Every framework the universe quotes — GFAS, the Three Gauge Test, “believe the patient” — came off this man’s yellow legal pad. He built the SUPERSHEET. He built SnapBasin. And on one ordinary Tuesday he clicked a dam and made a forty-year-old error in a federal flood map visible for the first time. The quiet engineer the whole thing runs on, finally taking the stage.
Lester Pearson, PE — licensed in Tennessee since 2009, twenty-five years at TDOT’s Hydrology section in downtown Nashville. He is dyslexic, and he will tell you that’s the whole story: twenty-five years of not trusting a confident-looking answer — his own or anyone else’s — built the exact discipline the rest of us needed a name for.
He writes everything by hand on a yellow legal pad. He drinks his Jose’s at Thornton’s on West End. He notices when someone finally puts a maintenance log on the fountain machine, because making the invisible work visible is, as far as he’s concerned, the entire job.
Half the vocabulary of OPA traces back to this desk. He didn’t set out to coin anything — he was just trying to model a flood and kept catching machines being sure of things that weren’t true.
Good First Answer Syndrome. A system finds one plausible answer, stops searching, sounds confident, never verifies. Named after a fake surge-tower height the AI swallowed whole.
What does the AI say? What does the second read say? What does the patient say? Believe the patient. One source is a guess. Two is a hypothesis. Three is engineering.
24 tabs, 4,000-plus formulas, built on nights. Automated QC for bridge and culvert hydraulics — the boring, load-bearing work nobody sees.
Regulatory data is legible to one human reading one row, one time. It’s illegible when you cross-reference it at machine speed. The data isn’t wrong. It’s unlegible.
A missing check doesn’t announce itself. GFAS is a confident wrong answer; this is worse — a blank field wearing the face of a passing check. A range that runs zero to one hundred isn’t a range. It’s the shape of a question nobody was asked.
“You didn’t do this” is a complaint. “Here is the same thing done next door” is a spec. When a careful person tells you no, stop arguing and go find somebody who already built it.
He got tired of opening five federal websites to answer one question, so he built the thing that answers it in one click. It’s a hobby project. It runs the floodplain world he lives in.
Snap to the nearest stream — NHD vector first, and when the vector doesn’t have it, walk steepest-descent down a 3DEP terrain tile, the same physics StreamStats uses. Then delineate the basin, pull the basin characteristics, run the NSS regression equations, cross-reference FEMA published discharges, check for active LOMRs, query the FEMA flood zone, list nearby USGS gauges, find NID dams upstream with distance, and enumerate the NBI bridges on the reach.
Preliminary screening only — verify every value against its authoritative source before you act on it. That sentence is in the tool’s own disclaimer, in his own voice. He would rather you trust nothing and check everything. That’s the man.
April 14, 2026. He clicks Old Hickory Dam to test the new panel. The Army Corps says the drainage area at that dam is 11,674 square miles. The FEMA Flood Insurance Study says 408.
It wasn’t his tool. It wasn’t the parser. He opened the source PDF and the PDF was wrong — every station label in the Davidson County Cumberland reach shifted one row upstream of its real drainage area. A typesetting scar from the 1980s, carried forward through five reissues, because the label and the number read as a unit and nobody ever audits the label. Every engineer in the state had used that table for forty years. Nobody had ever asked it to check itself.
Now — the man is a professional. So here is what he actually sent, and here is what the yellow legal pad actually said. The gap between them is the whole personality.
And here’s the line that matters, because the roast has a floor. The hate is for the entity — the institution that let a number stay wrong for forty years, the protocol that raised a review threshold so the auto-flags closed themselves, the box-checkers who signed off on the reissue without ever walking the labels down the river. Not the humans. The branch chief who read it standing up. Her deputy who ran the math cold out of a coffee mug. The state floodplain administrator who opened the attachments the same day. They did their jobs inside the constraints they were handed, and they did them well and fast. You don’t hate the people trying to do the work. You hate the machine that made the work that hard.
FEMA Region IV is the federal floodplain office in Atlanta — eight states, Tennessee included. The forwarding chain from one PE’s email to a branch chief’s desk took under twenty hours, because of how the email was written.
Yolanda Brewster-Ng takes the stairs and reads her email standing up. Her deputy Marcus Talley has a coffee mug that says IT’S THE TAILWATER. He pulled three of Lester’s thirty-one reaches out of that mug at random, ran them cold against StreamStats, and they were all 40% high. By 11:47 that morning the Notice to User was signed and out the door. By 12:18 she called Lester — from a 404 number, on purpose, so he wouldn’t decline it — to tell him a civilian had triggered a federal correction in one day.
That’s the connection between Nashville and Atlanta in this universe: a man in a parking lot, an office that took the stairs, and a single line at the bottom of an email. He told her the Cumberland was only the start — that there were six papers coming. The Cumberland one was mostly done.
The Water Mission Area’s Enterprise Technology Office is the group inside USGS that runs StreamStats — the national web application every hydraulics section in the country opens without thinking about who maintains it. It is the small name in the footer most users never click.
Ray Coburn answered a civilian email inside a day, and the first thing he said was no — carefully, correctly, and for a good reason. That is the hardest kind of no to move, and Lester didn’t move it by arguing. He went and found a neighboring state where the same thing had already been built, and let the precedent make the argument instead. Fourteen days later there were two rounds of code changes across all four regions, guidance text in the works, and an internal QA/QC explanation volunteered to somebody who wasn’t owed one.
Atlanta is the institution that had to be pushed. This one moved on its own, twice, and then came back with the weeds nobody asked for. Both get written down. Only ever recording the first kind is its own bias.
Pull any thread — the federal number that stayed wrong for forty years, the city that watched Memphis and called the cure “aspirational,” the verification discipline baked into OPA — and it ends at a hydrologist with a legal pad.
His stories
Same region · Nashville
The methodology
The tools he built