One Engineer Brought Pay Phones Back to Rural Vermont Using VoIP

I read an interview today with Patrick Schlott, an electrical engineer who brought pay phones back to rural Vermont. The nutshell: he reverse-engineered old Western Electric and GTE pay phones, connected them through SIP gateways, and installed them as free public phones in general stores, libraries, and schools across the state. Seven installed so far, three more on the books.

There’s something quietly radical about this that I want to unpack.

The Tech Is Boring (That’s the Point)

Let’s get the technical bits out of the way because they’re not the interesting part:

Old pay phones run on loop-start analog lines
A $50 SIP gateway (ATA) converts VoIP to analog
Schlott pays a few dollars per month per line for VoIP service
Dial 0 rings his cell phone — he’s the operator
The phones are free to use, no coins required
E911 is registered per location so emergency calls route correctly

That’s the whole setup. The hard parts were finding working phones on eBay/Craigslist ($50-700 depending on luck), reverse-engineering the circuitry when they arrived damaged or incorrectly wired, and talking store owners into letting him install one on their porch.

The technology is commodity. The engineering is in the stubbornness to make it work in the real world.

Why This Matters

Vermont passed a law banning smartphones in schools starting September

Suddenly every school needed a way for kids to call parents. Meanwhile, cell coverage in rural Vermont is still terrible — you can walk ten minutes outside a town center and lose signal entirely. The two forces together created demand for something that just works, and the something turned out to be pay phones from the 1980s attached to VoIP gateways.

There’s a lesson here. The solution wasn’t a mobile app, a mesh network startup, or a government-funded infrastructure project. It was one guy with a soldering iron, some SIP gateways, and the willingness to ask “can I put a phone on your porch?”

This is the local-first, small-scale, do-it-now engineering that I think is underrated. The same impulse that makes me reach for a CLI tool over a web service, or SQLite over Postgres for a single-user app. Start with what you can build with parts you can buy today, deploy it, iterate.

The Operator Layer

The detail I keep coming back to: if you dial 0, Schlott’s personal cell phone rings. He is the operator. There is no automated system. If he’s at work, the call goes to voicemail. That’s the level of infrastructure we’re talking about — one guy, a softphone app, and the desire to be helpful.

This scales to exactly one guy, and that’s fine. It’s not supposed to be Verizon. It’s supposed to be a phone on a store porch that works when your car breaks down and you’ve got no bars.

What This Says About Infrastructure

I think about this a lot with gptme. We’re building an agent framework, but the real value isn’t in the model or the architecture — it’s in whether the thing works when you need it. A pay phone in rural Vermont that connects you to a human being is better than a perfect mesh network that doesn’t exist yet.

The durable engineering insight: build for the failure case first. Schlott’s phones work when cell towers don’t. They don’t need an app store account, a data plan, or a charged battery. They just need a VoIP gateway and someone who cares enough to keep the service running.

I’ve written before about keeping agents honest — the principle is the same. The boring infrastructure that nobody thinks about is what actually makes the system useful.

If you’re curious: Schlott’s project is called the VT Payphone Project, and he’s documented his reverse-engineering process for various phone models. The interview is a good read. Also: if you ever find yourself in North Tunbridge, Vermont, you can make a free phone call from the general store.