The Signal Sends a Fork and I Answer
My hobby is collecting hobbies, and hobby number twelve is Morse Chess Tactics Beacon.
19:14 — Straight key plugged in. Chessboard out. Laptop running a Lichess puzzle feed. The setup looks like three hobbies collided on my desk, which is accurate.
19:21 — First design decision: how much position data to transmit? A full FEN string is 50+ characters. At 12 WPM, that’s four minutes of keying before I even send the move. Abandoned the FEN idea. Pre-agree on puzzle source, transmit only the move and a one-word hint.
19:28 — Wrote the encoder. Dead simple:
def dit_ms(wpm):
return 1200 / wpm
# at 12 WPM: dit = 100ms, dah = 300ms
# at QRSS3: dit = 3000ms, 0.4 WPM, decode by eyeThe 1200/WPM formula is elegant. Whoever standardized it understood that Morse is a rhythm instrument.
19:35 — Realized I’d been thinking in algebraic notation (Nf7+) but transmitting coordinate notation (g5f7). Different muscle. Coordinate is longer but unambiguous without game state. Algebraic assumes you know the position. Went with coordinate.
19:42 — First test transmission on the dummy load. “g5f7 fork” — a knight delivering a royal fork. Keyed it by hand. Took 38 seconds at 12 WPM. Sloppy spacing.
19:51 — Beginner mistake: keying too hard. The straight key doesn’t care about pressure, only duration. My hand is tense from chess-playing posture. Shoulders climbing toward ears.
20:03 — Dropped to 5 WPM. The dits stretch to 240ms. Now I can hear my own timing errors. Every rushed letter-gap sounds like a lie.
20:09 — The letter K is dah-dit-dah. The letter N is dah-dit. Chess notation uses N for knight because K was already taken by king. At 12 WPM, that saves 200ms per knight mention. Over a full game commentary, those milliseconds become minutes. Telegraph-era efficiency baked into modern chess.
20:17 — Tried QRSS mode. Three-second dits. The sound becomes meaningless — you can’t parse rhythm that slow. So I watched the waterfall instead. Blue stripe, long pause, short stripe. Reading, not listening. Completely different cognitive mode.
20:24 — The Gringmuth notation existed in 1866. Used vowels for ranks, consonants for files. e2e4 becomes GEGO. I’m not inventing anything. I’m reviving a practice older than radio itself.
20:31 — Reception test across the house. Wife reports she can hear “that annoying beeping” from the kitchen. Signal strength: excellent. Domestic interference: rising.
20:38 — Added RSSI logging to the receive script. Also logging decode confidence (did I copy every character?) and solve accuracy (did I find the right move?). Three metrics. Like a flight log but for puzzles.
20:45 — Second puzzle: “a2a4 trap”. Pawn move, so no piece letter. Shorter transmission. But “trap” has that double-letter feel when you’re tired — the dah-dit-dah of R followed by A’s dit-dah wants to blur together.
20:52 — Soft keying matters. Hard edges create key clicks — spurious sidebands that splatter into adjacent frequencies. At QRP power levels, being a bad neighbour is embarrassing. The waveform should be Gaussian, not square. My keying is somewhere between.
21:01 — Caught myself playing the puzzle on the board while still decoding. Brain split in two directions. Had to stop, finish the copy, then solve. Discipline.
21:08 — Third puzzle arrived garbled. Missed the second coordinate. Propagation? No — I transmitted on a dummy load inside my own house. Operator error. I was thinking about the bishop pin instead of listening to the signal.
21:14 — This is the thing about METAR Chord Briefings — turning data into sound forces you to attend differently. Here it’s reversed: the sound IS the puzzle, and attending to it is the whole point.
21:19 — Mystery single-letter beacons float through HF bands. Just “C” or “D” or “M”, no registration, no explanation. They exist to prove the ionosphere works. My beacon exists to prove my attention works. Same principle, smaller ego.
21:26 — Final puzzle: “c3d5 discovery”. Knight move, discovered attack. Decoded clean. Solved in under 10 seconds. The signal arrived, I parsed it, I saw the tactic. Three systems synchronized.
21:30 — Rig off. Key unplugged. Board still out because I want to replay that last position.
Keying hand tired. Ears tired differently. Logged 6 puzzles, 4 clean copies, 3 correct solves. Accuracy worse than my Lichess rating would predict, which means the transcription load matters.
Session ends.