People of Tinder dataset

People of Tinder is a dataset created in 2016 by software engineer Stuart Colianni that was intended 'to build a better, larger facial dataset' capable of distinguishing between male and female images.

The dataset comprised 40,000 images of people's faces - half women, half men - from the San Francisco area scraped from dating app Tinder.

Risks and harms 🛑

The People of Tinder dataset raised serious ethical and privacy concerns by scraping and sharing personal photos and information from Tinder profiles without users' consent, potentially exposing individuals to harassment, identity theft, and the unintended analysis of their personal data and dating preferences. 

Transparency and accountability 🙈

The People of Tinder dataset has several transparency limitations:

• Lack of consent. The dataset was created by scraping Tinder profiles without obtaining consent from the users whose data was collected, raising significant ethical and privacy concerns.

• Data collection. The methods used to gather the data, including the specifics of the scraping process and the criteria for selecting profiles, are not fully transparent.

• Dataset creation. There is a lack of comprehensive documentation detailing the dataset's creation, including the exact data points collected and the potential biases introduced during the data collection process.

• Potential for bias. The dataset may not represent a diverse range of Tinder users, as it is unclear how profiles were selected and whether the dataset includes a balanced demographic representation.

• Restricted access. The dataset is not openly shared, limiting the ability of researchers to independently verify or analyse its contents and further reducing transparency.