Leah Libresco Sargeant: Platinum Is Expendable. Are People?

‘Efficiency’ is a thrilling word in the executive suite but much less so on the factory floor.

Right now, someone in a distant office is asking: “Can’t we do it better with less?” Efficiency is the art of subtraction, and increasingly it is people, not things, that we aim to do without. Yet on the other side of the ledger is what efficiency buys—luxury goods that become cheap commodities, medical breakthroughs that become over-the-counter cures.

Teasing out what efficiency demands and what it offers us in return is the topic of Brian Potter’s new book, The Origins of Efficiency, which is part history, part handbook. Potter, a structural engineer and writer of the substack Construction Physics, offers a close reading of several major industrial breakthroughs. The victories won by efficiency are easy to underrate as we get used to our prosperous new normal, and Potter’s histories remind us that something as quotidian as a lightbulb is the triumphal arch of a fiercely waged war.

A modern incandescent lightbulb testifies to the grinding battle that material scientists waged against platinum. Platinum has an incredibly high melting point (3,225 degrees Fahrenheit) and thus could stand up to the intense conditions required to transmute electricity into visible light, but customers couldn’t bear the cost of the precious metal.

In 1890, platinum made up one third of the cost of the bulb. To cut that cost down, engineers first found a way to eliminate platinum clamps from the platinum lead-in wires. Next, the platinum lead-in wires dwindled to almost nothing, relying on other metals except at the exact point where the wire passed into the bulb. This last iota persisted until 1912, when a cheaper alloy was found that could do the job. This gimlet-eyed zeal for reduction is romantic at the lab bench and a blessing for consumers. But it’s hard not to suspect that people are the platinum of many modern processes.

Platinum isn’t disconsolate at being released from duty, but people have found it hard to bear our shift away from manufacturing. And now, many more workers fear displacement by artificial intelligence. Potter’s book touches only lightly on the human costs of streamlining and engineering breakthroughs. In his final chapters, he considers a future where automated home construction allows a simple home to be built in less than 24 hours. There are no humans on site.

Such a shift would be an earthquake for blue-collar workers, and the off-site jobs Potter imagines (remote compliance inspectors reviewing construction footage) clearly wouldn’t be a substitute. Potter’s focus on materials, not men, makes it hard to imagine handing this book to a Ford worker who had just been laid off the line.

Potter instead focuses on a different civilizational cost—that of not making breakthroughs.

One of the most moving victories of re-engineering that Potter describes is the production of penicillin on an industrial scale. The mold first discovered by Alexander Fleming was finicky, and the lifesaving antibiotic it produced was hard to isolate. As Potter recounts, it took a year to produce enough penicillin for the first human trial of the drug—a single policeman sick with a staph infection. The doctors had enough to give him a miraculous yet temporary recovery, but not enough to fully quash the infection. Desperate, they purified penicillin he excreted in urine and gave it to him again so that not a drop was wasted. Yet it wasn’t enough and the policeman died.

Repeated refinements gave us the present day’s abundance. The bacteria grew better thanks to an improved medium, but still only on the surface of the liquid. To get past this bottleneck, scientists scoured any mold they could find for a strain that produced penicillin while suspended—allowing the drug to scale in three dimensions. As Potter summarizes, the United States went from not having enough antibiotics for a single man in 1941 to producing enough for the entire Allied Armed Forces in 1943.

When a product is completely new, like penicillin, there aren’t entrenched interests who want to protect the old way of doing things. Speeding up penicillin production was pure gain, with no jobs lost and many created. But, as Potter’s examples show, the line between streamlining an old product and inventing a new one is blurry.

Making a product efficiently often requires making it into something new. The individually blown glass lightbulbs housing once-platinum wires were replaced by bulbs blown by a machine into a mold, displacing the glass blowers. The process was physically similar but standardized and automated, though it too was ultimately overtaken by something much stranger. The Corning Ribbon machine, invented in 1926, sent a current of molten glass flowing down a belt peppered with holes. Rather than gathering onto blowing pipes, it sagged through the holes into molds moving in sync with the ribbon of blazing molten glass, and was then blown out with precisely timed puffs. The first automatic bulb blowing machine, created in 1921, made blanks at the rate of 1,000 bulbs per hour. Five years later, the Corning Ribbon machine could produce 16,000 every hour.

When a bulb is made orders of magnitude cheaper and faster than the one that preceded it, it becomes a fundamentally different product. Efficiency overhauls aren’t just about increasing margins for owners, they’re about making it possible for certain products to exist at all. Although cost and personnel minimization is a part of industrial improvement, in Potter’s view, it’s not the fundamental driver of efficient processes.

A truly efficient system isn’t one that has been cut to the bone, it’s one that operates continuously. As Potter writes, it “continuously transforms inputs into outputs without any delays, downtime, waiting, unnecessary steps, or unneeded inputs. A steady stream of inputs goes in, and a steady stream of completed products swiftly and smoothly comes out.” This was what the continuous, glowing river of glass gave Corning.

The concept entails a few counterintuitive results. If you can massively improve one step of your process such that it punches blanks from steel sheets twice as fast, it’s no good to you unless you can speed up the next step to match. When any one step runs faster than the overall flow, you build up a buffer of waiting parts, increasing the amount of raw material you’ve paid for without being able to sell it. Excess stock also means you need to pay for a physically larger building.

A continuous flow process is a careful dance of interlocking rates and manipulations. The gains to production can be so large that they force consolidation. When the 1881 Bonsack machine consolidated every step of cigarette manufacture into a single, flowing process, it was so efficient that, by 1885, it took just 30 Bonsacks to produce enough cigarettes for the entire United States. Consolidating production is, at least initially, a boon for producers and employers.

Managing Efficiency

One drawback to continuous flow processes is that when companies need to expand to a new facility, it’s easy for a hyper-optimized process to fail. As Potter notes, modern semiconductor manufacture is so sensitive to environmental factors that yields change slightly with the cycles of the Moon and the menstrual cycles of female employees (the latter is believed to be because of hormonally induced fluctuations of the oil production of their hands’ sebaceous glands). Companies minimize variation where possible, and people are variation. Intel, for example, developed a “Copy EXACTLY!” protocol which requires each new semiconductor fab to exactly mirror an existing, successful fab, right down to paint colors.

Companies that work at a large scale often require apprenticeships in successful stores to pick up by practice what is hard to teach. McDonald’s needs new franchisees to see how the chicken is fried, just like fabs need their workers to see how silicon wafers are etched. When a company’s hyper-efficient machines let them downsize to a fraction of their former workforce, it diminishes the number of workers with deep knowledge to hand off to others. Pair consolidation with offshoring, and there are far fewer American workers with the know-how to reengineer or reproduce the process.

When consolidated processes go wrong, the results can be disastrous. An American formula factory’s contamination problem caused massive shortages in the middle of the pandemic, which were exacerbated by protectionist policies that barred Americans from buying European baby formula and restrictive vouchers for poor mothers that only allowed them to purchase specific brands of formula that suddenly did not exist. Many of the generic medicines we rely on are produced abroad with only intermittent inspections. When something goes wrong, it can trigger similar shortages. When something goes wrong and is not detected due to the difficulty of coordinating overseas inspections, Americans take pills that are little better than placebos.

Yet it’s hard to imagine reshoring all of these industries or forcing critical industries to be less efficient and more dispersed. It’s easier to prioritize knowledge transfer and inspections to mitigate the fragilities of consolidation. If you’re building to last over the long term, these priorities can be integrated into your process.

For an extreme example, look to the Ise Jingu grand shrine in Japan, which has been rebuilt from scratch every two decades for more than 1,300 years. Continuous renewal guarantees there are deeply experienced artisans with the capacity to rebuild the shrine when needed. America needs similarly deep knowledge of the processes we most depend on, even if we don’t shift the bulk of manufacturing back home.

Creating the Future

To be a major player in whatever breakthroughs come next, the U.S. also needs a predictable, growth-oriented regulatory environment that gives engineers room to capitalize on innovation. Many efficiency-oriented processes depend on scale and repeatability, so upfront investment only pays off if you can expect to make the same thing again and again indefinitely. Often, American regulations are too volatile or too aggressive to allow that to happen.

In the nuclear industry, changing and tightening requirements means that reactor designs resemble extremely expensive Etsy commissions rather than a smoothly running, repeatable industrial process. In home construction, over-specified and localized zoning codes prevent homebuilders from developing standardized homes that are compliant across the nation. When regulations or tariffs whipsaw as political parties exchange power, business executives can’t make investments for the long term.

Rather than aching to return to older, lower-yield processes, the U.S. should plan for a manufacturing future that frees American inventors to keep innovating. The upside of efficiency can be hard to anticipate—it was a short gap between penicillin as an artisanal product and an industrial one. We should expect that some medical and material rarities today will become cheap commodities within five to ten years. Artificial intelligence has the potential to speed up that search, especially when it comes to new drug discovery.

Potter’s book suggests that freeing entrepreneurs to innovate won’t just get us better versions of whatever we have now. It means we may create entirely new industries that make our children healthier and more prosperous than we thought possible. I don’t favor consumption for its own sake, but I do think we’re likely to see massive energy and medical breakthroughs before my children graduate from high school. And then they may work in fields we don’t yet have names for.

Streamlining will eliminate some jobs. Making formerly expensive inputs cheap may create new industries. Compensating the losers during the transition is a job for the government. It’s not the government’s job to foreclose future wins.