Where it goes wrong
Traditional teaching
Counting problems about dice, cards, and marbles: list outcomes, form the fraction, simplify. Correct, and strangely disconnected from the actual point — that probability is how you reason when you can't know what happens next.
The reteach
The Alice Method
Start with the deep idea: randomness is unpredictable up close and reliable from far away. Flip real coins — ten flips give chaos, three hundred give ~50%, and streaks of five heads happen naturally (this single demo inoculates against the gambler's fallacy). Then counting outcomes becomes the tool for predicting long-run behavior, not an end in itself. Expected value turns 'is this game fair?' into a computation a twelve-year-old can own.
Mental model
The long run
Probability is a statement about what happens over many trials, not a prophecy about the next one. A 70% chance of rain that stays dry wasn't 'wrong.' The question is always: if this situation repeated a thousand times, what fraction goes each way?
Transfer
Where this shows up for the rest of their life
This is the reasoning of medicine, investing, insurance, weather, and AI. A child who thinks in long runs evaluates risks ('rare but catastrophic vs. common but mild'), reads statistics in the news skeptically, and understands why casinos always win — literacy that most adults never acquire.
Watch for these
Common misconceptions
The misconception
“After five heads, tails is 'due.'”
The repair
The coin has no memory. Long-run balance comes from dilution by future flips, not correction of past ones. Watching a live simulation makes this visceral in a way lectures never do.
The misconception
“Unpredictable means unknowable.”
The repair
We can't call one flip, but we can say precise things about a thousand. That paradox — individual chaos, aggregate order — is the whole subject, and kids find it genuinely thrilling.
Seen, not said
The visual explanation
A live running-average chart: the proportion of heads swinging wildly for early flips, then hugging the 50% line as trials accumulate. The law of large numbers, watched in real time.
Try it
Interactive example
Flip 10, then 100, then 100 more. Watch chaos organize itself.
Probability is long-run behavior, not fortune telling
Ten flips are wild. Three hundred flips settle near 50%. Randomness is messy up close and orderly from far away.