What Watchmaking Taught Me About Writing a JHA That Actually Holds
- Chris Fredette
- 2 days ago
- 5 min read
I recently got into mechanical watches. Not smart watches, not quartz. The old kind, run by a wound spring and a few hundred tiny parts, that has been keeping time the same way for three hundred years. I fell down the rabbit hole the way most people do, chasing the look, and came out the other side thinking about job sites.
Here is the thing that got me. A mechanical watch has hundreds of parts, and every one of them is trying to fail. Friction wears the pivots. Temperature swings the mainspring's tension. Gravity drags on the balance wheel differently depending on how the watch is sitting on your wrist or your nightstand. The whole discipline of watchmaking is designing a system that stays accurate not because failure is impossible, but because failure is anticipated and compensated for at every single point.
Job sites work the same way. I think we sometimes forget it.
A Control That Only Works on a Perfect Day Is a Wish
We write a job hazard analysis like the goal is a perfect plan that nothing disturbs. Clean sequence, full crew, materials on site, weather cooperating. On paper it all works. Then the real day shows up.
The wind picks up. The crew is short two people because someone called in and someone else got pulled to another scope. The material that was supposed to be there Monday shows up Wednesday afternoon, and now everyone is compressing three days of work into one and a half to hold the schedule. That is the day your controls actually get tested. Not the calm one.
A control that only works on a calm, fully staffed, on schedule day is not a control. It is a wish. The whole point of watchmaking is that the watch keeps time on the day it gets bumped, dropped in a bag, left in a hot truck, and worn by someone who forgot to wind it until noon. A movement designed for a laboratory is useless on a wrist. A JHA designed for a perfect shift is useless on a job.
By the Numbers
Construction accounted for 1,075 worker deaths in 2023 according to the Bureau of Labor Statistics, more than any other industry sector.
Falls, struck by, caught in or between, and electrocution, the Focus Four, made up the majority of those construction fatalities year after year.
Fall protection, general requirements, 1926.501, has been OSHA's most cited standard for more than a decade running.
A large share of serious incidents trace back not to an unknown hazard, but to a control that existed on paper and did not hold under real conditions.
The Good Stuff Is the Redundancy
Open up a good movement and the thing that stands out is not elegance. It is defense. There is a device called a shock protection system, Incabloc being the famous one, that lets the delicate balance staff jump in its setting when the watch takes a hit, instead of snapping. The escapement releases energy in tiny controlled bites so the mainspring's power never hits the hands all at once. Better movements add a balance spring shaped to cancel out the errors that temperature and position would otherwise introduce. None of that makes the watch prettier. All of it keeps the watch honest when the world stops cooperating.
That is what a real control set looks like on a site. The good stuff is the redundancy. The second point of attachment so a single failure does not drop a person. The independent verification, someone other than the person doing the work confirming the energy is actually isolated. The stop work authority that does not depend on any one person being brave that day, because it is built into how the crew operates rather than resting on one guy's willingness to be unpopular.
A single control is a single point of failure. Watchmakers learned centuries ago that you do not bet a person's day on one part behaving. You assume the part will fail and you build the system so the failure does not reach the thing that matters. Swap the word part for control and you have the whole philosophy of a JHA that actually holds.
Every Complication Adds a Way to Fail
In watch language, a complication is any function beyond telling time. A date window, a chronograph, a moon phase. Every complication a watchmaker adds also adds parts, and every part is one more thing that can bind, wear, or break. The most reliable watches in the world are often the simplest ones, because there is less to go wrong. Complexity is not free. You pay for every feature in failure points.
Scope works the same way. Every time a job adds a complication, a second crane on the pick, a night shift running alongside days, a new subcontractor showing up with their own equipment and their own habits, it adds failure points to the whole system. That does not mean don't do the work. It means the JHA has to grow with the complexity instead of pretending the extra parts are free. When a job gets more complicated and the hazard analysis stays the same length it was on day one, the gap between the two is exactly where people get hurt.
A Movement Gets Serviced. So Should Your Controls.
Even a perfect movement needs service. The oils dry out. Wear accumulates in places you cannot see from the outside. A watch that runs beautifully for five years and never gets opened is not reliable, it is running on borrowed time, and the day it stops it will stop without warning. Watchmakers do not wait for the failure. They pull it apart on a schedule, clean it, and check the parts that have been quietly wearing the whole time.
A JHA written on day one and never revisited is the same watch nobody opens. The conditions it was built for have drifted. The crew rotated, the sequence changed, a control that used to work got quietly worked around because it added four minutes and everyone was in a hurry. The plan looks fine from the outside and is wearing out on the inside. The fix is not a better one time document. It is treating the hazard analysis as a living thing that gets opened, walked, and re-checked against the site as it actually is right now.
Key Takeaways
Design controls for the bad day, not the perfect one. If a control only holds when the crew is full and the schedule is calm, it is a wish, not a control.
Build in redundancy. A second point of attachment, independent verification, and stop work authority that does not rest on one person being brave are the equivalent of shock protection in a movement.
Grow the JHA with the complexity. Every added complication, crane, shift, or sub, adds failure points, and the hazard analysis has to grow to match instead of staying stuck on day one.
Service your controls. A hazard analysis that never gets reopened is wearing out on the inside. Walk it and re-check it against the site as it actually is now.
Design your controls the way a watchmaker builds a movement. Assume everything is trying to fail, build for the friction instead of the ideal, and hold the time anyway.
Comments