The editorial argues the watch community discusses this but the engineering community misses it: mainspring = power rail with load regulation via fusee, balance wheel + hairspring = reference oscillator (4-5 Hz crystal analog), and escapement = gating logic that doubles as clock gate and regulator. The mapping is exact, not metaphorical.
Ciechanowski's entire body of work (Engine, GPS, Gears, Bicycle, Light & Shadows, now Watch) is an implicit argument that manipulable 3D illustrations beat text-plus-diagrams for teaching mechanical systems. Letting readers drag the pallet fork and pause the impulse jewel mid-tick converts passive reading into a self-installing mental model.
By submitting Ciechanowski's piece to HN where it hit 652 points, the submitter endorses the format itself — these posts reliably top the front page without newsletters, ads, or promotion because the interactive medium does the persuasion work that static explainers cannot.
The editorial highlights that Ciechanowski has no newsletter funnel, no team, no ads, and no sponsor block, yet every post lands on the HN front page within hours. This is offered as evidence that quality and respect for the reader ('prose that assumes you can think') still beats the growth-hacking playbook.
Bartosz Ciechanowski published Mechanical Watch — another entry in his series of single-author, fully interactive explainers (Internal Combustion Engine, GPS, Gears, Bicycle, Light & Shadows). The watch piece hit 652 points on Hacker News, the latest in a streak where every Ciechanowski post lands on the front page within hours of going live. No newsletter funnel, no team, no ads, no sponsor block. Just draggable, rotatable 3D illustrations and prose that assumes you can think.
The content covers the standard mechanical-watch anatomy — mainspring, going train, escapement, balance wheel, hairspring — but the medium is the point. Static watch explainers describe the escapement and the reader nods and forgets. Ciechanowski's version lets you drag the pallet fork, pause the impulse jewel mid-tick, and watch the balance wheel's hairspring breathe. The mental model installs itself.
Here's the thing the watch community talks about and the engineering community usually misses: a mechanical watch is a complete clock distribution network, and every primitive on the silicon side has an exact brass equivalent. Map it:
- Mainspring = the power rail. Stored energy with a discharge curve. Watchmakers added the *fusee* — a variable-radius pulley — to compensate for the rail's voltage droop as the spring unwinds. That's load regulation. - Balance wheel + hairspring = the reference oscillator. A typical movement runs at 4 Hz (28,800 vibrations per hour); high-beat calibers run at 5 Hz. This is your crystal. The hairspring's elasticity sets the resonant frequency the way a quartz cut sets a crystal's. - Escapement = the gating logic. The pallet fork releases the escape wheel one tooth per half-cycle of the balance wheel. That's a clock edge, in a literal mechanical sense. It's also the only place energy moves from the spring to the oscillator, which means it's a regulator and a clock gate simultaneously. - Going train = the PLL divider chain. Gear ratios divide 4 Hz down by deterministic factors until the seconds hand runs at 1 Hz, the minute hand at 1/60 Hz, and the hour hand at 1/43,200 Hz. Every transition is a clock-domain crossing handled by rigid mechanical coupling — the brass equivalent of synchronous logic.
Christiaan Huygens shipped the first practical version of this architecture in 1657. The clock-domain crossing problem — moving state safely between domains running at different frequencies — was solved mechanically three centuries before anyone wrote a Verilog synchronizer. Modern digital designers reach for two-flop synchronizers and gray-coded FIFOs; the watch reaches for a wheel with the right number of teeth. The constraint is the same: do not let metastability eat your state.
Why then does a quartz watch hit ±15 sec/month while a high-grade mechanical hits ±5 sec/day — three orders of magnitude worse? The answer is oscillator Q-factor, the same parameter that picks your CPU's reference crystal. A balance wheel has a Q of maybe 100–300; a 32.768 kHz tuning-fork crystal has a Q north of 100,000. Temperature, position, lubrication state, and shock all perturb the balance wheel's effective spring constant. That's the analog noise floor that drove the industry to crystals in the 1970s — not because quartz was cheaper (it wasn't, initially), but because Q-factor wins.
The community reaction on HN captured the parallel: 'this is the best explanation of a frequency divider I've ever read, and it isn't about frequency dividers.' Someone else noted that the escapement's impulse-jewel geometry is the same problem as setup/hold margin on a flip-flop — there's a window where the input must be stable for the gate to fire cleanly, and outside that window you get jitter.
If you're doing hardware or firmware: this is the explainer to send a junior engineer who can't visualize what 'clock tree' means. The watch makes the abstract concrete — every term in a digital clocking textbook has a brass-and-jewel cognate that you can hold in your hand and pause mid-tick. That's pedagogically priceless, and it costs nothing.
If you're doing distributed systems: the watch is a useful sanity check on assumptions about synchronization. The fusee — the load-regulation pulley — exists because watchmakers learned the hard way that a 'constant' energy source isn't constant under real load. Distributed-systems folks learn the same lesson the same way, in production, every Black Friday. The watch's answer was a mechanical compensator; the cloud's answer is rate-limiters and circuit breakers. Same problem class.
For interviewing: if you're hiring for embedded, RTL, or systems roles, the watch piece is a better filter than most leetcode. Ask a candidate to map mainspring → escapement → going train onto crystal → PLL → divider tree. If they see the analogy in under a minute, they understand clocking at the level that matters. If they don't, no amount of textbook recall will save them at 2am when the PLL won't lock.
Ciechanowski's catalog has quietly become the canonical 'physics-to-engineering' reference for the field — better than most undergraduate textbooks because it respects the reader's intelligence and refuses to use the words 'simply' or 'just.' The watch piece's deeper lesson is that the hard problems in digital systems aren't new. They were solved, badly at first and then well, by people working in brass and steel because that's all they had. We made the primitives faster and smaller. We did not make them different.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.