Does Math Matter?

TBPN，科技每日秀 Substack 专栏。关注

Happy Tuesday.

We’re live on YouTube and 𝕏.

The current thing in tech and business is the drama around a mathematician seeming to accuse OpenAI of front-running him on solving the Navier-Stokes equation.

Today’s Lineup

Armada Co-Founder & CEO Dan Wright at 12:00 PM
Mobile Dev Memo Analyst Eric Seufert at 12:10 PM
Antioch Co-Founder Harry Mellsop at 12:40 PM
Cognition Co-Founder & CEO Scott Wu at 12:50 PM
OpenAI Co-Founder & President Greg Brockman at 1:10 PM
Forus Founder & CEO Sahir Jaggi at 1:30 PM
Split Pay Co-Founder & CEO Andrew Borovsky at 1:40 PM

Run of Show

Does Math Matter?

Tensions are high on the timeline this morning as the big AI labs race to prove Navier–Stokes. Clearly multiple teams are working on this with the best possible models and whoever publishes first will get a lot of clout among a very small group of mathematicians and AI researchers (provided the result is clean of course).

A similar drama played out around the 2025 International Mathematical Olympiad, where OpenAI and Google DeepMind both scored 35/42, or exactly gold-medal level, on the test. There was a bunch of back and forth because OAI validated with former IMO gold medalists while Google’s result was graded by the official rubric that year (internal to the organization that runs the competition). We called it “running the fastest 100m sprint in the parking lot while the Olympics happen in a stadium.”

So the timeline will debate exactly who got to the result first, how they did it, who they communicated around it, etc etc. The bigger question is does any of this matter? The Navier–Stokes equations themselves are already known. Engineers and physicists solve them numerically all the time for particular situations: airflow around a wing, water through a pipe, weather models, etc. That sounds like cause for optimism, better weather models can save lives during hurricane season, more efficient wings can save fuel and reduce carbon emissions, these are things every day people do care about, even if marginal improvements are barely noticeable and don’t exactly fit neatly on a yard sign. But proving Navier–Stokes is unlikely to have any measurable impact in these areas, because engineers already solve the equations numerically as mentioned above.

Clearly it matters for hype and vibes. Solving hard math problems is a good heuristic to show how quickly models are progressing, but benchmarks and raw metrics do something very similar. The number of people who will be convinced that AI is progressing quickly after hearing of a math breakthrough vs say, just hearing about the number of parameters in a model or amount of GPUs racked in a data center is very small.

All this is to say that computer use demos are much more meaningful to AI diffusion than math challenges. Watching AI actually do stuff on your computer (even moving the cursor) for you is much more of a “feel the AGI” moment for most people. So, we’re probably back to the drug development conversation for a while. Cure cancer, don’t make mistakes. Even that won’t solve everything though, I see people all the time bemoan the lack of “real world progress” despite the fact that GLP-1s seem to be completely on track to cure obesity in short order.

Clip Spotlight: StepStone Group’s Hunter Somerville says a “long overdue” reckoning is coming for SPV

"It's almost criminal how bad it is."

"The most egregious part of the entire continuum is SPVs."

"Even separate from the funds, it's the insane 4-10% management fees they're charging, and syndicating out to the long tail of investors that are out there."

"They're opaque, they're multilayered, people don't even know what shares they own underneath it... There will be a reckoning and lessons learned, and I think it's well overdue."