January 4, 2026
Messy jobs, messier comments
A New Year's letter to a young person
AI vs Your First Job: Twitter drama, big-city dreams, and "just get hired" energy
TLDR: Garicano urges young people to choose “messy” work humans still excel at as AI takes over simple tasks. Commenters mocked the “install Twitter” tip, split over big-city hustle versus second‑tier happiness, and noted early careers are about “who will hire me,” not theory—because your first choice matters.
Economist Luis Garicano’s New Year letter says: skip easy, single-task jobs AI will swallow and chase the messy many-hat roles humans still run. He pushes hubs like San Francisco or Paris and—controversially—“install Twitter.” The comments pounced. ahel roasted the Twitter tip as “doomscrolling as career advice.” dzonga warned that “go to the hubs” breeds echo chambers, while zelda420 claimed second‑tier cities bring happier lives: homes, families, and quirky hobbies instead of endless grind. Suddenly a career guide turned into a lifestyle cage match.
Practical voices cut through the glow-up talk: astura waved off “pick where you’ll learn most,” saying early on the only metric is “who will hire me.” RevEng dunked on the old promise that tech gives us more free time—“that never happens.” Still, many nodded at the core point: AI keeps eating simple tasks, pushing junior roles toward zero value, as Brynjolfsson, Li; Ramond found in customer support. But others noted laws and cautious bosses can slow the robots. The crowd’s verdict: pick messy, human work—maybe skip the bird app, don’t worship one zip code, and for your first gig? Say yes to whoever says yes.
Key Points
- •Knowledge work varies along a spectrum of “messiness,” with single-task roles being more vulnerable to AI automation than complex, multi-task roles.
- •AI currently requires human oversight due to error rates, but those rates are decreasing, especially in well-defined tasks.
- •Fields differ in error tolerance; some accept higher risks while others (e.g., diagnostics, corporate communications) are highly risk-averse.
- •As AI improves and legal requirements for human oversight lessen, unsupervised AI may displace junior roles (e.g., producing finished code), driving down prices.
- •Adoption is shaped by firms and governments; examples include Uber bans in parts of Europe and legal protections for notaries, which can delay displacement even in single-task roles.