Tinder CEO, in an interview with Fast Company, spoke about a hidden service algorithm that reduces users to one league in terms of attractiveness.
At first glance, all Tinder users are on an equal footing, however, according to the head of the company, Sean Rad, this is not so. Immediately after registration, each person in the service is assigned a hidden factor called Elo rating.
This term is borrowed from the world of chess. Elo rating allows you to choose a player close to the level of the enemy. In Tinder, the mechanism of the same name performs a similar function – it reduces people whose “toughness” is approximately at the same level: such partners are more likely to continue the relationship after the signal about mutual likes.
According to Red, the Elo rating is not a level of external attractiveness, but a level of desire. It takes into account not only the user’s photos, but also other factors: for example, the average time that potential partners spend in his profile, as well as how “desirable” the people he likes most, and how often he does it. It took about two and a half months to write the algorithm, the latest version of which was launched in November 2015.
Tinder Vice President Jonathan Badeen (Jonathan Badeen) compares the resulting system with the selection of opponents in online games: the better results the user shows, the more serious rivals he comes across in matches.
According to the service engineers, each svayp to the left (the partner did not like it) and each svayp to the right (partner liked it) is a signal that means that the second partner is more like the user than the first.
At the same time, the rating cannot be considered universal. Some girls tend to put huskies on the profiles of young people with a mustache, and some, by contrast, prefer the lack of facial hair. The task of the mechanism is not only to calculate a certain average coefficient of desirability, but to try to match the desires of one partner with the desired qualities of another.
As Fast Company notes, unlike Uber, Tinder does not plan to make the rating public. The very essence of the service is to eliminate the possibility of being rejected: the user does not know how many people said “no” to him, and how many – “yes”.
Similarly, Tinder tried to make Tinder a more reliable tool in finding a partner and the users themselves. For example, in early 2015, Justin Long, a programmer based on a face recognition system, created a bot that automatically likes partners of a given type and starts chatting with them.