Longreads + Open Thread

Longreads

• Abraham Thomas of Pivotal with another great look at data companies, this time focused on defensibility. A big focus of this piece is on how LLMs change the economics of data businesses, but as with other technological changes, the new thing is a great way to understand the status quo—crypto can teach you a lot about payments, banking, and currency, and in a sense we didn't really understand the PC and Internet revolutions until we speedran both a second time in the form of smartphones. It covers a wide range of models, from businesses whose most valuable data depreciates almost instantly (exchanges) to companies that really capture value by being in the middle of an expensive and data-adjacent process (Salesforce) to companies that are selling access to a static dataset of historical information in order to make a forward-looking decision (credit scores). Fun throughout, with many good concepts that will show up again.
• Ray Doraisamy interviews @its_not_real on software in oil and gas. If you think about the software you interact with, most of the user-hours are in consumer apps (social, streaming, search, a long tail of destinations for those) or email clients, IDEs, Excel, CRMs, and, again, a long tail. But weight your software interactions by dollars, and a lot more of them are the invisible infrastructure that powers everything else. Including, in this case, getting hydrocarbons out of the ground and converting them into useful energy. It's a good reminder about the fractal complexity of the world, and about how a large fraction of the software industry does work that's invisible to almost everyone unless something breaks.
• Vinson Cunningham in The New Yorker on the appeal of the New York Post. One of the features of populist right-wing journalism is that there doesn't need to be all that much pretense—the hot-button issues are literal, or at least have literal representation, and in a city as big as New York every day has a good chance of offering a little morality play. Climate change and inequality just don't offer as many stark and literal representations as crime or immigration. So a publication like the Post is naturally entertaining, even if it's a funhouse mirror for people who don't share its politics. The Post hasn't made sense as a business for a long time—second-place papers, even in very big cities, just don't have the economies of scale of the #1 player. But the Post has been owned by one ideologically-motivated heir or another since 1939, and of all the world's weird art projects subsidized by eccentric rich people, it's arguably the one enjoyed by the most people.
• Maggie Harrison Dupré has a good profile of a prolific purveyor of AI slop on Pinterest and other social networks. The piece makes an interesting point about the shape of the problem: the subject of the piece is a social media spammer, but he also sells courses on how to spam social networks with AI-generated images, and like most such people he's reluctant to give a breakdown of how much of his income comes from doing the work versus selling courses. But elsewhere: "'Ten pins daily is not going to cut it,' he adds, explaining that he posts around 80 AI pins a day in his efforts to manipulate Pinterest's algorithm—enough to get his pins to 'cruising altitude,' he says, but not enough to get hit with a spam notice by the platform." If being an actual human is proof-of-work, it's actually economically optimal for people who've learned to game the system to sell courses to other people who can do it: that's the only way they scale.
• In NYMag, James Walsh has yet another piece about higher education and LLMs, this one about how students are cheating their way through their classes (apparently classes that don't just use blue book exams or exams done on airgapped computers to determine most of GPA). One question that comes up in this is: what do the students, and the professors, think they're actually doing here? When systems can be gamed, it's worth asking what the system is trying to accomplish. In some domains, people are perfectly aware of this: chess players know that computers can beat them, but that winning is not the only point of chess; weightlifters realize that pulleys would make the deadlift a lot easier. The outcome is a proxy, not the end goal, so it's worthwhile to ask: if gpt-4o can write plenty of grammatically-correct sentences, summarize texts, write snippets of code, etc., why exactly is society devoting so many nice campuses and talented teachers to the project of teaching people to replicate the outputs of a cheap consumer product? And, if students aren't doing the work and professors can't tell the difference, what's the resulting credential really worth? (Note that in this story, some of the sources are better than others at delivering a polished story directly targeting someone who doesn't ask any follow-up questions.)
• In this week's Capital Gains, we look at a realist model of corporate control: the CEO is whoever can say "do this" and expect it to get done. (A lemma here is that not every organization has anyone in charge; inertia can be an incredibly powerful force sometimes.)

Books

How Star Wars Conquered the Universe: The Past, Present, and Future of a Multibillion Dollar Franchise: To be fair, Star Wars is not humanity's greatest artistic achievement. It's derivative in some ways, incomprehensible in others, ahead of its time in many technical respects, but not a great artistic achievement. On the other hand, I vividly remember watching The Empire Strikes Back in preschool, and I devoted a substantial share of my free time from ages ~10-12 to reading Star Wars Expanded Universe novels. (Takeaway: Thrawn is a pretty cool concept for a character, but if he were analyzing humans what would he look at? Would he contemplate the Mona Lisa? Watch every season of Cheers? Mainline the Twitter firehose?) Star Wars is compelling as a cultural phenomenon even if it's not especially significant as an artistic one.

The book bounces back and forth between telling the backstory and story of the creation of the Star Wars movies, and talking about various parts of the fandom, like the 501st Legion or the insta-fans who watched every single showing on the day Star Wars premiered in theaters.

That structure is a good representation of Star Wars itself, which is both a consumer product—movies, shows, games, theme parks, action figures—and a fandom. One thing that stands out is how contingent the whole thing is. All that infinite Star Wars lore on Wookieepedia starts with the first movie, which itself was a snapshot of a rapidly-evolving process where Lucas would assemble names and concepts from random real-world influences, Verbal Kint-style: Han Solo is named after solo cups, R2D2 comes from a shorthand reference to "second reel, second dialog," one of the extras in a previous Lucas movie had a friend named "Bill Wookey," etc. But basically every random visual, bit character, or offhand reference to prior events has now been fleshed out with a detailed, complicated backstory, a backstory so contradictory that much of it had to be declared non-canonical so Disney could keep its own stories straight.

This is an odd flywheel: Lucas threw lots of details into his universe in order to make it look like the story was taking place in a detailed world, not as an isolated little vignette. (This is roughly the same reason the characters' uniforms look grubby and the tools and vehicles are all banged-up—Lucas wanted a fictional universe that looked lived-in.) And then every bit of that mystery, like Obi-Wan's offhand reference to the "clone wars," or a random lizard skeleton on Tatooine, now have detailed backstories. In one version of the script, Han Solo's reference to doing "the Kessel Run in less than 12 parsecs" was meant to be understood, by Obi-Wan and thus by the audience, as Solo revealing that he had no idea what he was talking about. There are at least two books where he does the Kessel Run. There's something very Trumpian about Lucas' approach, where he throws out a lot of random ideas and listens to the crowd to find out what they want to hear more of.

All of this was partly accidental, but it turns out to be a repeatable business model: make an entertaining story with lots of evocative details, so the fans who keep watching will keep wondering about them. And then, one game, novelization, spinoff, or theme park experience at a time, sell them an explanation that resolves that mystery but ideally adds a few more.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Who should I talk to about undersea mining? It’s a tricky space, but there are a few fewer legal obstacles than there used to be.

Diff Jobs

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

• YC-backed startup using AI to transform how companies quantify and optimize engineering productivity is hiring formidable full-stack and AI engineers. Experience with React + Typescript, Go, or Python on the ML side a plus. (SF)
• A multi-stage, fintech focused investment firm is looking for an investment associate to support thematic opportunity identification, diligence, and execution. Investing experience OR high-growth operating/investment banking/consulting experience and demonstrated interest in fintech required. (NYC, London)
• A hyper-growth startup (10x growth in 9 months) that’s turning customers’ sales and marketing data into revenue is looking for a head of deployments who is excited to work closely with customers to make the product work for them. Experience as a forward deployed engineer and leading enterprise deployments preferred. (SF, NYC)
• Well-funded, fast-moving team is looking for a full-stack engineer to help build the best AI powered video editor for marketers. Tackle advanced media pipelines, LLM applications, and more. TypeScript/React expertise required. (Austin, Remote)
• A Google Ventures-backed startup founded by SpaceX engineers that’s building data infrastructure and tooling for hardware companies is looking for a software engineering manager with 7+ years experience building large scale distributed systems. (LA, Hybrid)

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.