What I'm listening to - May 2025

How to Build Endurance from Huberman Lab

It's easy to rag on Huberman as "bro science" but I still listen to episodes that sound useful. Here are my notes from this one:

• Endurance depends entirely on neurons that control willpower and muscular function. "We have this thing called the central governor, which decides whether or not we should or could continue or whether or not we should stop, whether or not we should quit."
• Muscular endurance is local and non-cardiovascular. You can build this with high-rep, low-weight exercises.
• Long duration aerobic exercise builds mitochondrial density. This is what lets you become more efficient. You can get the same output but burn less fuel.
• High intensity aerobic intervals helps your blood deliver more oxygen to the muscles and builds lung capacity so the mitochondria get the fuel they need. These are done in 3–8 intervals of 3-6 min work / equal recovery (e.g. mile repeats, 4-min bike intervals).
• High intensity anaerobic intervals help those mitochondria use as much oxygen as possible. The best way to build this is with 3-12 sets of something like 3:1 work:rest (e.g. 10 sets of 30sec hard, 10sec rest on an assault bike).

My biggest takeaway is that it's good to do a mix of exercise, including some (higher-rep) weightlifting and high intensity intervals, even if training for something like a marathon. Full episode here.

2027 Intelligence Explosion on Dwarkesh Podcast

Every Dwarkesh episode is a banger. This one is with the people behind AI 2027, Scott Alexander and Daniel Kokotajlo. Scott is the "blogger extraordinaire" behind Slate Star Codex and Daniel is an AI researcher who's made some very accurate predictions (see here).

There are some pretty wild forecasts around AI fully automating AI research and the impact on economic growth and national security. For example, they predict that the US government will seize car factories and convert them to robot production. We have a historical example when car manufacturers shifted to making bombers in WWII.

Definitely worth listening to and reading the article. Full episode here.

Forage: The Trillion-Dollar Opportunity in Restricted Payments

42 million low-income Americans rely on food stamps and Ofek Lavian saw this pain point first-hand by trying to integrate food stamps at Instacart. The legacy infrastructure required tons of compliance and policy work that led to the opportunity for a new startup: "The amount of compliance and work required to enable EBT is massive … it took us years before we processed our first dollar in EBT"

Great example of attacking a dusty industry (payments) by understanding a specific painpoint in the market (restricted payments like food stamps). There's riches in niches and love that Forage is tackling such a meaningful problem. Full episode here.

Epic Systems on Acquired

I'm just wrapping this one up. A few takeaways so far:

• A lot of Epic's success came from being a single, integrated system while other products were poorly integrated. Incredibly the founder Judith Faulkner had the epiphany on the right architecture in the mid-1970s and I love how trippy she sounds: "The sun was shining, I was dis-attentive, I was just sitting there, and suddenly it all came to me. Here’s how you build it."
• The story about how they got started with only $70k of outside capital (for 50% dilution!) is pretty wild.
• Awesome customer obsession. New hires do "immersion trips" where they spend time in "clinical settings like operating rooms to directly observe workflows".

Full episode here.