Where quotes go to die

Koby’s launch post hides a five-year archaeology, told tonight as origin story, and every layer is a failed system with the same cause of death.

Layer one: physical books, phrases typed by hand into phone notes mid-reading, stop, type, resume ▸ 2:39. Those quotes fed his old Instagram captions; the friction eventually killed the habit and, with it, physical books entirely. Layer two, circa 2020: photographing the Kindle’s screen and pushing the photos through Google Translate’s image recognition to extract text ▸ 3:52, OCR by bricolage, years before ChatGPT existed. “Llega un punto en donde no aguanta.” Layer three, last November: a proper Python extractor, 42 books, 1,026 highlights, published as a GitHub wiki. Its epitaph is the entry’s teaching: “como que nunca lo volví a abrir… realmente no sirvió de nada lo que hice” ▸ 6:14.

el archivo perfecto que no visitas vale cero →

Notice the wiki didn’t fail at capture, the data was all there, clean, organized. It failed at the requirement he only articulated after burying it: “la gracia es tener las frases disponibles y verlas frecuentemente” ▸ 4:46, quotes as ambient companions, some of them literal decision-making compasses (entry 75-1), not as a warehouse to visit. Nobody re-visits a warehouse. Availability beats archival: a knowledge tool succeeds not when it stores well but when it resurfaces well.

Which explains every Koby design choice that might look decorative: the landing page that is a feed of random quotes, the shuffle filter, the dark mode for bedside reading, the public URLs that let a quote travel. Storage was solved in layer three. Koby is what the five-year autopsy actually asked for: a place where the sentences keep walking past you…