It was 10:30 PM and bright daylight under a blue sky when my helicopter lifted away from Toolik Biological Station, just south of the Brooks Range in Northern Alaska. The permafrost tundra was a lime green below us—it was the middle of June, and this was the short window for photosynthesis. I had come from NSF headquarters in Arlington Virginia because I didn’t believe what I’d been told: that the NEON site here was nearly complete.

The reports said construction was well underway. According to the updates crossing my desk at the National Science Foundation, work at this particular remote monitoring station was progressing on schedule—tower foundation laid, equipment deliveries confirmed, site preparation complete. The National Ecological Observatory Network was building a continental-scale observatory, eighty field sites across America measuring everything from soil microbes to atmospheric carbon, from Alaska to Puerto Rico. Half a billion dollars. A thirty-year mission to understand how ecosystems were changing across an entire continent.

The helicopter banked north. We were close. I pressed against the window.

We were above the site’s GPS coordinates. There was nothing. Just tundra. I pressed the video function on my iPhone and collected the data.

No tower foundation. No equipment. No site preparation. The primeval landscape stretched unbroken to the horizon, exactly as it had for ten thousand years since the last ice age retreated. According to the paperwork, construction was well underway, and millions had been spent. According to reality, no one had broken ground.

That’s when I knew we would have to fire them and find another builder.

The problem started, as these things often do, with the best of intentions.

When the National Science Foundation conceived NEON in the early 2000s, the vision was breathtaking: a network of standardized ecological observatories spanning the entire territory of the United States, collecting identical measurements from tundra to tropics, mountains to prairies. For the first time, ecologists could compare apples to apples—soil microbes in Kansas versus Massachusetts, carbon flux in Alaska versus Alabama, all measured with the same instruments, the same protocols, for thirty years.

It was exactly the kind of transformational infrastructure science needed. And so NSF did what seemed entirely logical: we created a nonprofit corporation, NEON Inc., and populated its board with distinguished ecologists who understood the science.

They were brilliant scientists. They understood ecosystems, biogeochemistry, microbial ecology. They had collectively published thousands of papers, trained generations of graduate students, and spent careers asking profound questions about how nature works.

What they didn’t understand was how to manage a half-billion-dollar construction project.

By the time I arrived at NSF as Assistant Director for Biological Sciences in 2014, NEON was hemorrhaging money. The initial budget had been ambitious but defensible. Now we were looking at an $80 million cost overrun with no clear path to completion. Construction timelines had slipped repeatedly. Some sites that should have been operational were years behind schedule. And as I’d just discovered in Alaska, some sites that were reported as “progressing” didn’t exist at all.

The reporting problem was symptomatic of a deeper issue. NEON Inc.’s board met quarterly, reviewed progress reports, asked questions—but they were asking the wrong questions. They scrutinized scientific protocols: Were the soil samples being collected at the correct depth? Was the CO2 sensor calibration adequate? These were important questions, but they weren’t the questions that would determine whether NEON actually got built.

They should have been asking: Why is the tower foundation delayed by six months? What’s the critical path dependency? Where are the project management controls? Who’s accountable when a milestone slips?

These weren’t their questions because these weren’t their skills. You don’t learn construction project management by studying forest ecology. You don’t learn procurement logistics by measuring carbon flux. The board was doing exactly what they’d been trained to do—think like scientists. The problem was, NEON needed someone thinking like a construction manager.

When I got back to D.C., that video became the smoking gun and we began to clean shop. Not long afterwards, I walked out to the mall parking lot behind our building and dialed the NEON board chair to deliver the news: you’re fired.

The board didn’t go quietly—they felt betrayed. The board chair had been in a leadership role at NSF, himself. The other members were respected ecologists who had viewed me as an academic colleague on rotation to the leadership at the agency. But the math and state of play on the ground were unforgiving. At this rate, the project was going to crash.

At the same time, our oversight folks, including the US National Science Board, Congress and the Inspector General’s office all saw disaster written all over that video. The optics for the agency were appallingly bad.

The decision was easy, but institutionally it was a tricky one. For one thing, we were years into the construction and there was a very real risk of cancellation being forced upon us notwithstanding the sunk costs. I still really believed in the transformational potential for the project. Another problem was finding a new incumbent to finish construction. There wasn’t a lot of precedence for that, not only at NSF, but across the government. We were advantaged in that the funding vehicle was something called a ‘cooperative agreement’ and not a contract, but when we changed horses, we would be going full tilt—this had to be done at speed.

Congress helped. During my testimony to the House Science Committee, I was pushed hard on being willing to fire NEON.inc. After some hemming and hawing, while on the hot seat, I agreed: yes, that might be the right course. And then it was all made possible by our general counsel, who found the right emergency mechanism to both gage interest and award the partial cooperative agreement to Battelle.

Battelle brought decades of experience building big projects for the US government at the scale that NEON required. They approached the rescue with a ‘can do’ attitude, no doubt incentivized by the opportunity to compete for an operations and maintenance award upon completion of construction.

On the in-house side, we changed up things just as massively. The project management was moved to our Division of Biological Infrastructure where we had the human capital to work productively on the rescue. The new team included some of the most talented project managers I’ve ever worked with.

Today, NEON is fully operational. Commissioned in 2019, it’s producing open data as a multidimensional time series of how the United States biosphere is evolving across nearly 200 data products. It’s director and chief scientist is a scientific visionary with her own substantial track record of scientific discovery. Battelle got the job done and the rescue was a success.

But NEON was not unique. We constantly confuse domain expertise with managerial expertise. The job of R1 university president is now too far removed from the traditional academic pipeline that produced the farm team for those jobs. Similarly, physician leaders responding to public health emergencies must also have a sophisticated understanding of political nuance and public communications. If they don’t, then they risk both trust and the public welfare.

What was the lesson of NEON? Honestly the hard truth: sometimes firing the leadership is the most respectful option.