The Open Source Internet Is Here: A Thousand-Mile Bridge Attempt with Reticulum and HF Radio

Phones stay linked as long as the bill is paid; this project tries to remove the bill from the equation. Reticulum (a cryptography-based, open-source network stack that survives high latency and very low bandwidth) is the core. Background: telecom, energy and storage lean toward oligopoly, handling data opaquely to maximize profit — recent DePIN research points to community infrastructure as the counterweight. Think in three steps: 1) an edge node talks LoRa locally, 2) a gateway translates it into an HF frame, 3) the sky carries it to a distant community. Like a bike-share — no single owner, everyone carries one link. For example, a person who raises a 915 MHz antenna by two meters often sees three new neighbors appear.

Why Gatekeepers Exist

Gatekeeping is not an accident but a design choice. Today most communication passes through gates that decide what is allowed, keep records and charge for the privilege. Background: commercial telecom and mesh networks are not two answers to the same question; their infrastructure assumptions are opposite. Meshtastic (every node repeats — managed flooding) versus MeshCore (clients never repeat — flood once to discover a route) captures the split: one is flash-and-go, the other preserves airtime at scale. Transport-agnostic (not caring which bearer carries the bytes — Reticulum's core promise) makes the gate irrelevant. Like replacing a toll highway with a grid of side streets that keep their own signs.

Field proof was gathered on the road. Driving from Florida to New Hampshire with a magnetic 915 MHz whip and a small Neo One listening on the public Long Fast channel, the receiver logged nodes along the route; some cities such as Concord showed dense activity and more than 450 nodes appeared on a single protocol. That is what curiosity scaled looks like. Open source (design, fabrication and tooling all open) matters here: hardware is modified in the garage, antennas trimmed, software forked. For example, tuning a roof antenna can summon three new neighbors in the same block.

Community grows that way: a question to understand, a few friends, small steps. The study then asks how that accumulation survives a gap of hundreds of miles. Background: Reticulum starts with a default config and joins local peers; if one peer is a Transport Instance , it can carry you to a wider net. So a neighborhood mesh, with the right gateway, can be grafted onto the far side of a continent. Like tributaries feeding the same sea. For example, a node in Ankara and one in Hamburg can speak the same stack and let HF quietly bridge the gap.

The Limits of Three Established Paths

Three established long-range paths exist. One, towers : many are needed and they only pencil out for large carriers. Two, satellite (Starlink, Garmin, Iridium): strong off-grid but hardware is proprietary, monthly plans apply and each message pays a toll. An earlier test wired an Iridium modem into Reticulum and sent a message under open sky — it worked, but each exchange was billed. Three, community mesh (Meshtastic/MeshCore): real, local and owned, yet serious distance needs enough volunteers in between. Their shared limit is ownership: they lean on someone else's backbone.

Economics sharpens the limit. Tower builds demand capital; satellite subscriptions are recurring; mesh depends on volunteer density. Research on decentralized infrastructure notes that telecom scale advantages turn into coverage gaps in rural areas. Off-grid (operating away from urban power and backhaul) makes satellite look attractive, yet contracts and proprietary gateways recreate gatekeepers. Like lighting a room with a rented generator — useful in the dark, but someone else writes the bill. For example, a pop-up mesh at a festival vanishes when the crowd leaves.

A fourth path is proposed: privately owned networks that link across hundreds or thousands of miles without classic backhaul. Two conditions are needed: one, cross on a path you own; two, do it within the rules. The amateur band (licensed amateur frequencies) enters here — it requires public identification by call sign. Background: in 2023 the FCC removed the symbol-rate cap for data emissions and moved to a bandwidth limit, adding flexibility for modern HF data. That opened room for newer modulations on HF. Like changing a speed limit from vehicle count to road width.

Reticulum's difference becomes clear. The stack is medium-agnostic — LoRa, Wi-Fi, Ethernet, internet or a custom HF link are all just another path in the network. Instead of building a new isolated island, you can stitch existing community nets together. RNode (hardware born from the search for affordable, general-purpose long-range digital radios) embodies that idea: an open alternative to expensive, hard-to-source and power-hungry solutions. Like a rail switch that joins different gauges at one platform.

Ham radio is misunderstood as a retirement hobby or weather club; what it can do a Wi-Fi box cannot. HF (high frequency — 3 to 30 MHz that can travel far via ionospheric reflection) operates roughly 100 times lower than typical Wi-Fi and achieves skywave (reflection from ionized layers that drops hundreds of miles away) beyond the horizon. How does it work? 1) the transmitter creates a frame on HF, 2) the ionosphere reflects it as ionization shifts by day and night, 3) a distant receiver listens on the same band while it is open. Like skipping a stone — angle and timing decide if it reaches the far shore. For example, 14 MHz often opens near dusk while 7 MHz shines at the gray line.

Data on radio is not new. APRS (small VHF packets for location and weather — great locally) and FT8 (an HF digital mode that can be heard even in weak conditions) already exist. But APRS lives on VHF and leans on local repeats; FT8 is built for short exchanges, not networking. The aim here is not to write yet another HF messenger; it is to design a custom packet that carries a Reticulum frame over HF. Then HF becomes not a new island but a quiet bridge between islands that already exist. For example, a person using LoRa in one town can reach another town's very different link without touching HF at all.

Heart of the Bridge: Custom Packet and AirMay Lite 2

The custom packet must carry what the bridge needs: a Reticulum envelope (encrypted, routed frame), gateway identity and call-sign identification in the clear. APRS is not built for that — it is for location and short status. The bridge packet must preserve the network envelope intact. Background: Reticulum is designed for high latency and tiny bandwidth, so even one packet carries value. Like an envelope inside an envelope — the inner stays encrypted while the outer is readable on air. For example, a 40-byte Reticulum announcement can fit in a 1–2 second HF emission.

The key hardware is AirMay Lite 2 (a Hermes-Lite 2–based, 0 to 38.4 MHz open-source software radio built around a computer, not a microphone). Traditional HF rigs are built for a person with a mic; this one is built for software: code generates the exact frame, puts it on air and decodes it. The Hermes-Lite lineage, with more than 500 units built, signals scale. Alongside lie boards for Bluetooth, Wi-Fi, LoRa, Halow and Iridium, but they either stay local or lean on someone else's backbone. Analogy: a typewriter versus a computer — one moves keys, the other moves files.

The build log ran on a three-week ship promise. Learning to read the waterfall, listen for APRS traffic, catch a real FT8 decode, watch a band close and reopen, and verify antenna tuning became the curriculum. The ground plane remained confusing, but the rest sharpened: touching coax can shift center frequency, antenna length and placement decide match. For example, if touching the feedline moves the peak, revisit ground and feedpoint. These small details are invisible in a lab yet break a link in the field.

Proof That Skipped Off the Sky — and the Gap Left

Remote proof arrived first as listening. The receiver setup decoded a digital transmission from Bogota, about 1,600 miles away, proving we can push beyond the horizon. Skywave here is not just physics but timing: right frequency, right hour and right antenna must coincide for a packet to land on the far shore. Like catching wind in a sail — the wind is always there, but angle opens the path. For example, 7 MHz often opens at sunset while 14 MHz works better midday.

The bridge itself was shown in four steps: 1) an ordinary Reticulum node creates a message, 2) a licensed gateway converts it into a public, call-sign-identified HF frame and transmits, 3) a distant receiver decodes and rebuilds the payload, 4) it is delivered to an ordinary Reticulum node on the other side. In this chain few people ever touch HF; most communities stay local. In other words, HF becomes the quiet wire that holds two local meshes together.

The missing link stayed logistical. A real skywave test needs someone hundreds of miles away listening on the same frequency with a decoder while the band is open; a $37 receiver, antenna and computer could be enough — no second AirMay or even a transmit license required — but place and time must align. The three-week clock ran out on a Sunday; instead of waiting for a perfect result, the log shipped the good, the bad and the ugly. For example, the idea of a Reticulum announcement born in Germany bouncing off the sky into North America is now an open call for a partner.

An honest ledger noted the irony: while trying to cancel the middleman, subscriptions to fiber, Verizon, Starlink and Iridium piled up — the would-be quitter became the best customer. That friction is the lived cost of decentralization. Open source makes cheap hardware reachable, yet spectrum, licensing and weather remain hard limits. Like baking your own bread while still renting someone else's oven — the first loaf still depends on the outside.

The vision is simple: a free network we actually own, built with tools by the people, for the people. If a Reticulum node in Germany can bounce off the sky and land in a North American Reticulum net, that changes the map. Hardware is ready, code is open, knowledge shared; now someone on the far side of that map must hold the other half of the test. When an experiment is shared before it looks perfect, the community itself becomes the test — scaling not by one person's stubbornness but by a culture of testing together.