Summary: A probability estimate should move when new evidence arrives: the prior becomes a posterior. Every test also has a confusion matrix (false positives and negatives), so a negative test isn’t certainty. In medicine this reasoning is often missing: risk is quoted from population priors while evidence already in hand is ignored.
Key ideas
- Genetic-counseling example: carrier screening usually tests only one partner. When a relative’s partner tested negative but their child was born deaf, the counselor discussed only the prior risk. She ignored that a hearing first child itself lowers the chance the mother carries the gene; twenty hearing children would lower it substantially (explainable—152-biohacking-like-a-data-scientist).
- Every diagnostic test has false positives and false negatives; treat results as evidence that updates a probability, not as verdicts (explainable—152-biohacking-like-a-data-scientist).
Disagreements & open questions
Takeaways
- When given a medical risk number, ask whether it accounts for evidence you already have (prior results, family outcomes) (explainable—152-biohacking-like-a-data-scientist)