Philosophy
Ten principles we design against.
These are constraints, not slogans. Each one rules out product decisions that would be easy, profitable or conventional — and each one is meant to be checkable against what actually ships.
- 01Not a course marketplace
Nodes before courses
Knowledge is a graph of ideas, not a shelf of products.
A knowledge node is the first-class object. A "course" is only a temporary path between nodes — useful, but never the thing you own.
- 02No hours-watched metrics
Learning is structural change
We measure how your understanding is built, not how long you sat still.
Progress means clearer structure, better explanations, steadier reasoning and transfer to new problems — not watch time, check-ins or play counts.
- 03No certificate theatre
Capability through evidence
Ability comes from work you can point at.
Projects, tasks, traceable contributions and real feedback build a record. Unverifiable badges and stacked certificates do not.
- 04Not one-off transactions
The learning relation is the core
One durable relationship holds everything together.
Matching, goals, tasks, messages, feedback, progress and proof all organise around a single LearningRelation that evolves over time.
- 05Not a plain chat app
Messages are collaboration events
A message can be a goal, a task, a submission or a milestone.
Conversation is structured. A piece of evidence or a round of feedback is a first-class event in the thread, not a file lost in chat scrollback.
- 06No follower counts
Trust accumulates over time
Consistency earns credibility. Popularity does not.
Teacher standing grows from verified email, shown work, learner feedback and sustained contribution — never from follower counts or a black-box ranking.
- 07No endless feed
The interface projects structure
You should be able to see the shape of your learning.
Every screen makes the goal → task → evidence → feedback → change chain visible. The interface is a window onto structure, not an information feed.
- 08No streaks or dark patterns
Anti-addiction by design
We do not compete for your attention.
No infinite scroll, no streaks to protect, no anxiety-inducing red dots, no variable rewards, no mechanics that exist only to extend a session.
- 09No oracle answers
AI assists, never declares truth
If a machine suggests something, it shows its work.
Any AI may summarise, structure or explain — always with sources, stated confidence and your ability to correct it. The MVP ships no autonomous AI at all.
- 10No platform lock-in
Your data stays yours
Export it or delete it, whenever you want.
Settings give you a full JSON export and real account deletion. Recommendations explain themselves and can be turned off. No lock-in.
Consequences
What these principles cost us.
Principles only mean something when they rule things out. Here is what we gave up.
No engagement growth loop
Streaks, daily goals and push-to-return nudges reliably increase retention metrics. We do not use them, so our numbers will look worse than products that do.
No marketplace economics
Taking a cut of lesson payments is the obvious business model for a product like this. The MVP has no payments at all, which means no take rate and no transaction flywheel.
No verification badge to sell
We could run identity checks and charge teachers for a trust badge. Instead profiles are openly self-described, and we say so plainly wherever a profile appears.
No viral invitation mechanics
No contact-book uploads, no "people you may know", no referral rewards. Discovery is deliberately narrow: exact handles, invite links and genuine learning context.
On AI
Node Learn ships no autonomous AI in the MVP. If we add assistance later it will summarise, structure or explain — always showing its sources and its confidence, and always correctable by you. A model will never be positioned as the authority on whether you have understood something. That judgement belongs to you and your teacher.