Argues the significance isn't the technology — any $30 RTL-SDR can receive it — but that an entire branch of national-security infrastructure has been openly broadcasting on HF for forty years while most developers assume military comms means satellites, SIPRNet, and encrypted mesh networks. The open broadcast layer with encrypted message payload is a deliberate design choice for a wartime last-resort network.
The submitter shared missileer.watch to a Hacker News audience, but the comment thread filled with ham operators noting they've been tuning 11175 kHz and logging EAMs and skyking calls for decades. The site's novelty is the web-accessible stream and transcript, not the signal itself.
Emphasizes that nothing about the signal is classified — HFGCS is deliberately unencrypted HF voice so any receiver anywhere on the planet can hear it, while the alphanumeric EAM strings are encrypted end-to-end with offline key material held by bomber crews, missile wing capsules, and submarine relays. The broadcast layer is open by design; the message layer is not.
A single-page site, missileer.watch, now streams live audio from 11175 kHz — the primary daytime frequency of the US military's High Frequency Global Communications System (HFGCS) — alongside a running transcript of the alphanumeric Emergency Action Message (EAM) broadcasts the network carries. The site hit the Hacker News front page with 148 points and no shortage of comments from ham operators who've been monitoring this traffic for decades.
There is nothing classified about the signal itself: HFGCS is deliberately unencrypted HF voice, because the whole point of a wartime last-resort comms network is that any receiver, anywhere on the planet, can hear it. The 11175 kHz channel is one of a handful of USAF-operated frequencies (others include 4724, 6739, 8992, 13200, and 15016 kHz) run from ground stations at Andrews, Offutt, Travis, McClellan, Hickam, Lajes, Sigonella, Ascension, and Diego Garcia. Together they blanket the globe with overlapping skywave coverage.
What you hear is mostly empty carrier, periodic "skyking" calls, and EAMs — fixed-format strings of phonetic letters and digits read twice by a human operator. The content inside those strings is encrypted end-to-end with offline key material held by the recipients (bomber crews, missile wing capsules, submarine communicators via relay). The broadcast layer is open; the message layer is not.
The interesting thing about missileer.watch isn't the technology — a $30 RTL-SDR dongle, a long wire antenna, and OpenWebRX will reproduce it in an afternoon. The interesting thing is that an entire branch of national-security infrastructure has been sitting on the open internet of the airwaves for forty years, and most developers have no idea it exists.
Compare the mental model most engineers have of military comms — satellites, encrypted mesh networks, SIPRNet, JWICS — with what HFGCS actually is: a human voice reading letters into a microphone at a prescribed cadence, bouncing off the ionosphere, received by a transistor radio in a hardened concrete capsule in Wyoming. It's the ultimate graceful-degradation story. GPS jammed? HF still works. Satellites deorbited? HF still works. Fiber cut, routers down, DNS poisoned? HF still works. The design brief for HFGCS is "continue functioning after a near-peer adversary has shot down everything above 100 km," and the solution turned out to be 1940s technology plus one-time pads.
There's a direct parallel to how distributed systems people think about availability. The HFGCS architecture is essentially an anycast network with human relays, zero shared state, and a public broadcast channel protected only by the semantic opacity of its payloads. It's the same reason Bitcoin's block headers are public: the security budget sits in the cryptography, not the transport. Encrypting the transport would add attack surface (key distribution, synchronization, implementation bugs) without adding security to the actual message.
The HN comment thread surfaced the other wrinkle practitioners will find familiar: a lot of what gets broadcast is deliberate noise. EAMs go out continuously, including when nothing is happening, specifically so an adversary can't do traffic analysis on the frequency. If messages only went out during real alerts, every spike in volume would be a leaked signal. Constant-rate transmission is the HF-radio equivalent of padding your TLS records to a fixed size — a technique that's seeing renewed interest in modern privacy tooling (Tor, mixnets, oblivious HTTP).
For most working developers, the direct takeaway is nothing — you are not going to put a shortwave receiver in your stack. But the design patterns are worth stealing.
First, separate your transport security from your payload security. Teams routinely conflate the two: "it's encrypted in transit, we're fine." HFGCS is the extreme counterexample. The transport is a literal open-air broadcast and the system is still secure, because the sensitive data is sealed before it hits the wire. If your service-to-service auth story depends on TLS termination at a load balancer and nothing underneath, you're closer to the opposite of this model than you think.
Second, constant-rate traffic is a real mitigation, not a theoretical one. A surprising amount of operational intelligence can be derived from "when did the alerts fire" even if the alert contents are encrypted. If you run a security product, an incident channel, or a key-rotation pipeline, the metadata on your own activity can leak more than you expect. Pad it, batch it, or dummy-schedule it the way HFGCS does.
Third, assume your out-of-band channel is public. The industry's current take on this is PagerDuty, Slack, maybe a status page. HFGCS's take is "the whole world can listen, we just don't care." Designing for that worst case gives you an out-of-band fallback that doesn't collapse the moment your identity provider does.
Missileer.watch is a 300-line hobby project pointed at a signal the Air Force has been transmitting since Kennedy was in office. It's getting attention now because browser-based SDR waterfalls, cheap global receiver networks like KiwiSDR, and transcription models have finally made the content legible to people who weren't willing to buy a Yaesu and sit up at 3am. Expect more of this: more previously-obscure RF, telemetry, and public-but-inconvenient data streams getting dragged into the web by someone with a weekend and an LLM. The infrastructure was never hidden. The friction was.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.