Stripe CEO Patrick Collison: Classics and Common Contexts Will Make a Comeback
Patrick Collison shared an article by Alec Thompson in Asterisk magazine titled "The Destruction of Scottish Classics," stating that discussions about classics, culture, and common contexts will return to the forefront in the coming years. The reason is that many narrowly defined skills are being handed over to artificial intelligence, the extreme de-Arnoldization, and the educational processes described by Hannah Arendt are beginning to recede. He wonders about the causality between PISA scores and the reforms mentioned in the article, and claims that it is an exciting time to engage in educational theory, with the best insights often coming from unexpected places.
Thompson writes about how Scotland dismantled its once world-leading secondary school curriculum around 2011, replacing content knowledge with skills-based teaching. Open-ended exams, such as higher-level essays, were removed due to their difficulty and excessive openness. After the reforms, reading, writing, and math abilities significantly declined, and socioeconomic gaps did not narrow, but even expanded within Scotland. Scotland's PISA scores have been in continuous decline since the 2000s; from 2012 to 2022, math scores equated to a loss of about 16 months of learning, and science scores to about 18 months. England's scores remained relatively stable, while Scottish 15-year-olds have clearly fallen behind, despite higher per-student funding. The Curriculum for Excellence was implemented around 2010, and the decline began to become systematic around 2012.
Arendt wrote in "The Crisis in Education" that education is a political order issue: adults must pass on the existing world to the next generation, rather than first dismantling it into skill modules. Matthew Arnold's tradition emphasizes common culture as the material for judgment. Collison describes AI taking over narrow skills as a reverse pressure: after skills are taken by models, if schools abandon shared texts, the common context will become thinner. He has long advocated for a reading list in Silicon Valley, viewing "classics" as an inheritable context for the tech circle, rather than mere decoration.
The causality is still not nailed down. PISA is a cross-national snapshot, with curriculum reforms, teacher training, testing systems, and the pandemic overlapping. Collison sees the unresolved causality itself as a theoretical opportunity: those who can distinguish between reforms and scores will be writing the next round of educational design specifications.
In market mechanisms, attention is shifting from skill products to content products. Buyers are schools, publishers, and foundations defining "what is still worth teaching" in the model era; sellers are designers of decade-long skill courses. Beneficiaries are compilers of classic texts, core knowledge courses, and interdisciplinary old curricula; those under pressure are reform bureaucrats and educational companies that write transferable skills as the sole goal. Funding will not immediately leave online practice platforms, but elite employers and university admissions will re-weight common reading. Pricing occurs in course catalogs, not in app stores.
Source: Public Information
ABAB AI Insight
Collison interprets the Scottish curriculum reform as a controlled experiment: dismantling the classics while retaining skills has not improved scores or equality. Stripe sells a common protocol for global payments; he applies the same preference to education—without a shared context, collaboration costs rise. After AI takes over narrow skills, even the best prompts cannot replace the shared reading that the "half-sentence casually mentioned in class three weeks ago" relied on. The de-Arnoldization once framed classics as exclusionary devices; he now frames classics as layers of social memory that models cannot generate.
The capital path is shifting from skill software back to text compilation. Edtech charges by the number of exercises, while classics are preserved by bibliographies and oral tradition from teachers. Scientific funding initiatives like Arc and Fast Grants, in which Collison is involved, tend to bet on content infrastructure when evidence is lacking. The motivation is to prevent the next generation from having only tools and no quotable sentences. Resources are shifting from curriculum consulting firms to educational researchers capable of writing causal identifications—his point that the best theories often come from unexpected places implies not leaving orders to existing education reform consultants.
A similar structure is seen in England's insistence on a clearer knowledge curriculum while PISA remains relatively stable, and the back-and-forth after the Common Core in the U.S. The skills movement promised to narrow gaps, but Scottish data shows the opposite direction: when high-status students drop faster, the gap may temporarily "narrow," only for lower-status students to follow suit. The industry phase belongs to control: models will take exams, and schools must decide what human memory must retain.
Structural judgment indicates that technological replacement forces a reconstruction of content. The mechanism is that once skills can be outsourced, the residual value of education is reduced to a common world that cannot be prompted. Without classics, citizens only share model outputs; with classics, models have a human context to align with. PISA is a delayed instrument, while curricula are the switches that should have been activated a decade earlier.
ABAB News · Cognitive Laws
Once skills are taken by models, what remains for schools is common memory.
Dismantling difficult-to-change exams does not equate to dismantling difficult-to-teach gaps.
Classics are not decoration; they are compressed packages that existed before collaboration.