Novus Athenaeum

Novus Athenaeum is an early-stage concept for an adaptive learning system centered on scientific reasoning. Rather than organizing learning primarily around facts, grades, or a single ability score, it is designed around a repeatable inquiry loop: observe a phenomenon, ask a question, make a prediction, run a fair experiment, examine the evidence, revise an explanation, and apply the idea in a new setting.

The first planned experience uses changing shadows. A learner notices that a shadow changes, predicts what will happen when a light moves, helps decide what should remain fixed, tests the idea in a deterministic simulation, and compares the outcome with the original prediction. An incorrect prediction is not treated as failure: recognizing contradictory evidence and revising the explanation is itself valuable evidence of scientific reasoning.

My work so far has focused on defining the product, curriculum model, learner-evidence rules, safety boundaries, and candidate system architecture. The design separates authored curriculum and deterministic simulations from optional language-model assistance. The model may eventually help with bounded Socratic dialogue or classify responses into authored labels, but it is not allowed to choose the curriculum or assign mastery. The same learning flow must continue to work when the model is unavailable.

The project is currently in discovery and technical validation, not implementation. The repository contains requirements, a detailed changing-shadow vertical-slice specification, a candidate evidence-scoring model, API and data contracts, privacy constraints, test plans, and specialist review findings. There is not yet a runnable learner interface, deployed application, pilot result, or validated claim of educational effectiveness. The next stage is to prototype and test the shadow simulation, learner experience, evidence model, persistence guarantees, accessibility, and privacy controls before application development is treated as production-ready.