Three weeks into a new job, I got blamed for a sustainability plan I didn’t write. It turned out the Board was right to be angry
I’d been in this room quite a few times. It was on the 12th floor of the old Arlington NSF headquarters and housed a 15-meter-long conference table and chairs for an audience of perhaps fifty. But on this November day in 2014, it was my first time in the hot seat. I was giving a report to the US National Science Board on progress with the troubled iPlant Cooperative—a large cyberinfrastructure investment of my directorate at the University of Arizona to serve the plant biology community.
Before I was more than a couple of paragraphs into my presentation, there was trouble. There was something called a ‘sustainability plan,’ and apparently, the iPlant people had not taken it seriously. As far as they were concerned, the money would keep flowing (from NSF or somewhere) forever. It got heated. I was three weeks into my new job as the head of biological sciences and had no clue that the Board took its oversight role seriously. I’d spent the previous 16 years as a university institute director (with decanal rank) and had absolutely zero experience with either large cyberinfrastructure projects or sustainability plans.
Under direct questioning, my lack of experience and preparation became obvious. A board member was so angry that she told me to my face, in front of the assembly, that my responses were “personally offensive to her.” I apologized, calmly, but she didn’t look appeased.
It turned out that the Board was incensed not only because the plan was bad—there was no credible revenue model, no institutional home identified, and no transition timeline. The Board was furious because it was clear that the plan wasn’t serious. Rather, it was a cynical box-checking. If it had been in the present day, there would have been accusations of using ChatGPT. So, it wasn’t a communication failure. It was a substance failure. The iPlant team hadn’t taken the instructions seriously.
It took two more years, but the story does have a happy ending. The iPlant team at Arizona took the feedback seriously and reorganized the project, rebranding it as CyVerse to serve a much larger scientific community beyond plant biology. Crucially, the team brought in leadership beyond Arizona and rethought the value proposition. The expanded team added science-education expertise from Cold Spring Harbor Laboratory and HPC capacity from TACC, the Texas Advanced Computing Center. The NSF remains the key investor, but the funding base has broadened beyond the Biological Sciences Directorate. The focus these days is on democratizing high-performance computing, providing access to US researchers who need compute power but lack it at their home institution. The box-checking exercise from 2014 eventually became a real institutional commitment. And even while I was still at NSF, the results became clearly visible, not only to those of us in NSF leadership, but also to the community.
The iPlant case wasn’t an anomaly. I’ve watched the same scene play out at two other $50M+ projects since.
How does this happen? Why spend months working on a grant renewal only to deliver a package with a fundamental scale mismatch? What produced the delusion that $10M+/year operations can land softly without years of serious groundwork?
It turns out that most academic scientists don’t plan for endings. At the psychological level, sustainability planning feels like admitting defeat. From a cultural standpoint, the grant cycle rewards acquisition rather than stewardship. And there’s always the deferral pattern: “We’ll figure that out in year 8.”
All of this results from a fundamental skills gap: excellence in STEM grant writing (what we commonly call ‘grantsmanship’) differs from business case writing. In fact, none of these domains— institutional negotiation, stakeholder development, or revenue modeling—are present in the science training pipeline. PIs face what Donald Rumsfeld would have called the unknown unknown: they don’t know what they don’t know.
So, what would good look like here? First, on projects like iPlant, with a typical ten-year lifespan, sustainability should be a priority for a dedicated team in year 3, not year 8. That team should include a named senior transition role with real authority — someone whose job is explicitly to plan for the project’s end, not its continuation — analogous to a Chief Risk Officer. In the NSF ecosystem, a midpoint review should have real teeth — a required external review panel with a go/no-go funding decision — long before things reach the National Science Board for oversight. Finally, there are models worth borrowing from: the nonprofit sector and major research infrastructure projects like CERN and the Vera Rubin Telescope have learned, sometimes painfully, how to plan for institutional continuity at scale.
The Board member who chastised me wasn’t wrong — even if she came very close to ending my tenure at NSF three weeks after it began. She was angered by cynicism born of ignorance — and that sequence matters. The box-checking comes second. The failure to understand why the box exists comes first. It’s part of why I took the Chief Risk Officer role on my current NSF-funded project seriously.
The problem is very real. All current large-project PIs should be asking themselves what happens to their project when the federal funding ends — and they should have a real answer, not a paragraph of verbiage.
The Board was watching then. Someone always is.
Discoverers and Discoveries 