Two Minutes: The Part of the Design Review Nobody Budgets For [2026]

The Pipe

Two Minutes: The Part of the Design Review Nobody Budgets For

The enclosure knowledge is real, it is specific, and it is almost never in the room where the decision gets made.

Written by Scot Free

A meeting room in Building 2 of the Meridian campus. On the table, a parking sensor: a sealed puck about the size of a hockey puck, magnetometer inside, built to be set into the pavement of a single parking space. A few hundred feet away, in the corporate parking garage, more of them were being tested.

Our visitors that day were a father and son from a small shop outside St. Louis. They were in the building for a different reason, as part of the narrowband IoT initiative, and the puck was just on the table.

The son picked it up. He turned it over. He looked at it for about two minutes.

Then he said: "The design is upside down. This will leak."

And then, without much of a pause: "Here's a better design."

He was right

The pucks leaked. I wrote about that deployment and its arithmetic in the Project Smart City opener, so I won't relitigate it here. The company behind them had raised about $13.9 million and made a genuinely clever call on the sensing: one magnetometer instead of a stack of sensor types, which cut the hardware cost by roughly 90 percent against an industry standard near $250. That was real engineering and it deserved the coverage it got.

The seal was not in the coverage. The seal is never in the coverage. And the seal is what decides whether a sensor in Denver pavement is a ten-year asset or an eighteen-month consumable.

This piece is not about who was wrong. It is about the two minutes, and what has to be true for someone to have them.

Why he could do that

The shop was Labyrinth Technologies, out of Hazelwood, Missouri. Ted Stegeman is the CEO; his son John is the CTO. At the time of the St. Louis press coverage in January 2020, John was 23.

Their public record is a street-lighting system. Downtown STL Inc. ran an open design competition in 2015–16 after a civic push to make downtown safer, and Labyrinth won it. They deployed 135 smart streetlights on Market Street as a $400,000 demonstration project, then the program expanded across a 360-square-block area. Their luminaire lights street and sidewalk independently with dual-zone optics, and their controller platform, Xanthon, came out of that same project. The city's own CTO, Robert Gaskill-Clemons, later said Labyrinth took five months to get new small-cell, streetlight and utility-pole specifications approved by the Board of Public Service, a process that normally takes far longer.

That is the résumé line. Here is the part that matters for the two minutes: they put their own hardware outdoors, in a river city, on public poles, and then they had to keep it alive. High-pressure sodium fixtures replaced. Concrete poles with cores too small to retrofit. High-voltage supply that has to be converted down to run low-voltage IoT devices. Rain during a demonstration on Clark Street before a Blues game. Every one of those is a chance to learn something about water, thermal cycling, gasket compression and drainage that does not appear in any datasheet.

John did not intuit the failure. He recognized it. He had, by then, already been designing and building things in the physical world for most of his life: a FIRST robotics team he started in high school, Mars rover builds for a NASA exhibit at the St. Louis Science Center before he was 18, and then a company's worth of outdoor enclosures. When he turned that puck over, he was comparing it against a mental library of parts that had already failed on him.

That is what expertise in this layer actually is. Not cleverness. A catalog of specific failures, most of which were never written down.

It didn't start with John

When I sent Ted a draft of this piece, he wrote back with a story I didn't have.

His father started in the family machine shop, Ted's grandfather's, at six years old. He learned to weld and to machine before school had much of a claim on him. Then he went into the Navy as a machinist's mate, in the engine room of a destroyer. At nineteen, in the middle of the Pacific, he was tearing down and rebuilding diesel engines, and he knew more about them than the officers and the formally trained men around him.

Not because anyone had taught him more. Because he had done it more.

That is the same argument as this piece, two generations earlier and an ocean away from a parking garage. The engine room did not run on whoever held the rank or the certificate. It ran on whoever had actually had one apart.

Count it out. A great-grandfather's machine shop. A machinist's mate on a destroyer. Ted. And John, turning a puck over in Building 2. Four generations, one skill: read the part in your hands.

Why the room usually can't

Think about who was in that decision chain. Venture investors, who diligence the market, the model, the team, and the traction. Corporate development, which diligences the strategic fit. Product, which owns the roadmap. Engineering, which owns the electronics and the firmware. Every one of those functions was competent and every one of them was looking at a different question.

None of their processes has a line item that reads: hand the unit to someone who has pulled a flooded one out of the ground in February.

There is a deeper reason than oversight. The knowledge is tacit and it is negatively coded. It exists as "we stopped doing it that way," and the reason often lives in one person's memory of a warranty return. Nobody publishes the enclosure that failed. Vendors certainly don't. So the only reliable way to access it is to put the object in the hands of someone who maintains that class of object and watch what they do in the first two minutes.

This is where the usual skepticism advice runs out. You can audit a number without being a domain expert: check where the data came from, who got counted, what got left out. That works on a claim. It does not work on a part. To audit a part, you need someone who has replaced one.

This is every deployment, not just sensors

The same asymmetry runs through industrial and municipal telemetry everywhere:

• The plant electrician who knows which conveyor trips when the humidity comes up, and which panel to open first.
• The water operator who knows which lift station floods and in what order the alarms lie to you.
• The fleet tech who knows which harness chafes at 200,000 miles because he has cut three of them open.
• The line lead who can tell you that the vibration reading on unit 4 has always run high and it has never meant anything.

Every one of those people holds calibration data that no sensor is producing, and almost none of it reaches the dashboard. Worse, the dashboard is often built to replace their judgment rather than to encode it. That is backwards. The first thing a telemetry program should do is interview the people who currently detect the failures, because they are the existing detection system and you are about to compete with them.

The Underground Take

Here is the rule, and it costs nothing: a design review is not complete until someone who maintains that class of object has physically held one. Not seen the CAD. Not reviewed the spec. Held it, turned it over, and been asked what they think. If that person does not work for you, borrow one. Ask the utility, the fleet shop, the public works department, the vendor's own field service crew.

What I remember most about that afternoon is not the verdict. It is that Ted could not stop talking about how fast his son had gotten there. I understand that pride better now. He was not only proud of John. He was watching the family trade show up in someone else's meeting room, the same thing his father carried into an engine room at nineteen. That is what a family shop is: knowledge handed down close enough to the work that it transfers by watching.

A round of venture funding buys a lot of things. Engineers, tooling, a supply chain, a press cycle. It does not buy the two minutes. The two minutes have to be earned by someone, somewhere, who has already lost units to the thing you are about to build. Sometimes they were earned three generations before anyone walked into the room.

Labyrinth's streetlight had its own economics, and the payback math on that product is a separate story. That one is coming in the Energy edition of Project Smart City.

Sources

• St. Louis Public Radio, St. Louis on the Air, "How This Father-And-Son Team Invented St. Louis' Latest Smart Technology," Jan. 24, 2020 (Ted and John Stegeman; John's age and FIRST robotics and St. Louis Science Center background; Downtown STL Inc. project scope).
• Labyrinth Technologies company news releases, 2017–2019 (135 streetlights on Market Street; $400,000 demonstration project for Project #LightMySTL; second-generation dual-zone luminaire; Clark Street demonstration).
• Labyrinth Technologies, "Labyrinth Completes Development of New St. Louis Lighting Infrastructure Standard," 2019 (City of St. Louis CTO Robert Gaskill-Clemons on the five-month specification approval; 60,000 streetlights; concrete pole and voltage constraints).
• Labyrinth Technologies, Innovations page (Xanthon connected technologies).
• Denverite, Nov. 2016, and Next TV, Feb. 2018, for the parking sensor's public claims and acquisition, as cited in full in the Project Smart City opener.
• Stegeman family history as told to the author by Ted Stegeman, September 2026, used with his permission.
• The meeting room account is the author's firsthand recollection. The quoted verdict is as remembered; John Stegeman reviewed this account in September 2026 and had nothing to add.

Corrections and additions: Scott@IoTunderground.com.