An interactive educational game that teaches you how ranking algorithms work by challenging you to rank posts better than raw percentage-based sorting. Learn why statistical confidence matters more than simple approval rates.

This game demonstrates why the Wilson Score Interval algorithm is superior to naive percentage-based ranking for user-generated content. Through hands-on examples, you'll discover:

• Why an 89% approval rating with 200 votes beats a 95% rating with 20 votes
• How small sample sizes make perfect scores unreliable
• Why bigger sample sizes provide more reliable rankings
• How statistical confidence intervals improve content ranking

1. Read the Challenge: Each round presents up to 3 posts with their upvotes and downvotes
2. Think Like a Ranking Algorithm: Consider both approval percentages AND sample sizes
3. Drag to Rank: Arrange posts from highest-ranked (top) to lowest-ranked (bottom)
4. Submit Your Ranking: See how well you did compared to the Wilson Score algorithm
5. Learn from Results: Study the key insight explaining why the correct ranking works

Traditional ranking uses simple percentages:

19/20 = 95% 👍 (but only 20 votes!) vs 178/200 = 89% 👍 (200 votes!)

Wilson Score considers statistical confidence:

95% with 20 votes = uncertainty 🤔
89% with 200 votes = very confident 📈

Higher sample sizes reduce uncertainty, making rankings more reliable.

• Interactive Learning: Drag-and-drop interface with immediate feedback
• Curated Examples: Real-world scenarios that illustrate ranking algorithm concepts
• Score Tracking: See how your intuition compares to mathematical algorithms
• Educational Insights: Each example includes specific lessons about ranking
• Responsive Design: Works on desktop and mobile devices
• Accessible: Keyboard navigation and screen reader support

The Wilson Score interval calculates a confidence interval for a Bernoulli parameter (like approval ratings). The score represents the lower bound of this interval, which accounts for:

1. Sample Size: More votes = higher confidence
2. Approval Rate: Higher percentage = higher score
3. Uncertainty: Small samples get penalized for statistical uncertainty

Wilson Score = (p̂ + z²/(2n) - z√(p̂(1-p̂)/n + z²/(4n²))) / (1 + z²/n)

Where:

• p̂ = upvotes/(upvotes+downvotes)
• n = total votes
• z = 1.96 (for 95% confidence interval)

• Original Wilson Score Paper
• Reddit's Ranking Algorithm
• Understanding Confidence Intervals

Challenge your intuition. Master ranking algorithms. Build better feeds. ✨