Longreads + Open Thread

Longreads

• Ben Brubaker has a great piece in Quanta on the discovery of the fifth Busy Beaver number. This is a CS/math construct—given a Turing machine with N possible rules, how many steps does the longest non-halting program take before it stops? This problem is trivial for sufficiently small N, rapidly gets harder, and at some point veers off into impossibility. There's also a nice combination of theory and brute force: the winning candidate was first proposed as a solution in 1990, but this wasn't proven until this year. A lot of cleverness, a lot of tedium, and a problem that turned out to be divisible into chunks that different researchers could focus on—it's a good story about how research gets done, and a reminder that there's still more math to learn.
• Santi Ruiz in Statecraft on How to Catch a Lab Leak, the story of a multi-decade investigation into an anthrax outbreak in the USSR. The Soviet explanation was that it was caused by tainted meat, but there was circumstantial evidence pointing to an accidental lab leak. This piece is a fun combination of science and diplomacy, and a look at late Soviet dysfunction. Here's the story of how the Soviet minister of health got his start: "During the revolution, in Moscow, if you wanted certain kinds of jobs, you had to queue: if you wanted to be a carpenter, you wait in the long line, you get to the table, you sign up. So he and his friend wanted to become blow torch welders, but they got in the wrong line by mistake, into the line to become doctors. They say, “So what the hell? We'll become doctors instead of welders.” That's how he becomes a doctor, and eventually minister of health."
• Kevin Erdmann questions the narrative that the financial crisis was caused by excessive homebuilding. He points out that housing supply growth in the most active housing markets had been elevated for a long time before the bubble period, making it hard to attribute the crisis to merely building houses where there wasn't demand for them. But this picture is incomplete: higher spending on housing, and speculative purchases of homes, can add to GDP growth—construction workers and mortgage brokers all get paid, and spend money, and newly-wealthy homeowners may be tempted to cash out some of their equity and spend it. So there can be a setup where a a state's housing-demand fundamentals are partly a function of housing demand itself.
• Pradyumna Prasad and Jordan Schneider write in praise of RAND, one of the most accomplished think-tanks of all time. One important point they make is that staffers at a military-funded think tank during the height of the Cold War were quite ideologically aligned, and it's easy to get more done if there isn't much disagreement on what the long-term aims are. (That's a magnitude, not a direction—if everyone's aligned in their pursuit of a bad idea, they'll be a lot more effective at accomplishing that, too.) But one of the difficulties think tanks run into is that the pure pursuit of optimal policy means they have minimal leverage for implementing their ideas, and at least some of the time, the reason those ideas didn't get put into practice in the first place was political.
• Maxwell Tabarrok has a fun argument against the "burden of knowledge,", the view that progress slows down in many fields because there's simply too much to learn before you can start making meaningful contributions. What pushes against this burden of knowledge is a particular kind of knowledge—the kind that explains many different observations with the same general model. In some fields, that's true, and a few broad laws can replace a lot of anecdotal observation. But a lot of that falls into the trap of losing the Beginner's Mind. A computer scientist could, in theory, sum up their entire body of knowledge by saying "Consider what you could build by putting together NAND gates," but the profundity of that statement is proportionate to how much computer science the person hearing it already knows.
• This week in Capital Gains, we look at what people mean when they talk about "strategic" decisions. There's a wide range of things you might mean by "strategy," and this makes it a good dodge. But it's a useful concept, because sometimes businesses need to focus less on immediate opportunities and more on what maximizes the scope of future ones.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Earlier this week, we wrote about this piece questioning AI economics, which has been making the rounds recently. The Diff counterargument is that some of AI's benefits will show up as cost savings at non-AI companies, not as explicitly AI-related revenue. Have you seen any interesting examples of non-AI companies saving money with LLMs or other AI tools?

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• A company building ML-powered tools to accelerate developer productivity is looking for someone who would rather write extremely technical English than code, but not bad enough to take a haircut on comp. (Washington DC area)
• A defense and public safety startup sitting at the intersection of OSINT and facial recognition is looking for product-minded fullstack engineers who are interested in national security and novel applications of OSINT data. Python and Typescript experience is a plus. (Remote)
• A mission-driven ed-tech startup is deploying AI tutors to revolutionize education. They have a strong customer base in Latin America and they're in need of a driven founding engineer. Key skills: TypeScript, React, Node.js, Postgres. Spanish and Portuguese proficiency preferred.(Remote, East Coast time zone)
• A startup building tools to enable people build communities and own their own data is looking for a site reliability engineer with deep Kubernetes experience. (Remote, SF preferred)
• A company building the new pension of the 21st century and building universal basic capital is looking for fullstack engineers with prior experience in fintech. (NYC)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.