In a fascinating blog post, Stefan Schüller unpacks how this came to pass, in Switzerland, a country known for its impassable mountains and its impossible national telco (Swisscom):

https://sschueller.github.io/posts/the-free-market-lie/

Schüller describes the Swiss system as a kind of Goldilocks approach that's midway between two failed systems: the American "free market" system and the German state provision system.

2/

Most people in the US can't get fiber at all, and if you can get it, it's probably 1Gbit, and available from a single provider (that's nearly my situation in Los Angeles, where I can buy 2Gbit symmetrical fiber from AT&T, who run a shared connection on old Worldcom fiber they've lit up).

3/

Some (very foolish) people say that Starlink represents a competitive alternative to fiber. This is nonsense - first, because Starlink is another natural monopoly (how many competing satellite constellations can we cram into stable orbits before they start smashing into each other?), and second, because satellite is *millions of times slower* than fiber:

https://www.somebits.com/weblog/tech/bad/starlink-nov-2022-data-caps.html

4/

In Germany, most people also have a single provider, and the connection they get is shared, and caps out at 1-2Gbit.

Meanwhile, the Swiss can get connections that are *far* faster, and cheaper. How did they do it?

For starters, the Swiss recognized what any Simcity player knows: fiber is a "natural monopoly." It doesn't make any sense to build multiple, competing fiber networks - any more than it would make sense to build multiple, competing sewer systems or electric grids.

5/

In the US, private fiber providers get city permits to dig up the roads and lay their network. If you have two competing networks, they dig up the road twice.

You'd think that the (more regulated) Germans would lay a single network, but they, too, have multiple, competing networks.

6/

German regulators have a complex set of priorities and constraints: to encourage competition, they promote the idea of competing networks in competing trenches, often just meters apart (rather than on competing services running over the same fiber and/or fiber run through the same conduit - pipe - laid in a single trench).

7/

This makes setting up fiber extremely capital-intensive, so Germany backstops this system with "essential facilities sharing" - a rule that requires the incumbent (formerly state-owned, now partially state-owned) Deutsche Telekom to offer space in its conduit to smaller ISPs that want to thread their own fiber from their data-centers to their customers' homes.

8/

This is a good idea in theory - but in practice, DT has largely captured its regulators and so it is free to place all kinds of administrative hurdles in the paths of competitors seeking to use its lines.

The result is that Germans can get fiber from multiple, heavily capitalized network providers who overbuilt redundant systems under the city streets, squandering capital digging trenches that they could have spent on providing faster and/or cheaper connections.

9/

Meanwhile, in the US, they leave this all up to "the market" (though, of course, there's no way "the market" could get fiber laid down without public participation, because the clearing price for privately negotiated licenses to dig up every street in town is "infinity").

10/

The US is dominated by a cartel of massive incumbents: there's AT&T (formerly a regulated monopoly that was so entangled with the US government that it was effectively a for-profit state enterprise) and the cable giants, Comcast and Charter, who divide up the country into exclusive territories like the Pope dividing up the "New World."

11/

These companies generally enjoy regional monopolies, which means they're less interested in making *profits* (money you get by mobilizing capital) than they are from extracting *rent* (money you get from sweating assets).

12/

For example, when Frontier went bankrupt in 2020, we got to look at its internal bookkeeping system, and learned that the company treated 1m customers who had no alternative carriers as special assets because it could charge them more for worse service and poor maintenance:

https://pluralistic.net/2022/12/15/useful-idiotsuseful-idiots/

13/

This means that US fiber networks tend to be *under*built (the opposite of Germany's overbuilt networks), meaning that even if you're buying "gigabit" fiber, you're probably sharing that one gig connection with your whole block or neighborhood, so you only get your nominal throughput at weird hours when all the other subscribers aren't streaming Netflix.

14/

(Note that there *are* cities in the US with a better situation; particularly cities served by Ting, which is owned by Hover, the amazing domain registry. Ting operates an excellent mobile carrier *and* a fiber networks in many cities. If you are lucky enough to have Ting as an option, then you should *treasure that option*.)

So, that's Germany and America. What did they do in Switzerland?

15/

For starters, they ran a four-strand, *dedicated* line (an insulated wire with four separate strands of fiber in it) to every house. That wire terminates at your wall with a "neutral, open hub." Any carrier can provide service over those four strands: Swisscom (the incumbent, majority state-owned carrier); Init7 or Salt (national, commercial carriers); or a local ISP.

16/

Each of the strands in your neutral hub operate independently. That means that you can switch from one carrier to another with a click. You can also run two or more carriers' signal through your hub, meaning that you can try out a new carrier before canceling your old one. The carriers compete on price, speed and customer service - but they don't compete on who can actually connect your home to the internet.

17/

The origins of this excellent system are in 2008, when Switzerland's Federal Communications Commission convened a roundtable to determine the future of the country's broadband. Incredibly, it was Swisscom that pushed for the multi-strand, dedicated fiber system, on the grounds that anything less would lead to monopolization.

18/

I say "incredibly," because in all my travels over the past three decades, a single encounter with Swisscom stands out as the most absurd and backwards run-in I ever experienced with a telco.

It was while I was working as EFF's delegate to the United Nations in Geneva, as part of an infinitesimal coalition of digital rights group convened by James Love and Manon Ress of Knowledge Ecology International.

19/

Geneva is not a forgiving city for someone working for a cash-strapped NGO: it's a city where everyone (except you) is on a lavish expense account courtesy of a national government, or (better still) an industry body that lobbies the UN.

My usual daggy two-star hotel (which cost as much as a four-star in London) didn't have its own wifi: instead, you signed on through Swisscom, which did not offer its own payment processing.

20/

To get onto the Swisscom wifi, you had to buy a scratch-off prepaid card that was good for a certain number of hours or minutes. The hotel was *always* sold out of these cards.

So my normal ritual upon my arrival in Geneva was to scour the tobacco shops around the train station for scratch-off cards.

21/

Normally, this would take four or five tries - the shops would either be completely sold out, or would only have the two-hour cards (needless to say, these were a *lot* more expensive on a per-hour basis than the one-day and multi-day cards).

On one trip, though, *all* the shops were sold out of these cards, so I skipped breakfast the next morning to wait outside the doors of the Swisscom offices.

22/

These opened five minutes late (the only business in Switzerland that wasn't achingly prompt!). The clerk let me in eventually, but when I approached his counter, he made me trudge to the opposite end of the room to take a number (I was the only person in the shop).

After an ostentatious delay, the clerk called out "Numero un!" and I went up to his counter and asked for a three-day card. No dice, he was sold out. Two-day cards? Nope. One-day? Uh-uh.

23/

He only had two-hour cards, too. Literally, the Swiss national telco had *run out of integers*.

This incident stuck with me so durably that I wrote it into my third novel, *Someone Comes To Town, Someone Leaves Town*. You can hear me read that passage here:

https://pluralistic.net/2020/08/17/aura-of-benevolence/#sctt-slt

24/

So it's frankly *amazing* to me to learn that Swisscom - who will forever be synonymous in my mind with the most catastrophically stupid internet delivery system imaginable - demanded this anti-monopoly fiber rollout.

But - as Schüller points out - Swisscom's foray into uncharacteristic reasonableness was short-lived.

25/

By 2020, the company had regressed to its mean, and was demanding an end to the neutral, four-strand, point-to-point system, petitioning for regulatory permission to switch to a cheaper, slower, shared hub-and-spoke system. This system wouldn't just be slower - it would also require all of Swisscom's rivals to rent access to its fiber, with Swisscom having the final say over who could compete with it and how.

26/

This went all the way to the Swiss federal courts, who ruled that Swisscom had failed to demonstrate "sufficient technological or economic grounds" for the change and fined the company CHF18m for wasting everyone's time with this stupid idea (that is, "violating Swiss competition law"). And so it is that, in 2026, you can get 25Gbit symmetrical fiber throughout Switzerland. *Wunderschön!*

27/

Schüller closes out his piece with a set of recommendations for countries hoping to replicate Switzerland's broadband miracle: open access to physical infrastructure; point-to-point service; neutral fiber standards; municipal fiber; and strong antitrust enforcement to keep the incumbent carriers in line.

These are great recommendations; they address the contradiction of regulated monopoly telcoms provision.

28/

On the one hand, these networks *are* natural monopolies, and they can only exist with extensive government intervention (at a minimum, to clear the way for poles, trenches and conduit for the physical fiber).

On the other hand, telcoms (especially broadband) play an important role in the political realm, because broadband connections are essential to civic and political engagement.

29/

You can't turn people out for a protest, or run an election campaign, a referendum, a ballot initiative, a regulatory notice-and-comment campaign, or even a campaign to get people to a public meeting or listening session without broadband.

This means that state-provided broadband is an incredibly tempting target for political corruption and regulatory capture.

30/

Think of all the terrible things that governments are doing with broadband regulation today, like Trump demanding that service providers turn over the identities and locations of his political enemies so that ICE can hunt them down and kidnap or murder them; or "age verification" systems that accumulate mountains of easily raided personal information on adults *and* children.

31/

Do you want Trump's FCC chair Brendan Carr setting content moderation policies for your internet connection? The guy who wants to pull TV and radio stations' broadcast licenses if they criticize Trump and Israel's catastrophic Iran war?

https://www.techdirt.com/2026/03/17/brendan-carr-pretends-to-be-tough-demands-broadcasters-support-disastrous-war/

Do you want your local ISP being run by your mayor? I mean, sure, there are *some* reasonable mayors out there, but imagine if your ISP was managed by Eric Adams, Boris Johnson...or *Rob Ford*:

https://www.patreon.com/posts/rob-ford-part-1-111985831

32/

Saying that broadband should be run "like a utility," raises more questions than it answers. I, too, want broadband run "like a utility," but that doesn't mean that I want the whole show to be provided solely by my federal or municipal government. A "utility" model for broadband *should* mean running conduit to every home in town, with point-to-point connections that deliver broadband via a municipally owned network - but not *just* that.

33/

The municipal network should *also* offer "essential facilities sharing" in two forms: first, they should allow anyone to set up an ISP by renting shelf-space in the municipal data-center and installing their own switches that can provide internet to anyone in town. This would let large *and* small companies set up ISPs, as well as co-ops and nonprofits, or even tinkerers wanting to provide access to a group of friends.

34/

Beyond that, the city should rent space in the conduit itself, to support point-to-point links beyond those offered by the city - for example, between a university campus and an offsite supercomputing center, or two buildings owned by the same company, or even as a parallel set of fiber connections run by someone who's fed up with getting their internet service from Eric Adams.

35/

This is a "pluralized" utility model: one that involves the city in providing infrastructure at several layers, as well as a "public option" - but which doesn't allow a city that's in thrall to Moms For Liberty to decide what you can say on the internet.

This principle generalizes beyond internet provision, too.

36/

Many people have observed that social media, with its strong "network effects" (meaning its value increases as more people use it), could be a "natural monopoly" and want a social media "utility." I can see the reasoning there, but if there's one thing we've learned from zuckermuskian legacy social media, it's that centralized control over speech forums is a moral hazard and an attractive nuisance.

37/

It's a political prize beyond measure, and it attracts all sorts of skullduggerous bids to suborn it and harness it to some political faction.

But there's a pluralized utility model for social media, too, thanks to modern, federated social media systems like Mastodon and Bluesky. These are open platforms that can support multiple, interconnected servers that all talk to one another.

38/

Unlike, say, Twitter, where you can only talk to other Twitter users, federated social media allows you to talk with anyone on *any* server, provided they want to talk with you.

As with fiber, a "utility" model for federated social media would feature public intervention at multiple layers of the system. Governments could (should!) run their own servers, providing the canonical source of government information.

39/

They can also provide turnkey cloud services for people who want to start their own services - and they can spin out the code that goes into these services into free/open source projects that others can use (and contribute to). Governments could support people who are trying to migrate off of legacy social media (for example, through library workshops and helplines), and pay to label and tag media (for example, media that is compliant with the public education curriculum).

40/

Governments could also offer public servers where you could sign up to get online - and because federated social media makes it easy to move your account from one server to another, it would be easy to move from that server to one run by a nonprofit, a co-op or a business:

https://pluralistic.net/2025/06/25/eurostack/#viktor-orbans-isp

41/