In the ship model shop at Naval Information Warfare Center (NIWC) Pacific you’re more likely to hear machines whir than people. Ben Wong probably won’t say much because he works alone. You might hear radio station Magic 92.5; “Let’s Groove” by Earth, Wind & Fire plays. With the 50-year-old, solid-steel metalworking machines, framed film prints on the walls, and relative lack of screens, it’s one more detail giving the hazy impression you’re not in 2024 anymore.

Trace the moment back a little and you’ll find the work in that shop looks much as it did when Wong first joined in 2008 as an engineering technician after working as a machinist for the San Diego Department of Public Works. In its simplest terms, his job is to build model ships to sit under the arch up the hill so other engineers can test its antennas before they’re installed on a real ship 48 times the model’s size. In broader terms, it’s to save the Navy time, money, and to perfect high-frequency radio communications on its ships.

Wong fields questions in the form of digital blueprints: Here’s how we want to place the antennas, will it work? And he builds his part in the answer with wood, brass — malleable, rust-resistant, and easy to clean — and tiny antennas you could hold between your thumb and index finger. Antenna placement, surrounding add-on compartments, and anything solid can affect how an antenna receives radio waves, which means part of Wong’s job is building by hand miniature fixtures like satellite dishes, delicate enough to top a dollhouse.

Except the dollhouse is a ship, and it’s the key to a math problem instead of a toy: What’s the fewest number of antennas, and where should we put them, for full high-frequency communications coverage?

It may be the Navy’s singular question still answered by hand measuring, hand sanding, with wood and brass and a unique sort of professional labor enduring within these walls yet fading to history otherwise. Here the skill of solving Navy problems more with hands than computers is alive and raucous, spanning generations of model makers who dedicated their careers to this shop.

“This job requires an open imagination,” Wong said, “because there’s never a right way or wrong way of machining. Everybody machines differently. I’ve learned a lot from other journeymen,” some of whom we’ll meet soon. “That’s where I’ve learned a lot of my skills, just kind of talking to people and working together.”

Keep tracing backwards and you’ll meet first Jun Peralta, who shared the shop with Wong before retiring in 2018; then Bob O’Neill, who worked in the shop 1986 to 2008; then Fred Blas, 1976 to 2002. O’Neill and Blas saw the model range’s transformation from a zenith arch to a composite-material tripod arch in the 1990s, right around the time people first said computers would make model makers jobless — a prediction each subsequent modeler proved wrong.

Now they’re all back in this shop, pulling a sort of reverse interview: “You want to know the three most common questions we get from people who come in here?” O’Neill asks, first firing off one, admittedly, on my list.

I cross off “Do you build models for fun at home?” — they don’t. “Do you go home and write?” they ask. “That’s like asking an auto mechanic if he goes home and works on cars,” Blas says. I say sometimes, but come clean that I don’t go home and write about antennas for fun.

“Do they float?” They don’t — their bottoms are hollowed out for wiring. And, “They pay you to do this?” Hearing them talk about it, it’s almost a valid question. For the years their tenures overlap, going to work meant enjoying the satisfaction of the same hands-on craftsmanship hobbyists do at home for free, all among friends who’d prove to be lifelong.

O’Neill and Blas, for instance, were coworkers first, then neighbors starting in the late 1980s. They carpooled to this interview, which turns out less like an interview and more like a reunion. I ask Wong what kind of music he plays in the shop and he mentions Kool & the Gang’s “Cherish,” apt for the moment.

I ask Wong what he was doing last time he was happy at work, and he says he’s happy any time a model build helps solve someone’s problem. “Especially if it’s a rush job and they need it right away — getting it built well and in time without any snags brings me satisfaction. Like, ‘I did the job right. He met his deadline because I helped out.’”

I ask why brass models work better than digital ones and they tell me to ask Jodi McGee, head of the Electromagnetics and Advanced Technology division. She says, “Since I started at the Center almost 30 years ago, people have been saying that computational modeling should be able to replace brass modeling ‘in a few years.’ Periodically, our engineers check in on the progress of computational modeling for high frequency. It’s getting closer than ever, but there are still a few very challenging problems that prevent us from giving up brass models.”

NIWC Pacific’s design engineers do use computational electromagnetic modeling for predicting antenna performance in some situations, but it’s still a challenge in the two megahertz to 30 MHz band — classified high frequency, but a relatively low-frequency band with long signal wavelengths between 10 meters and 150 meters long. Because the wavelengths are so long, amounting to a significant proportion of the vessel’s size, the entire ship functions as one big antenna, a synergy of surface currents flowing over its complex shape. If that synergy isn’t accounted for, it can affect operational performance.

For now, brass modeling just predicts that operational performance better, capturing both the sum effect of the antenna network’s parts, and the minute details of its more complicated parts.

A fan antenna, for instance, could have six wires fanning out and running down the deck topside on each side of the mast. If you know the electrical impedance of the fan antenna — how resistant or reactant it will be to current running through it — before computer modeling, you can tweak the computer model to account for it, sure. But the point of modeling antenna configurations on Navy ships is to prove effectiveness before spending the time and money to build the antenna — or even the ship. And so brass modeling still wins.

The brass models are also cost effective. “We’ll build a ship model that can be reused over the life of that ship class, which may be 30 years or more,” McGee said. “Our Nimitz-class aircraft carrier model was built more than 50 years ago and is still in use.”

And when ship classes undergo modernization, modelers can validate planned changes won’t impede communications before any metal is cut in the shipyard. Effective models can prevent costly rework on actual ships, both in terms of labor and impacts to operational fleet readiness.

Brass modeling also taps more accessible skillsets, rare as they are. “Brass model antenna measurements are fairly straightforward, whereas computational modeling in this challenging frequency band practically requires a Ph.D. in electromagnetics to perform simulations and interpret results. So we’re still at least ‘a few years’ out from fully transitioning from brass to computational modeling for high frequency.”

For now it’s Wong and the Gang, passing the torch one by one, soon from Wong to the model maker he’ll train as his replacement before he retires in two years. Wong has trade expertise to pass on, which he describes as old school, requiring a more nuanced touch than can be input via computer numeric control. By comparison, larger, automation-friendly machines at the Center’s machine shop are fit for mass production; here, one needs to be comfortable both using manual machines, such as a manual machine lathe, and tools — sanding, carving, and soldering one-off models that will be used for a lifetime.

“One of the fun things about that is seeing your product through, start to finish, step by step,” Blas said. “In the machine shop, you develop the skills to know which steps to take. Here, those skills are the same, it’s just that some of those steps are on smaller machines, and some are by hand.”

The group seems to share this ethos: one of precision, dedication, high attention to detail. “And camaraderie,” Blas adds. They say there simply wasn’t enough room in the shop to not end up friends.

Before I leave, O’Neill tells me to visit the Miniature Engineering Craftsmanship Museum in Carlsbad, so I do, marveling with my own division head at the tiny artistry and raving about the satisfaction that comes from using your hands to turn basic materials into something new. Looking at pictures and stories of people who spent decades building intricate dollhouses, functioning miniature engines — and thinking about the vibrant personalities back at the shop — it strikes me there’s an intangible, personal quality to models, marked by the modelers who make them.

We get to the model ship section and learn about William Tompkins, who, starting as a teenager, built more than 307 ship models at 1:600 scale. He was so good at it that naval intelligence personnel, shocked at seeing accurate representations of then-secret radar antennas hanging out on a model in a Los Angeles department store window, interrogated him as a suspected spy. He wasn’t a spy — just smart — so they asked him to join the Navy, just 17 years old.

He went on to build exceptional careers both in the Navy and in support of government projects after, a major contributor to plans for Apollo space missions. He looks a little like our technical director, whose dad interned in the model shop in the 1960s; I’d swear there was a relation if they didn’t have completely different last names.

Somewhere in this history — no one can say when — brass models began to fill first “the barn,” a shed next door for storing models still used for testing, then overflowed into the “bone yard” — a ship cemetery for retired models so eerily captivating that passersby, complete strangers, have pulled over to the side of the road to ask how to acquire one of those old model ships.

Before the barn and the bone yard, before Wong and Peralta and O’Neill, Blas’ time in the shop overlapped with Joe Havlick, the very first, who started in 1951 and retired in 1979. Havlick would come along for the model range in its second form, the wooden arch in 1948. It’d only take a handful of years for ship modeling to prove its cost effectiveness; when USS Mt. McKinley (AGC 7/LCC 7) was recommissioned in 1951, engineers at the Antenna Model Range proved they could reduce the number of antennas needed on Mt. McKinley by two thirds.

Between Wong and Havlick there’s a small club of 30 people who spent anywhere from a few months to a few decades here. Before all of them, all the way back at the start, there were just three 100-foot-tall telephone poles that simulated incoming radio waves at various frequencies, and engineers on the ground below doing pretty much the same thing they’re doing now. There were modelers and machinists, some middle-aged and some yet to be born, some for summer internships and some for lifelong careers, all set to intersect and overlap here at the bottom of the hill. Some of their friendships would span nearly 40 years.

If I could, if it were open to the public, I’d prescribe an afternoon in the model shop as a retreat from constant change — energizing but not often enough leaving time to look back and appreciate the ingenuity that came before. Here innovation has been suspended in time since the late 1940s, when they first found a solution hardy enough to survive more than 70 years of subsequent breakthroughs. It leaves one wondering whether we should measure our innovations more by their staying power — how much we get it right the first time — than by their novelty.

For now, just know there’s a place under that big white arch, a time capsule for a niche sort of person doing a niche sort of work more enduring than rumors about computers making another trade obsolete, where craftspeople do things the old way, simply because it works.