Ask HN: Selling software to company I work for as an employee
44 points by apohak 3d ago 51 comments
Ask HN: Go deep into AI/LLMs or just use them as tools?
160 points by pella_may 20h ago 126 comments
Ask HN: What's your favorite architect/editor pair with Aider?
13 points by 34679 13h ago 1 comments
Bloom's two sigma problem (2020)
8 Tomte 1 5/24/2025, 7:25:35 PM nintil.com ↗
I've been working on an implementation of mastery learning and other related techniques called Trane (https://github.com/trane-project/trane/) for the past three years or so. Mastery learning is the main one, but it also integrates spaced repetition, interleaving, mixing difficulties, and reward propagation (doing well or bad in an exercise affects how related exercises are scheduled).
I think it works pretty well, but you need to pair it with proper pedagogy of the skill you want to learn and the proper curriculum. The latter is the hardest part, so it's being my main limitation. I've used some external resources to build courses, and they work well, but obviously it would work much better with a full curriculum built from the ground up.
Currently working on Pictures Are For Babies (https://picturesareforbabies.com/), which is meant to do just that for literacy. I am hoping to do a first release soon. As for the motivation angle, the solution in this particular instance is fairly simple. Use the software to enforce scheduling andpedagogy,y and a human tutor to provide emotional and social support. This division allows any literate person to become an effective tutor with a few hours of training.
I am hoping that the average student can complete the whole curriculum in five years. That would mean that (assuming they start at between 4 and 5 years old), the average student would have college-level reading and writing skills by the time they are nine or ten.
Most complete explanation so far is in the pedagogy page: https://picturesareforbabies.com/home/pedagogy/