The Unfalsifiability of Security Claims
summary
Argues that claims of necessary conditions are unfalsifiable: you can observe insecurity but never prove a system will never fail, so design rules become tautologies. Uses passwords and masking to show implicit assumptions (leaked hash, plaintext storage, attacker access) determine whether a countermeasure helps. Shows costly overkill persists because self-correction is one-sided and recommends framing fixes as conditional statements (outcome better given a b c d), estimating plausibility, and ask what would convince you you are wrong so measures become testable.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:13The Unfalsifiability of Security Claims
- 2:01Jacob Nielsen
- 4:34A Secure System Must Defend against all Possible Attacks Including those Unknown to the Defender
- 12:40How Do I Check that the Design Goals That I Built In Are Actually Necessary
- 15:11Fallacy of Denying the Antecedent
- 15:54Is Password Reuse the Real Threat Factor
- 21:32Unfalsifiable Claim
- 50:58Conclusions
- 1:07:01Safety Analogy
related talks
58mDefending Security is Probabilistic, Not Deterministic: Get Over It
51mNavigating Faulty Assumptions and Magical Thinking in Cybersecurity
29mSecurity by(e) Design?
48mKeynote Presentation: Escaping Cantor's Find-Fix Cycle
16mDesigning Holistic Security Programs
18mDesigning for Security Outcomes
38mDo You Feel Secure?
53mOptimistic Security