An Unscientific Investigation of Tinder’s Algorithm
Do you ever wonder if you’re using Tinder more than you would like or need to? Well, I do. Using machine learning, I decided to explore whether a recommendation algorithm could learn my swiping preferences and whether dark patterns may be hidden in Tinder’s design.
_________________________________________________________________
PSA: It is a key value for us at AIxDESIGN to open-source our work and research. The forced paywalls here have led us to stop using Medium so while you can still read the article below, future writings & resources will be published on other platforms. Learn more at aixdesign.co or come hang with us on any of our other channels. Hope to see you there 👋
Swiping cramps
I’ve been using Tinder for the last 2 years and since my first swipes, I’ve had a hunch that Tinder loads the deck against me to keep me on the app for longer. I promise I’m not salty about not getting enough matches, I just feel I could be getting these matches with less swiping. Currently, I average 9 left swipes (rejection) to 1 right swipe (hopeful acceptance), and would argue I don’t have a specific “type”.
However, a definite side effect of an app that shows you the world and values quantity over quality, is that you develop some conscious and unconscious bias on which way you’re swiping. Whether that would be profiles with sporty photos that irrationally make me feel lazy or profiles with bios that are only used to plug Instagram profiles, there are certain things I always half-consciously swipe left on. All this may sound shallow to non-Tinder users, but online daters will all agree this is the reality of dating in 2020.
So I asked myself:
Why is it that, in an age where Spotify can accurately create playlists based on my music taste and Youtube is able to feed me video after video I will passively consume, Tinder only shows me profiles I want to swipe right on 10% of the time?
It’s important to say here that Tinder does use an algorithm to help curate peoples experience on the app. They claim it helps people match more frequently, get off the app and meet IRL. However, they are very vague on how this algorithm actually works, but they do say that the more you use the app, the more matches you get. Who would’ve guessed?
So, if Tinder has an algorithm that is so focused around matches, why does it show me profiles I don’t want to match with 90% of the time. Is Tinder’s algorithm just bad? Am I more random and spontaneous than I think? Or am I just weirdly picky and hard to please?
Recommendations matter
My hypothesis/conspiracy theory is that Tinder knows who I am likely to swipe right on and rations out those profiles, and in an attempt to gamify the experience, keep me swiping for longer and hook me to their app. A tactic akin to slot machines, video game loot boxes called “variable rate reinforcement”, and one of the widely recognized dark design patterns.
“The player is basically working for reward by making a series of responses, but the rewards are delivered unpredictably… Dopamine cells are most active when there is maximum uncertainty, and the dopamine system responds more to an uncertain reward than the same reward delivered on a predictable basis.”
Dr Luke Clark, director at the Center for Gambling Research at the University of British Columbia
Methods of investigation
To test this hypothesis, I decided to assess whether there was a distinct pattern of my “Matches” compared to my “Left Swipes”.
To work out these patterns, I trained various machine learning models with data from both my “Matches” and random “Left Swipes”. I split them into Pictures, Bios, Music and in the end, the full profile.
Consequently, I tested the generated results made by these models side by side, choosing my favourite of the two options presented each time. If I choose the result generated by the model trained on my “Matches” more often, then it would prove my match preferences were learnable by an AI and hence, Tinder could consistently fill my feed more with people I would like to swipe right on.
Generating (Frankenstein’s) Profile Pictures
Profile pictures are a key part of the Tinder experience — it’s the first thing you see and I would be lying if I didn’t say it’s probably the most important aspect of a profile. Tinder evidently thinks so too, as throughout the apps short history it has continued to increase the size of the profile picture in its UI.
So to generate fake profile pictures, I trained a Stylegan2 Model on RunwayML with 212 images of the faces of my “Left Swipes” and a separate model on 212 “Match” faces, both for 5000 steps.
Due to my small data set, the resulting generated images were very abstract. However, going into the test I was pretty confident I would be easily able to discern which generated images were based on “The Perfect Match” model, based on the colour, composition and frankensteined facial features.
This was not the case — during the test, I only selected “The Perfect Match”-generated picture 5 out of 10 times. Not too promising.
Tool used: https://runwayml.com/
Generating Profile Bios
Bios are a variable part of a Tinder profile: some users writing only their height, others a thesis on their perfect partner and a good proportion of users have no bio at all. For me, a bio is an important part of my swiping decision, but no bio is definitely better than a bad bio. Hence, out of my 212 matches, only 116 had bios. I also found 116 bios from random people I would normally swipe left on. These were used to train the Tensor flow-based Textgenrnn word-level model for 100 steps.
Again, as in the previous step, the generated results were very abstract but definitely recognisable as Tinder bios, even if they made no sense.
During the test, I went for “The Perfect Match”-generated bio 7 out of 10 times. However, 2 were easy to spot because I clearly recognised some generated text from the datasets, so the real result was 5 out of 8.
Tool used: https://colab.research.google.com/
Generating “Our Song”
The final piece of a Tinder profile is music, as users have the option to add an Anthem song, as well as a feed of their current most-listened artists. For my test, I focused solely on the user’s Anthems as these are more common and they would give me cleaner data points. For me, these anthems are quite important and help to add detail to my picture of someone.
Through brute force, I compiled a playlist on Spotify of the 99 songs from my “Matches” and 99 others from random “Left Swipes”. I then used Spotify’s similar playlist feature via Skiley to generate new playlists. I was worried that the diverse data sets used would lead to very random playlists but the resulting playlist felt very close to the input data.
During the test, I was shown the song name, artist and album artwork but not allowed to listen to the song because the majority of the time this is how I engage with Tinder profile Anthems.
As a result, I picked the “The Perfect Match”-generated song 6 out of 10 times.
Tool used: https://skiley.net/
The Final Test — Comparing fully generated profiles
For the last part of the test, I compared full profiles of generated photos, bio and anthems. This was most true to life of the test, as the combination of all the different points gave me a really clear picture of a profile. As such, I managed to pick “The Perfect Match” generated profile 8 out of 10 times. However, because of the bios, I again clearly recognised 2 of the generated bios from the data sets — the real result was 6 out of 8.
Conclusion
In the end, I managed to correctly pick “The Perfect Match” generated profile 61.1% of the time. This was all done with very crude models (except Spotify) and tiny data sets. Whereas Tinder would have access to more sophisticated models, a lot more data and real data scientists. Hence I would tentatively say my hypothesis is correct. If Tinder wanted to, I believe they could learn my preferences and show me profiles I would match with at a much more consistent rate.
There are many reasons why I think they choose not to implement a more accurate and user-focused algorithm — perhaps it would ruin the fun of the app or maybe people wouldn’t opt for subscription plans if they found matches so easily. Cynically, I believe they tune their algorithm to balance out the outcomes of Matches, Fun and Profit. They are a company after all, not a benevolent cupid armed with digital arrows.
Closing Thoughts
At the end of the day, I feel my results were quite obvious and entirely predictable. However, I feel that fact I was able to create this experiment and put Tinder on trial with no Machine Learning and very little coding experience, tells us of a new paradigm emerging. One where companies or governments can no longer hide behind or outright blame their algorithms for anti-user features or more importantly discrimination. People (Companies or Governments) make choices about what they want their algorithm to do, and now with new easy to use the software we can check that we agree with their choices.
On a personal level, what I am also taking away from this experiment is a greater understanding of Machine Learning but most importantly a deeper awareness and questioning of my conscious and unconscious biases.
Meet the author
Fred Wordie is a critical designer based in Berlin. His work uses design fiction to explore how we as a society engage with technology and social issues. His work can be found here.
About AIxDesign
AIxDesign is a place to unite practitioners and evolve practices at the intersection of AI/ML and design. Currently, we are organizing monthly virtual events, sharing content, exploring collaborative projects, and looking to developing fruitful partnerships.
To stay in the loop, follow us on Instagram, Linkedin, or subscribe to our monthly newsletter to capture it all in your inbox. You can now also find us at aixdesign.co.
