Suppose you and a hundred friends are setting up a colony on Mars. Other colonists are handling food, water, and shelter — the big stuff. All you need to do is set them up with working cellphones.

Of course, there's no AT&T on Mars, so you'll have to set up your own cell towers, but probably not as many as you'd think. You can stick to a Wi-Fi network on the main base with just a few towers for outer regions, putting up a new one every time a new region is explored. Since you own the towers, you can also skip charming earthly quirks like texting caps and tethering charges, just instituting basic fees to cover access to the network at large. (It's a market-based colony.) Anyone calling within the station can do it entirely on the Wi-Fi network.

You also won’t be privileging voice calls, so you don’t need a distinction between your phone number, your email, and your screen names on various services. You can rely on just a single data channel and a single handle, which might also cut down on overall usage since you're not being nudged towards bandwidth-intensive services like voice calling anymore. And with the bulk of the traffic flowing over internal Wi-Fi, there will be a lot less data to be moved overall.

The point being, your Martian cellphone carrier could very well end up better, cheaper, and more sensibly organized than the one you're dealing with on Earth.

This, roughly, is the starting point for Signal, a new service currently in public alpha. From the outside, it looks like an ambitious kind of phone-service-meets-mesh-network. Instead of working by phone numbers or emails, you get a new Signal screen name prefaced by a ^ symbol, the equivalent of Twitter's @. Send a message to ^russellbrandom (or russell.theverge.com if you want to tie it to a domain), and the web’s DNS infrastructure will route it to my Signal server, which can forward it to my phone, my tablet, my computer, or wherever else I've programmed messages to go. Instead of separating out texts, calls, and third-party app messages, they all end up in the same place.

More importantly, most of it’s free. The main offering is free voice calling to anywhere in the world, like Skype but over traditional phone lines. In the coming months, Signal will roll out a dialer replacement for Android, along with an app for iOS and a browser plugin for the web. If you want a traditional data plan or a connection to a traditional telco network, you can buy it from SignalThis, a third-party company that’s launching as a competing service inside the Signal network, betting that customers will pay more for a more extensive network. It won’t be free, but it may still end up cheaper than a standard phone bill.

That might seem simple enough, but behind the scenes, Signal is doing some pretty weird stuff. The company is signed up as a non-profit virtual operator with Sprint, so if you want to use it as a simple pay-as-you-go phone, you can. (Signal is hoping for deals with the other major carriers as the platform expands.) But if you want to move away from the carrier, shifting more and more of your coverage to local Wi-Fi networks, Signal makes it much easier than a conventional carrier. You can also cut Signal's hardware out of the picture completely, hosting your own Signal server and relying on your own Wi-Fi-powered cell tower for connectivity. Signal has designed the towers to talk to each other, so you could eventually call a friend without leaving Signal’s infrastructure at all. "It will be exciting to see the peer-to-peer usage between users hosting their own signals," says founder Andy Myers. "That's really the goal of all this." It’s the Martian scenario or, as Myers puts it, a cell network for the zombie apocalypse.

That restructuring also means you can pull off tricks that would be impossible on a conventional phone network, like issuing a universal do-not-disturb or limiting certain callers to certain devices. Your texts and calls are all managed by the same central Signal server, so it can route them however it wants. That gives you total control over how and when you get notifications. If you want a total call blackout between midnight and 5AM, all you have to do is tell the server.

The most immediate problem is adoption. Myers’ plan will work great once he’s got 100,000 users and a Signal tower on every block, but reaching that critical mass is notoriously difficult, as hundreds of failed apps and social networks can attest. At the moment, Myers is banking on ambitious gearheads and open standards to fill the gap. Anyone can build an app or a business on top of Signal’s architecture, and the organization is currently trying to enlist as many partners as it can before it launches its beta in May. Ultimately, Myers sees Signal growing into an open protocol, a kind of HTTP for the mobile age, but before it gets there, he’s going to need a lot of consumers on his side.

The bigger problem is, well, the phone industry. Myers says he expects carriers to welcome Signal with open arms, but so far, Sprint is the only one that’s taken the bait. "Sprint is really the most innovative carrier out there in many ways," Myers says — but not coincidentally, they’re also one of the smallest of the Big Five. Signal’s offering a better way to move data, but it’s not at all clear the carriers will be interested, and a small nonprofit like Signal won’t have enough clout to make them change their minds. It’s an exciting idea, but the telecoms have been successfully fighting off this kind of disruption for decades now.

Which puts us back at the beginning, building cell towers for Mars. Signal’s grand vision sounds incredible, but Earth might not be ready for it any time soon.