Vibe Courses is a library of short, free lessons that stop at exactly that moment. Each one is a small app built around a single hard idea — you drag it, predict it, break it, and the understanding comes from doing that.
Both come straight out of lessons in the library: a figure that answers back, then one question about what you just saw.
Is college BPM above some cut? Move the cut. Everyone left of the line gets one average, everyone right gets another — and the error they leave behind falls.
Move the cut to the far edge and one side holds a single prospect, whose own average fits him perfectly. Why do trees usually block that?
A perfect fit on one prospect is memorising, not learning. That is exactly what the minimum-per-leaf rule prevents.
The error does fall — that is the trap. A group of one fits his luck exactly, and luck does not repeat on next year’s class.
For one prospect the model has never seen, each dot is one tree’s guess. The forest predicts their average. Move across the figure to add trees.
The forest beats even its best single tree. Why not just find that best tree and keep it?
Being best on this class is partly luck. The average does not depend on picking a winner — which is why it holds up on prospects it has never seen.
Which tree wins depends on this particular class. On next year’s prospects a different tree would win — the average is the safe bet.
Two steps from two lessons. The rest of the library works exactly like this.
Each entry is its own small app, not a chapter. Move the figure on a card to see what you would be doing.