← Back to incidents
Tinder's Hidden Elo Score Algorithm Exposed for Reinforcing Dating Biases
MediumFast Company revealed Tinder's secret Elo score system that ranked users by desirability, creating potential bias in dating matches. The algorithm was deprecated following public backlash.
Category
Bias
Industry
Technology
Status
Resolved
Date Occurred
Jan 1, 2016
Date Reported
Mar 15, 2019
Jurisdiction
US
AI Provider
Other/Unknown
Model
Elo Score Algorithm
Application Type
embedded
Harm Type
reputational
People Affected
50,000,000
Human Review in Place
No
Litigation Filed
No
algorithmic_biasdating_appselo_scoringtransparencydiscriminationsocial_platforms
Full Description
In March 2019, Fast Company published an investigation revealing that Tinder had been using a secret algorithmic scoring system called an 'Elo score' to rank users by desirability since approximately 2016. The Elo rating system, originally developed for chess rankings, assigned numerical scores to Tinder users based primarily on how many other users swiped right on their profiles. Users with higher scores were shown to other highly-rated users, while lower-scored users were primarily matched with others of similar ratings.
The investigation revealed concerning implications about how the algorithm functioned. Users who received more right swipes would see their Elo scores increase, making them more likely to be shown to other popular users. Conversely, users who received fewer right swipes would see their scores decrease, creating a feedback loop that could trap less popular users in a lower tier of the platform. This system effectively created a hierarchy of desirability that users were unaware of but significantly impacted their experience.
Researchers and critics identified potential for the Elo system to perpetuate and amplify existing societal biases around race, age, body type, and socioeconomic status. Studies of online dating had already documented patterns where users of certain racial backgrounds received disproportionately fewer matches. The Elo system could exacerbate these disparities by algorithmically relegating users who faced bias to lower tiers, reducing their visibility to the broader user base and potentially reinforcing discriminatory preferences.
Following the Fast Company investigation and subsequent criticism from users and advocacy groups, Tinder faced significant backlash over the lack of transparency in its matching system. The company initially defended the algorithm as improving user experience by showing people others at similar levels of engagement. However, critics argued that users had a right to know how their profiles were being ranked and that the system created unfair advantages for already privileged users.
Tinder eventually announced in March 2019 that it had deprecated the Elo scoring system in favor of a new algorithm that the company claimed was more focused on user activity patterns rather than popularity scores. The company stated that the new system considered factors like recency of app usage and mutual interests rather than pure desirability rankings. However, Tinder provided limited technical details about the replacement algorithm, and some users remained skeptical about whether the fundamental bias issues had been addressed.
Root Cause
Tinder implemented an Elo-based scoring system that assigned numerical desirability ratings to users based on right-swipe rates and other engagement metrics. The algorithm perpetuated existing societal biases by creating feedback loops where users deemed less desirable received fewer matches, further lowering their scores.
Mitigation Analysis
Regular algorithmic auditing for disparate impact across demographic groups could have identified bias patterns. A/B testing with alternative matching algorithms focused on compatibility rather than popularity scores would have provided fairer alternatives. Transparency about matching methodology and user controls over matching preferences could have reduced harm and improved user agency.
Lessons Learned
The incident highlighted the importance of algorithmic transparency in platforms that significantly impact users' social and romantic lives. Hidden scoring systems can perpetuate societal biases at scale, and users deserve to understand how algorithms affect their experiences on platforms they rely on for meaningful connections.
Sources
Tinder's Secret 'Desirability Score' Revealed
Fast Company · Mar 15, 2019 · news
Tinder's algorithm rewards you for being picky — but only if you're a man
Vox · Mar 15, 2019 · news
Powering Tinder: The Method Behind Our Matching
Tinder · Mar 15, 2019 · company statement