990s, through a combination of direct subsidies, tax breaks, and rights-of-way grants. The Telecommunications Act of 1996 was supposed to foster competition. Instead, it enabled a consolidation wave that left most US markets with one or two viable broadband providers — a textbook duopoly.
The Swiss model inverts the American assumption that private companies should own the physical layer. In Switzerland, the municipality builds the fiber (often alongside other utilities like water and power), owns it permanently, and operates it as shared infrastructure. ISPs then compete on the service layer — routing, peering, pricing, customer support. This is structurally identical to how roads work: the government builds and maintains them, and private companies (trucking firms, delivery services, ride-shares) compete on top.
The results aren't subtle. Switzerland consistently ranks in the top 3 globally for average broadband speed. The US, despite spending more per capita on telecom infrastructure than almost any other developed nation, ranks in the low 20s depending on the measurement. The gap is widening, not closing. The 2021 Infrastructure Investment and Jobs Act allocated $65 billion for broadband, but early implementation reports suggest the money is flowing through the same incumbent carriers who failed to deliver on previous subsidy rounds.
What makes the HN discussion technically interesting is the upload speed dimension. Most US cable and DSL connections are heavily asymmetric — Comcast's standard plans offer 1 Gbit/s down but cap upload at 35 Mbit/s. This asymmetry was a deliberate architectural choice baked into DOCSIS cable standards, optimized for a consumption model (streaming video) rather than a production model (hosting, uploading, real-time collaboration). Symmetrical fiber isn't just faster — it enables fundamentally different usage patterns that asymmetric connections make impossible regardless of download speed.
If you're a developer in the US, the upload bottleneck shapes your architecture in ways you might not consciously register. Self-hosting is impractical when your upload is 35 Mbit/s. Pushing large container images, syncing monorepos, running CI locally, streaming high-resolution pair programming sessions, serving a home lab to the public — all of these are either painful or impossible on typical American residential connections. The US upload deficit doesn't just slow developers down; it drives architectural decisions toward centralized cloud hosting that wouldn't be necessary with symmetrical infrastructure.
This has real cost implications. A developer in Zurich can run a legitimate home server on a 25 Gbit/s symmetrical connection for $50/month. An American developer doing the same work rents a cloud VM for $50-200/month because their home upload makes self-hosting non-viable, then pays again for the home connection they can't fully use. The Swiss developer has one bill. The American developer has two — and the cloud bill scales with usage.
For teams doing remote work with large assets — game studios, ML teams pushing model weights, video production — the upload gap is even more consequential. A 10 GB model checkpoint uploads in about 3 seconds on a 25 Gbit/s connection. On a 35 Mbit/s US residential upload, that same transfer takes 38 minutes. That's not a performance difference. That's a workflow difference.
The municipal fiber model also has implications for edge computing and decentralized architectures. When every residential connection is symmetrical multi-gig, peer-to-peer protocols, federated services, and edge nodes become practical at the household level. The US upload bottleneck effectively centralizes the internet by making endpoints too slow to serve — which conveniently benefits the cloud providers and CDNs who fill that gap.
The post's title — "The Free Market Lie" — is provocative but precise. The US broadband market hasn't functioned as a free market in any meaningful sense since at least the early 2000s. Incumbent ISPs hold government-granted monopolies or duopolies in most markets, use regulatory capture to block municipal broadband initiatives (over 20 US states have laws restricting or banning municipal broadband), and face no meaningful price or quality pressure from competition.
When US cities have built municipal fiber — Chattanooga, Tennessee being the poster child — the results mirror Switzerland's: faster speeds, lower prices, and genuine retail competition. Chattanooga's EPB offers 25 Gbit/s symmetrical for $300/month (the 1 Gbit tier is $68/month), and the economic development impact has been substantial enough that other cities cite it as a model. But Tennessee's state legislature, lobbied by AT&T and Comcast, passed laws preventing EPB from expanding its network beyond its existing service area.
The pattern is consistent: wherever the infrastructure-vs-service separation is implemented, outcomes improve. Wherever incumbents control both layers, they optimize for margin extraction rather than capability. This isn't an ideological claim — it's what the data shows across dozens of implementations in Europe, Asia, and the handful of US municipalities that managed to build before state preemption laws blocked them.
The $65 billion in US federal broadband funding currently being deployed is the largest infrastructure investment since the original AT&T monopoly buildout. Whether it produces Swiss-style outcomes or repeats the 1996 Telecom Act's failures depends entirely on ownership structure. If the money flows to incumbents who own the physical layer and face no open-access requirements, expect the US to still be debating upload speeds in 2035 while Swiss teenagers saturate 50 Gbit/s connections. The engineering isn't hard. The physics is solved. The question is whether the entity that owns the pipe has any incentive to make it fast — and in the US, the answer has been "no" for thirty years running.
Most states in America ban municipal fiber. They saw EPB (Chattanooga) and said, no, we must make sure that doesn't ever happen again. That is how 'free' market is done in US, all the rules are to make sure the richest people become richer.
Swiss here. Just a minor clarification on the article: fibre is not available everywhere in Switzerland. Actually the rollout has been quite slow, and chances are that if you live in a rural or suburban area (like me) or in an older building then you might not have fibre, and are limited to (fairly
I don't understand the desire (fetish?) for high speed home Internet connections at home.I have 25 Mbps up. 10 Mbps down. Have had it for years. It's fine.It's fine when both my wife and I are working from home and doing calls. It's fine for software development. It's fine f
In Canada our internet became much faster for cheaper with better customer support when the government allowed competition from smaller players. Telecom also got better when they allowed a foreign competitor to compete against the government mandated oligopoly. But the market is still heavily regula
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
In my small island community, I participated in a municipal committee whose mandate was to bring proper broadband to the island. Although two telecom duopolies already served the community, one of them had undersea fiber but zero fiber to the home (DSL remains the only option), whereas the other use