This Fucking Guy: Nathan Myhrvold

I first heard about this fucker circa 2009, and he rubbed me the wrong way then. And since then, he's coninued to vie for a This Fucking Guy treatment. I am happy to oblige.

The Man Who Confused Diameter for Radius

Before we get to the patents, a case study in what the persona actually looks like when it's operating in the wild.

In 2016, Nathan Myhrvold decided that NASA didn't know how to measure asteroids. Not metaphorically. He is not an astronomer, has never worked in planetary defense, and had exactly one credential relevant to the argument he was about to pick: he is extremely smart and has never once let that go unnoticed. He'd been sniffing around NASA's NEOWISE mission data, the program responsible for cataloging more near-Earth asteroids than any other observatory in history, and concluded the whole thing was built on statistical garbage. So he wrote a 110-page paper, skipped peer review, and posted it straight to arXiv, because apparently the process that catches your mistakes is for other people.

It did not go well. NASA's own planetary defense chief told the New York Times, with the kind of patience usually reserved for a smart nephew who just discovered Ayn Rand:

"He's a very smart man, but that doesn't make him an expert in everything."

NEOWISE principal investigator Amy Mainzer's team went through the paper and found, among other things, that Myhrvold had confused diameter with radius in one of his own formulas. Not a subtle statistical quibble. A ninth-grade geometry mistake, in a paper he used to accuse a NASA team of not understanding basic physics. Asked years later why he kept fighting a dispute that had, by that point, eaten two years and a FOIA campaign, Myhrvold's own answer was "bad judgment," offered as a joke. It got through peer review eventually, in Icarus in 2018, and there was in fact a real methodological point buried in there. That's not the story. The story is the process: parachute into a field you don't work in, force a team of actual experts to spend years responding to you as if you were their peer, and when the dust settles on "partially correct," carry it around like a Nobel.

Hold that shape. You're going to see it again, and next time the stakes aren't asteroid diameters. They're every company that ever had to write a check to make Nathan Myhrvold go away.

The Brochure

The version of this story Myhrvold would tell you, presented fairly enough that taking it apart means something.

The official biography is genuinely impressive and that's the problem. Physics PhD by 23, postdoc work with Stephen Hawking, Microsoft's first chief technology officer, the guy who helped build the company that ate the 1990s. Then he leaves in 2000 and reinvents himself as a one-man Renaissance, and this part of the résumé is also, annoyingly, real: Modernist Cuisine, his six-volume, 2,400-page cookbook built on food science rather than folk wisdom, won two James Beard Awards and is taken seriously by actual chefs. He publishes peer-reviewed paleontology papers on dinosaur tail biomechanics. He co-founded Global Good with Bill Gates, aimed at malaria eradication and vaccine cold-chain logistics in the developing world, which is the kind of thing that makes it structurally hard to just call the guy a villain. He sits on TerraPower's board, Gates's actual nuclear reactor company, doing actual nuclear engineering.

The pitch, in other words, is: I am so brilliant that I overflow into everything, and you should trust the overflow. It's a good pitch. A tech journalist mocking Intellectual Ventures' new "VP of Global Good" hire in 2011 put it about as well as it can be put:

"it's like Darth Vader doing charity work"

Which, sure. But the joke only works if Darth Vader has a day job, and this is where the brochure stops being the whole story.

Twelve Hundred Shell Companies and a Losing Bet

The mechanism. This is the part that should actually make you angry, not the cookbook.

Here's the day job. In 2000, Myhrvold and three co-founders started Intellectual Ventures in Bellevue, Washington, with a business model built on one insight: patents are assets, assets can be aggregated, and aggregation at sufficient scale is its own kind of leverage. Two decades later, three of his four co-founders had left or scaled back their involvement, leaving Myhrvold as the only one still showing up. IV doesn't build anything. It buys patents, mostly from other people's inventions, and by 2012 estimates from Stanford researchers put the pile at 30,000 to 60,000 patents. By 2018, Forbes was citing estimates north of 95,000. That made IV, for a while, the fifth-largest patent portfolio held by any U.S. company.

The part that took actual investigative work to surface is how IV structured all of it. Law professor Robin Feldman and patent attorney Tom Ewing spent years tracing the ownership and found at least 1,276 shell companies, up from an earlier count of 954 as researchers kept finding more, each one holding a slice of the portfolio behind a wall of paperwork specifically built so nobody outside IV could tell what it owned or who was licensing what. When Xilinx got sued and tried to figure out who actually held the patents against them, the trail split into seven different IV-affiliated shells, and a California judge ended up shipping part of the case to Delaware because the ownership was too tangled to sort out where it stood. That's not bureaucratic sloppiness. That's a moat, built out of LLCs, specifically so that the people IV squeezes can't organize a defense against a target they can't even identify.

The tactic Feldman and Ewing name for how this gets used is called privateering, and it's worth spelling out because it's genuinely clever in the way a good con is genuinely clever. IV approaches a company and demands a license on a patent. If the company says no, IV sells that specific patent to a more aggressive third-party troll, one with nothing to lose and every incentive to sue for a number much higher than the original ask. Once the target has felt what that costs, IV comes back around and asks again, this time about a different patent in the pile, having just demonstrated exactly what happens if you keep saying no. It is a protection racket wearing a licensing department's name tag, and it's why "I'm just investing in preferred stock," Myhrvold's go-to defense for what IV does, is such an insulting comparison. Preferred stock doesn't come with an enforcer. When challenged on this at a tech conference in 2012, his response was that Microsoft, Apple, and Facebook were all buying up huge patent portfolios too, so:

"They're doing what I'm doing!"

Which is true, and also the single worst defense available, since "several of the wealthiest companies on Earth also engage in extractive patent hoarding" is an argument for burning the whole practice down, not a character reference.

Now for my favorite part, the part that should be the actual headline every time someone writes about this man: it didn't even work. Myhrvold raised close to $3 billion across two twenty-year patent funds from institutional investors, including university endowments and, in an irony that writes itself, Microsoft, Apple, and Sony, some of the very companies IV was busy suing. By 2016, the larger fund, Invention Investment Fund II at $2.3 billion, had posted an annual return of negative 12.59 percent and handed back just 37 cents on every dollar invested. The smaller fund, $590 million, was at negative 36.73 percent, having returned two cents on the dollar. Two years later those numbers still hadn't recovered, and Forbes noted, with the specific dryness reserved for financial journalists watching a rich man's hobby cost other people money, that Myhrvold seemed to be spending a great deal of his time on his own passions, like a 2,400-page cookbook, a 2,600-page bread guide, and picking fights with asteroid scientists.

So here's the actual shape of the man, stated plainly instead of implied through a brochure: he built a 1,200-shell-company extraction machine that shook down operating companies for licensing fees under threat of manufactured litigation, and even that fundamentally predatory business lost his own investors' money for a decade while he was off doing the thing everyone was supposed to find charming.

The Persona Was Always Doing Cover Work

The through-line to now, which is the part almost nobody has connected.

Here's what the polymath act is actually for, whether or not Myhrvold ever consciously designed it that way. It's a laundering process. Feed in "structured 1,276 shell companies to extract money from other people's work while losing his own backers' money" and what comes out the other end is "eccentric genius who wrote a cookbook and argues with NASA." That conversion is the whole value of the persona, and it explains something that happened very recently and that almost nobody noticed.

Documents that surfaced between 2024 and 2026 place Myhrvold in Jeffrey Epstein's black book with extensive contact information, on Epstein's flight logs for two trips in 1996 and 1997, and as a contributor to Epstein's 2003 birthday scrapbook, where he's listed as a friend. On their own, in isolation, that's the same low-grade guilt-by-proximity that touched a long list of powerful men who crossed paths with Epstein in the 1990s. What makes Myhrvold's case different, and what actually deserved more attention than it got, is reporting from Seattle Times journalist Alex Halverson, picked up by KUOW in February 2026, documenting an email correspondence between Myhrvold and Epstein running from 2011 to 2017. That's not a pre-conviction acquaintance. That's five years of ongoing personal contact with a convicted sex offender, well after his 2008 conviction, in a register the reporting describes as chummy and, at points, sexually joking. There's no allegation in the reporting that Myhrvold did anything unlawful, and it's worth being precise about that rather than implying more than the record supports. But five years of that kind of ongoing familiarity with Epstein, specifically, is not nothing, and it is not a story about a guy who innocently sat next to the wrong person at a TED dinner once in 1998.

Myhrvold's spokespeople have given reporters the same statement every time this has come up: shared interest in science philanthropy, TED-adjacent overlap, regrets ever having met the man. It's a boilerplate defense that gets harder to credit the more documents surface showing how much contact there actually was. And the response to all of it, in a media environment where far more marginal Epstein connections have ended board memberships and torched reputations, has been almost total silence. One regional radio segment. No resignations. As of that same reporting, Myhrvold is still CEO of Intellectual Ventures and still vice chairman of TerraPower's board.

That's the persona doing exactly the work it was built to do. A patent troll with a five-year Epstein correspondence is a genuine scandal. A "Renaissance Geek" who happened to know a guy is a Tuesday. Nobody had to lie for that trick to work. They just had to let the cookbook do the talking.

Still Running

No manufactured ending here. Just the state of things.

Nathan Myhrvold is not stupid, and he's not even wrong about everything he's ever said. That was never the indictment. The indictment is a specific, repeatable pattern: take one real credential, apply it with total confidence to fields where it doesn't transfer, force the people who actually know what they're doing to spend years responding to you as a peer, and walk away treating "partially correct" as full vindication, whether the field is astrophysics or patent law or, apparently, personal association with a predator. The shell companies are still active. The fund losses were never repaid. The board seats are still his. Nobody's coming for any of it, and there's no reason to think anyone will. The mechanism is still running exactly the way it was built to.

This fucking guy, amirite?

Keep the lights on

Like what you've read, but not able to switch to paid? Why not buy me a coffee instead! All money goes to help defray the expense of running the site.