Trending: One year later, Microsoft AI and Research grows to 8k people in massive bet on artificial intelligence

Nathan Myhrvold
Nathan Myhrvold shows off a fragment of the Chelyabinsk meteorite in his office at Intellectual Ventures in Bellevue, Wash. (GeekWire photo by Alan Boyle)

BELLEVUE, Wash. – Millionaire techie Nathan Myhrvold is used to stirring up controversy over issues ranging from patent licensing to dinosaur growth rates, but now he’s weighing in on an even bigger debate: the search for potentially hazardous asteroids.

In a 110-page research paper posted to the ArXiv pre-print server and submitted to the journal Icarus for peer-reviewed publication, Myhrvold says the most comprehensive survey of near-Earth asteroids ever done, known as NEOWISE, suffers from serious statistical flaws.

“They made a set of numbers that look right, They have what Stephen Colbert calls ‘truthiness.’ But that doesn’t mean they are right,” he told GeekWire today during an interview at the Bellevue headquarters of Intellectual Ventures, the company he founded.

On the other side of the debate, NEOWISE’s principal investigator, Amy Mainzer of NASA’s Jet Propulsion Laboratory, says it’s Myhrvold’s numbers that don’t look right.

“The paper contains multiple mistakes, including the confusion between diameter and radius (which is by itself enough to render the results wrong),” she wrote in an email to GeekWire. “Nonsensical asteroid diameters are presented throughout by the author.”

Mainzer noted that Myhrvold’s paper has not yet gone through formal peer review.

On one level, the controversy focuses on the scientific equivalent of inside baseball – ranging from whether an equation refers to diameter or radius, to how much a principle known as Kirchhoff’s law of thermal radiation should be taken into account when judging how big an asteroid is. But on another level, it plays into far bigger topics – such as the reproducibility of scientific findings, and how best to find killer asteroids before they find us.

Mainzer’s next proposed mission, an infrared space telescope known as NEOCam, would build on NEOWISE and intensify the search for asteroids. Right now NEOCAM is one of five finalists vying for NASA’s nod and up to $425 million of funding. NASA is due to announce who’ll get the money in September.

Will NEOCam’s telescope win out? Or is a ground-based observatory known as the Large Synoptic Survey Telescope, or LSST, good enough to do the job? That’s part of the subtext for Myhrvold’s asteroid studies.

“I am in favor of people finding asteroids. And that absolutely could include NEOCam,” Myhrvold said. “But if LSST is already going to find a subset, you should make sure that your instrument is ideal for the remainder.”

Myhrvold has been interested in asteroids for decades, going back to his days as a student at Princeton, and later as Microsoft’s chief technology officer. Now that he has a net worth estimated at $650 million, with an innovation factory humming along at Intellectual Ventures, he can afford to indulge that interest (along with paleontology, modernist cuisine and much, much more).

He began his deep dive into asteroid data analysis a year ago, after the nonprofit B612 Foundation asked him to back its plan to put an infrared telescope into space to identify potentially hazardous near-Earth objects, also known as NEOs. Several multimillion-dollar projects have been proposed for such searches, including B612’s Sentinel Space Telescope as well as NEOCam.

Myhrvold decided to try comparing the capabilities of those projects with the LSST, which is currently under construction in Chile and is due to start science operations in 2022. A couple of Myhrvold’s fellow Microsoft veterans, Bill Gates and Charles Simonyi, have contributed millions of dollars to support that project.

Myhrvold’s first study, which appeared in March in the Publications of the Astronomical Society of the Pacific, concluded that the LSST provided the best option for the asteroid search, in part because it’s already being built. He said proposals for additional space-based telescopes would have to be carefully checked “to assess the actual value added,” and called for the development a standardized method to compare proposals. (B612 didn’t get the money.)

Then Myhrvold started asking questions about the NEOWISE asteroid data. He said he couldn’t get answers from Mainzer or other members of the science team.

“I tried and tried and tried and tried,” he said. “I tried through third parties, and the third parties were told, ‘Oh, he’s a crackpot,’ basically. That’s what I got.”

“And they’re not wrong,” he added with a laugh.

So he conducted a ground-up analysis himself – using the same raw data that the NEOWISE team started with, from NASA’s Wide-Field Infrared Survey Explorer (also known as WISE). In the newly posted paper, he reports that he got dramatically different results, with margins of error that amounted to as much as 300 percent.

Myhrvold said the problems appear to be due to NEOWISE’s “very funky, ad hoc, invalid set of statistical analyses.” He said there wasn’t enough allowance made for the effect of reflected sunlight on near-Earth objects, or for variations in NEO composition. Some of the diameter measurements appeared to have been taken from earlier missions.

“They ought to get that data as right as they can, including that they ought to be open to criticism,” he said.

But Myhrvold is coming under criticism as well. After taking a look at the paper, Mainzer said the funkiness appears to be due to Myhrvold’s misunderstandings and mistakes. She said a follow-up study showed that NEOWISE’s measurements of asteroid diameters agreed with independent infrared observations to within a 10 percent margin. And she said some readings from earlier missions were used for calibration, which would explain the apparent duplication that Myhrvold noticed.

“All of this points to the fact that the paper requires following the normal scientific process of peer review through an independent journal,” she said.

UCLA astronomer Ned Wright – who was the principal investigator for the WISE mission – said he spotted several problems with Myhrvold’s paper, including the equation that confused radius with diameter. Today, Myhrvold agreed that he had to make at least one revision, and acknowledged that there may be more to come.

Right or wrong, Myhrvold said he’s glad to serve as a “gadfly” for science. In an earlier debate over dinosaur growth rates, his complaints about statistical methods led to corrections in previously published studies.

“If every single thing I bring up is wrong, but that causes a better explanation, a more thorough understanding – if it causes them to publish enough of their data analysis that I can replicate it, for example – that is a victory for science, even if I’m wrong,” Myhrvold said. “I’ve made myself a target for them to fling mud at. ‘Oh, he doesn’t understand this, he doesn’t understand that.’ We’ll see.

“My life is not going to change dramatically one way or the other. Like anybody, I’m embarrassed to do things that are wrong. But if you can’t admit that, you’re not a scientist. There’s a related thing in business. You know, I’ve been in the technology business for a long time. People talk about ‘risk-taking, risk-taking.’ Some people talk about risk-taking, and then when something screws up, they say ‘What the f—, what happened?'”

“What part of risk-taking do you not understand?”

Subscribe to GeekWire's Space & Science weekly newsletter

Comments

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.