# The Cambridge Chronicle > Source: https://karte.cc/pg/newspaper Saturday, March 28, 2026 — A generated newspaper profile for Paul Graham. ![Paul Graham](https://unavatar.io/x/paulg) ## Front Page ### LISP PIONEER TURNS ESSAYIST TURNS STARTUP ORACLE: PAUL GRAHAM STILL WRITING THE FUTURE From Viaweb’s $50M Yahoo! exit to Y Combinator’s 5,000 startups, the polymath’s fingerprints are everywhere Paul Graham, the 62-year-old programmer-writer-investor whose fingerprints are on some of the most consequential technology companies of the 21st century, is spending 2026 rewriting the rules of work itself. His latest essays—*Founder Mode* (2024) and *How to Do Great Work* (2023)—have become lodestars for founders and artists alike, distilled from decades of building, funding, and writing. "The companies that have stayed great were run by founders who never fully delegated," he observes in *Founder Mode*, a phrase now echoing in Silicon Valley boardrooms and Berlin co-working spaces alike. > The single trait that separates founders who make it from founders who don’t is being relentlessly resourceful—not smart, not visionary. ### Y Combinator’s Arc still powers Hacker News after 20 years The Lisp dialect Graham co-designed in 2001 remains the invisible engine behind one of the internet’s most influential forums, a testament to his belief that "the web got worse" while Hacker News stayed fast, clean, and human-moderated. ### Founder Mode debate splits Silicon Valley: ‘Hire good people’ is dead, says Graham In a 2024 essay that went viral, Graham argued that the conventional wisdom of delegation fails at scale. His prescription—founders must stay deeply involved—has sparked both admiration and backlash among venture capitalists and operators. - Born 1964 in Weymouth, England; raised in Pittsburgh - Cornell undergrad (philosophy + computer science), Harvard PhD in CS under Guy Steele - Co-founded Viaweb in 1995; sold to Yahoo! for ~$50M in 1998 - Still writes essays at paulgraham.com; new pieces appear when an idea is actually new Mood: Cambridge fog lifting to reveal a skyline of Lisp parentheses Advertisement: WANT TO WRITE LIKE GRAHAM? Learn Lisp in 30 days or your money back. (Disclaimer: No money will be refunded.) ## Features ### THE GRAHAM METHOD: HOW TO DO GREAT WORK (AND WHY IT’S GETTING HARDER) A 2023 essay becomes a quiet manifesto for the anti-algorithm age Graham’s *How to Do Great Work* is less a how-to than a diagnostic tool for modern life. It begins with a simple premise: great work is not about talent or luck, but about choosing a field where you have natural aptitude, staying in it longer than is reasonable, and noticing what others miss. "Bus ticket collectors are not eccentrics; they're prototypes," he writes, arguing that obsession is the only reliable predictor of future excellence. > Choose a field you have a natural aptitude for, that interests you most, and that offers the most scope to do great work. Then do something hard. ### Profile: The Lisp partisan who built the web’s most durable forum Hacker News, Graham’s 2007 experiment in minimalist social news, remains one of the internet’s last bastions of slow, thoughtful discourse. It runs on Arc, a Lisp dialect he designed with Robert Morris, and is moderated by a tiny team whose human touch keeps the signal-to-noise ratio alive in an era of algorithmic outrage. ### Behind the scenes: Graham’s editing ritual—rewriting essays five times Graham’s essays begin as ‘bad first drafts’ that get rewritten up to five times. He reads them aloud to find sentences that aren’t really sentences. The result: essays that feel conversational but are honed to surgical precision. - *How to Do Great Work* is his longest essay, distilled from decades of observation - Hacker News still loads in 200ms because it was never updated - Graham’s first drafts are ‘usually wrong’; fifth drafts are ‘usually shippable’ - He reads every essay aloud to catch unnatural phrasing Mood: A quiet study at dusk, the hum of a Lisp REPL in the background Advertisement: ESCAPE THE ALGORITHM: Read Hacker News the way Paul Graham intended. (No ads. No trackers. Just you and the parentheses.) ## Opinion & Letters ### WHY AI WRAPPERS WILL FAIL (AND WHAT WILL SURVIVE) The startup oracle on the next wave of durable companies Most AI startups today are ‘wrappers’—thin layers over commoditized models that will soon be free. The survivors won’t be the ones with the best prompts, but the ones solving a specific, painful problem with care. "The model layer will commoditize; the product layer is where the durable companies will live," Graham argues. "This is true for every prior wave of platform shifts too—mobile, web, cloud." > The interesting question isn’t whether AI replaces programmers—it’s what the new bar for great programming looks like once the trivially-typable stuff is free. ### A reader writes: "Graham’s ‘Founder Mode’ is just common sense dressed as heresy" The letter argues that Graham’s critique of delegation is obvious to anyone who’s ever worked in a real company. Graham’s response: "The advice is common sense. The mistake is assuming that common sense is common." ### Counterpoint: "Founder Mode’ romanticizes founder tyranny" Critics argue that Graham’s model risks glorifying micromanagement and founder cults. The counterpoint concedes that Graham’s insights are valuable but warns against conflating founder involvement with founder control. - Graham funded ~5,000 startups at YC, including Airbnb, Stripe, Dropbox, Reddit, Coinbase - He stopped actively picking startups in 2024; most of his time is now spent writing - Sam Altman took over YC in 2014; Graham calls him ‘a better operator than I am’ - Graham’s Viaweb playbook—small checks, 3-month programs, demo days—has barely changed in 20 years Mood: A lively debate in a Cambridge café, espresso shots flying like parentheses Advertisement: THINK YOU’RE RELENTLESSLY RESOURCEFUL? Try our 30-day challenge: build something no one asked for. (Spoiler: No one will.)