BrianPotter

The Engineer Who Asked Why

He spent fifteen years building things. Then he spent five years explaining why building things is so hard. The second project turned out to be more useful.

Brian Potter is a structural engineer by training and a writer by compulsion. His newsletter, Construction Physics, publishes long-form essays about buildings, infrastructure, and industrial technology - the kind of writing that asks how we went from erecting the Empire State Building in 410 days to spending 1,500 days on 432 Park Avenue, and refuses to accept "it's complicated" as an answer.

He started it in September 2020 with zero subscribers, no marketing plan, and one engineer's lifetime of frustration. Today it has 72,000 free subscribers and sits at #35 among all technology newsletters on Substack. Derek Thompson, author of Hit Makers, has called him "the single best writer in America on how important things get built."

That phrase - more complicated than they seem - is a fair summary of Potter's entire intellectual project. He approaches construction the way an epidemiologist approaches disease: not to find one villain but to map an entire system of incentives, institutions, regulations, material costs, and human behavior. The result is writing that rarely arrives where you expect.

Why Buildings Are Stuck in Time

That gap is the engine of everything Potter writes. Manufacturing got cheap. Agriculture shed 80% of its workforce. Semiconductors got 300 million times cheaper per transistor. Construction just... didn't. And nobody had a satisfying answer why.

Potter spent a decade and a half as a structural engineer looking for that answer from the inside. He worked as a consulting engineer, a subcontractor, and a design-builder. Each vantage point gave him a different angle on the same paradox. What he found was not incompetence or laziness but something more interesting: a system that rationally resists change because every incentive points the wrong direction.

The data tells a story that the industry rarely tells about itself. Skyscraper speeds in New York declined 50% since the 1960s. Individual construction tasks have not gotten cheaper since at least the 1950s. Brick prices have not declined since the mid-19th century. And prefab construction - the supposed silver bullet of housing innovation - typically costs no less than conventional building. The math, as Potter puts it, just does not work out.

Another Day in Katerradise

Before the newsletter, there was Katerra. Between 2018 and late 2020, Potter managed the structural engineering team at Katerra's Atlanta office. Katerra was a SoftBank-backed construction startup that raised over $2 billion and employed 8,000 people across nine countries. It was going to vertically integrate and industrialize everything. It was going to fix the industry.

He does not write about Katerra with bitterness. He writes about it the same way he writes about everything: with curiosity about systems and a determination to understand why outcomes diverge from intentions. The Katerra piece became one of his most-read early essays - not because it was a hit piece, but because it was the most honest accounting anyone had written about why a well-funded, intelligent attempt to change construction failed anyway.

The frustration was the newsletter. Construction Physics launched in September 2020, while Potter was still at Katerra, as a way to document what he was learning about a system that resisted change even when change was well-funded and earnestly attempted.

186 Essays, 600,000 Words, One Question

Construction Physics publishes roughly once a week. Each essay takes Potter two to three weeks of research. He works on multiple pieces simultaneously. At 600,000 words across 186 essays, he has written the equivalent of eight full-length books about industrial systems - most of which are not about construction.

The name is somewhat misleading. Yes, the newsletter covers housing costs, prefab, and construction productivity. But it also covers semiconductor fabs, WWII aircraft production, the electrical grid, solar manufacturing, jet engines, and the fundamental question of how any production system gets cheaper over time. "Construction Physics" is better understood as a frame: the physical constraints of building things, and why those constraints are so hard to overcome.

The common thread is learning curves: how do production processes get more efficient, and when do they stop? Potter is obsessed with cases where costs fell dramatically - solar panels, semiconductors, wind turbines - and equally obsessed with cases where they didn't - construction, nuclear, titanium. The asymmetry is the story.

The Origins of Efficiency

50 Things, 600,000 Words Condensed

In April 2025, after writing the equivalent of eight books about industrial systems, Potter published "50 Things I've Learned Writing Construction Physics." Tyler Cowen featured it on Marginal Revolution the same day. It was written in a rush after a work retreat, when he was, as he put it, crunched for time.

It is one of the most information-dense essays you can read in thirty minutes. Selected findings:

15 Years Building. 5 Years Explaining.

Potter graduated from Georgia Tech with a civil engineering degree in 2007 and went straight into precast concrete. He spent over a decade rotating through the industry's different angles - consultant, subcontractor, design-builder - collecting a systems-level view of why construction is so resistant to change.

How He Actually Works

Most technology newsletters cover software. Most infrastructure analysis covers policy. Potter covers neither and both. He brings an engineer's rigour to economic history, a historian's patience to technical detail, and a reporter's instinct for the specific fact that collapses a received narrative.

His research process is not fast. A single 5,000-word essay takes two to three weeks. He works on multiple pieces simultaneously. By the time a piece publishes, it has gone through primary sources, industry reports, academic papers, and usually a few rounds of conversation with people in the industry.

He is also unusually willing to kill a popular narrative with data. The prefab housing boom? Doesn't pencil. Three Mile Island as the death of nuclear? The orders had already stopped. The Ise Jingu shrine, rebuilt every twenty years for 1,300 years, is his favorite example of tacit knowledge: construction skills disappear not when they stop being documented but when they stop being practiced.

He is also quietly funny. A man who once ran an eBay automation side hustle for designer men's clothing, generated several hundred dollars a month, and concluded the ROI was too low for the maintenance time - then applied the same analytical framework to billion-dollar industries - has a consistent sense of what he finds interesting.

Where to Hear Him Think Out Loud

Links & Profiles