Entry 226-2 Build in Public 2 min ↩ back to the timeline

The curriculum that could teach Python

Building the curriculum surfaces an architecture confession and a moat claim. The confession: he wanted Severo to stay a simple structured chatbot, but using it for French proved that without a guide nothing sticks, so 'la complejificación es inevitable'; the consolation is that starting simple distilled the core worth complexifying. The claim: Gemini reviews the finished curriculum and observes it would work for any subject, Python included, echoing the old Pipo dream, and the moat argument writes itself: Duolingo could copy this, but only by restructuring a rigid product under millions of users who would riot.

video fuente → Source video thumbnail
Source transmission · “0 to 1 Million” diary

// trace: where this idea came from

The universal curriculum starts taking real shape, exercises sliced by CEFR level and by skill ▸ 2:07, with pronunciation deliberately parked: Gemini still congratulates you for saying anything ▸ 2:31, Azure’s recognizer covers only a handful of languages, and the standing rule is nothing that doesn’t work for all of them ▸ 3:04. Julia banks a future unit for the shelf (seed 226-1): regionalisms, a year or two out ▸ 4:06.

The confession is architectural. He wanted Severo simple, a structured chatbot and nothing more, but being his own user broke the wish: learning French inside it, things half-stick and there is no north to hang them on ▸ 5:04, so “la complejificación es inevitable” ▸ 6:35. The consolation clause matters as much: starting simple wasn’t a waste, it distilled what deserves to be complexified, “las bases ya están” ▸ 6:48, the same reason he waves off the socio’s request for mockups first, move the pieces, feel it, then decide ▸ 7:48.

la complejificación es inevitable; la base ya está destilada →

Then the reviewer’s aside that reframes the company: Gemini, reading the finished curriculum, notes it would work for any subject, Python included ▸ 8:35. Julia hears the echo at once, “me acuerda un poco a Pipo” ▸ 9:04, geography, history, math on the same engine. And the moat argument closes it: Duolingo gestures at the same territory, but its architecture is rigid, copying this would mean restructuring almost everything under users who would riot at the change, “ellos pueden correr menos rápido que nosotros” ▸ 10:06. The small team’s advantage isn’t speed of typing. It’s having nothing installed that resists being rebuilt…

// continued in

no entry has continued this idea yet: the arc is still open

la lista es el activo →

Continues

One distilled teaching per entry, straight from the video diary. Leave your email and get the next one the day it lands.