Why Pairwise Comparison?
Star ratings are easy, but they're noisy: people use different scales, ratings inflate over time, and averages hide real preferences. Rankl captures cleaner signal by asking one simple question: which is better?
We use the Elo rating system (originally developed for chess) to turn many matchups into a stable ordering.
The Elo Algorithm
Every item in a Rankl game starts with a rating of 1200. When you vote for one option over another, the winner gains points and the loser loses points. The magic is in how many points change hands.
The Expected Score Formula
EA = 1 / (1 + 10(RB - RA) / 400)
Where EA is the expected score, RA is the rating of player A, and RB is the rating of player B
The Rating Update Formula
R'A = RA + K × (SA - EA)
Where K is the adjustment factor (32), SA is the actual score (1 for win, 0 for loss), and EA is the expected score
The beauty of this system: if a highly-rated option beats a low-rated one, few points change hands (it was expected). But if an underdog wins, many points transfer. This creates more reliable rankings that help answer the questions that matter with less noise.
Self-Healing Rankings
Adaptive K-Factor
The K-factor determines how much ratings can change per vote. As items accumulate more votes, we can decrease the K-factor to stabilize rankings while still allowing gradual shifts. This helps mitigate the impact of malicious voting attempts.
Rate Limiting
We implement rate limits on each user to prevent vote manipulation. This ensures that no single actor can artificially inflate or deflate rankings, maintaining the integrity of the collective voice.
Perfect for Internet Voting
Pairwise Comparison Eliminates Choice Overload
Instead of ranking 20 options at once, voters simply pick between two. This reduces cognitive load and produces more thoughtful, accurate preferences.
Every Vote Updates Rankings Immediately
Every vote updates the ratings immediately. Upsets move scores more than expected wins, so informative matchups improve the ordering faster.
Harder to Game Than Polls
Because ratings are relative and self-correcting, it's far more resistant than traditional voting systems. You'd need to win against highly-rated opponents consistently—which only happens if you're actually great.
Our Goal
To create an open online ranking system that we can actually have confidence in. A place where the cream rises to the top, where collective wisdom prevails, and where you can help answer the questions that matter with less noise.