Show HN: How Random Are You?
This interactive 'Show HN' challenges users to pick random numbers, only to have an algorithm predict their choices with unsettling accuracy. It's a clever and engaging demonstration of how profoundly non-random human behavior truly is, and why it's so hard for us to simulate true randomness.
The Lowdown
Ever wonder how random you really are? This interactive web experiment, 'How Random Are You?', puts your internal random number generator to the test. Users are asked to vocalize numbers from 1 to 9 as randomly as possible, while a sophisticated 'house' algorithm attempts to predict their every pick.
- The Challenge: Users continuously vocalize numbers from 1 to 9, aiming for randomness.
- The Predictor: A 'house' algorithm employs ten small predictors simultaneously, trusting whichever has been most accurate recently.
- Adaptive Learning: The system learns from the user's patterns, with a warm-up period of 10 numbers to understand individual tendencies.
- Cognitive Bias: A primary predictor identifies numbers the user hasn't said recently, capitalizing on the human tendency to 'even out' their choices rather than picking truly independently.
- Psychological Basis: The house gets a head start from a psychology study involving 418 people who each picked 200 numbers, comparing user performance against this baseline.
Ultimately, the experiment reveals the inherent biases in human attempts at randomness, showcasing how predictable our 'unpredictable' choices can often be.