10 things I consumed this week
How to be more agentic, and why this Biblical rain isn't solving the California drought
This week, I had the odd phenomenon of having a Substack article go to #1 on Hacker News. It was very surprising (I didn’t expect it, nor do I know who posted it originally.) It felt like a spurt of addicting energy that I’ve been missing for a little while — it’s such a flashback to my time managing Hacker News comments on my product launches at Stripe. I owe a lot to HN (it was what convinced me that Stripe was a “thing” back in the day), and it’s so interesting to see how the forum has evolved over the years.
And of course, welcome to anyone new to Working Assumptions, I’m so happy to have you here. Please reach out to say hi!
What I read this week:
Why record rain hasn’t washed away California’s water woes: It’s been raining for three straight weeks in California. I have been relatively miserable about it, except for the fact that California needs water, and the rain might be helping the state recover from drought. It turns out that’s not the case: we might be getting a lot of rain, but we haven’t been able to store the water so that we can use it during less-rainy times. All of the water is apparently running straight into the ocean.
Open Source AI Business Models and Brand Moats: I’ve always been skeptical of “open core” companies (Hashicorp, Mongo, etc) because they are so ineffective at value capture. The company gets incredible engineers to work for a very long time on an ostensible public good, and then attempts to monetize some apps and “upgrades” atop the core. The user is getting the vast majority of the value from the free thing! Why is this a good business!? However, it also seems possible that open source foundational AI models might be able to reverse that trend. This article from the always thoughtful Michael Dempsey explains the open source business model for AI companies and why, this time, it just might work (albeit — with a very unsavory-feeling thinkfluencer brand moat!)
Enterprise sales meets product development: My friend Sebas is one of the most insightful people I know and an excellent engineer. He writes very clear, tactical advice on how to navigate the choppy waters of enterprise sales and SaaS, which reminds me of this (intentionally provocative) PG adage: “Enterprise software companies aren't technology companies, they're sales companies, and sales depends mostly on effort.”
How to rediscover childlike wonder: Apparently there are eight wonders of life that often propel people into states of awe: moral beauty (witnessing the virtue of others); collective effervescence (often experienced in large groups like those at weddings, rallies or sporting events); nature; music; visual design (beautiful buildings, paintings and the like); spiritual and religious experiences; life and death; and epiphanies.
8 Lessons from Curb Your Enthusiasm: I love Larry David — but so does everyone (except Elmo, who he punched?!) If you don’t watch Curb, I’m sort of envious of you because I wish I could watch it for the first time all over again. This post highlights the best things about Curb framed as lessons, most specifically the value of developing your own taste, and the idea that success can come later in life.
Stratechery — the view of Meta from the top: Ben reviewed Meta’s latest earnings with a special guest, deep diving into the ads business and compare/contrasting Meta with Snap. Snap’s earnings are down ~30% in part because they can’t bring in users from high ARPU (average revenue per user) markets. A quote I thought was interesting talked about a big reason Snap’s limitations being their use of external cloud providers. Will we see people build their own data centers in the AI era?!:
With their [Snap's] cost structure, too, particularly with their outsourced compute, bringing on users is a bit of a dangerous game for them [Snap].
ES: And it always has been. That was one of the key risks when they went public, is they were totally dependent on external cloud providers, they didn’t build this infrastructure and that’s one of the issues with being sub-scale. They even mentioned in the investor letter that their cost of revenue per DAU went up because they’re trying to compete with Facebook or Meta in terms of building out these ML models to do personalization at scale, to do better ad ranking, and those things have an absolute cost. They don’t necessarily scale with DAU, they have an absolute cost in just investing in that stuff, and it’s just a much bigger burden for a Snap than it is for a Meta.
LVMH — Acquired Podcast. Ah, Acquired. My favorite podcast — it feels like I’m hanging out with my smart friends. I loved the LVMH episode even though it’s an endearing eyeroll to hear them discover handbags for the first time (at one point they compared LV fashion shows to Monday Night Football…men!!) The part I like the most was when they discussed the benefits of the conglomerate model for LVMH and the difference between “premium” and “luxury.” Luxury has both economies versus diseconomies of scale: luxury has to remain limited in production, you can’t overproduce the good. Walking the line to reap the conglomerate benefit while not over-reaping is very difficult. As an example, you can’t share designers between Fendi and Celine, but you might be able to make a collective ad buy, or a collective department store negotiation, with all of the brands. I wonder how much we can borrow from this idea with software businesses — we often talk about the misery of starting an internal startup within a big company, and I think the LVMH style conglomerate model could help us carve out a better operating path.
How to be more agentic. Cate Hall “was a Supreme Court advocate and the number one female poker player in the world; started art and perfume companies; and cofounded Alvea, a pandemic medicine company, when it set the record for the fastest startup to take a drug to clinical trials.” For all of these achievements, Cate credits her ability to be more agentic (to have more “manifest determination to make things happen.”) This article includes her tips for growing your own agency.
On Putin’s interview with Tucker Carlson: Did you want a long lesson in Russian History? Boy, do I have the interview for you. Noah Smith has a good post reviewing the interview and his reactions.
“Underdog” relationships: I think one of my least favorite qualities of famous/successful people is their inability to adjust to being a “top dog” instead of an “underdog.”