Comment by PaulMest | Hacker News Reader

Comment by PaulMest | Hacker News Reader

Parent Full thread

I wrote a Tinder bot a couple of years ago and ran some interesting studies:

1) I placed a male profile in ~10 different cities, auto-swiped for a few days, and waited for matches and messages to flow in. I found the match/message rates to be significantly different for each city. Sample size of ~34,000 right swipes.

2) A/B test two female profiles with exact same photos and bio except 1 was a CEO and 1 was Graphic Designer. Swiped on 1000 men in NYC and 1000 men in SF.

3) A/B test two female profiles with exact same photos and bio except 1 was listed as 29 and 1 was listed as 31. Swiped on 1000 men in NYC and 1000 men in SF.

Through my experience in this, I can see the methodology they used in this study could be flawed. Specific concerns:

* While I don't have any real inside knowledge into Tinder's "recommendation" algorithm of who they show to you, I assume there is a strong preference to show active users first. So if they are swiping on hundreds of thousands of profiles, they are probably burning through the legitimately active users in a region pretty quickly... that's one reason why they get a lot of matches quickly and not as many matches over time and end up with a 0.6% match rate.

* The number of messages that guys send (at least as of 2 years ago when I ran my studies) is wildly more than women. The male profile was seeing an 11.8% match rate, and a message rate of 15.3% from the matches for an effective message rate of 1.8%. Whereas the women were getting an 81.5% match rate, with 63.9% message rate from those matches for an effective message rate of 52.1%.

* I believe there is now rate-limiting on the right swiping, so after ~100 or so right swipes, you have to wait 12 hours until right-swiping again. Really not sure how they could have made it through hundreds of thousands of profiles unless they paid for the premium membership.

Anyway, interesting stuff regardless, happy to release more data when I have time, if there is interest.

Honestly curious: Do you feel guilty for wasting the time of 4,000 men? To me, it feels selfish to value your own learning over the time of thousands of other strangers. Of course, you and others may feel differently than me. What's your approach for thinking about where to draw the line?

Edit: To those downvoting, could you explain why? I'd be happy to improve my comment, if I knew how. I am honestly interested in discussing the ethics of bots that pose as humans.

I was a fairly active real user of Tinder for a while. Though, I noticed that a lot of my matches were from Spam bots who were trying to get me to go to malicious links or promote businesses. I wasn't overly annoyed by them and would report those bots to Tinder as "feels like spam". I felt like this was a known part of the ecosystem and my bots didn't feel too burdensome or evil.

The line that I drew at the time was that, aside from the initial match, the bots were never going to engage in any on-going way with the matches. So the bots could receive messages from matches, but would never send messages or post Moments (which was a feature at the time to effectively message all of your matches with a photo).

If you're a man on Tinder, you play a numbers game and assume a relatively low response rate. It doesn't take to long to send that first message

This argument is fallacious because a real person, upon being sent a message, is not obligated to reply, irrespective of a mutual match having been obtained.

By questioning the bot author over the morality of their methodology, your post appears to suggest that people who don't reply to messages would be equally perceived as having wasted the time of message authors, which is a view that your downvoters find troubling. That may not have been your intention, given your disclaimer and edits, but I can see why it could be construed that way.

Thank you for your reply. To be honest, I still don't see why an obligation to reply makes my argument fallacious. Email spam is universally considered bad, despite the fact that no one is obligated to reply to email spam. Maybe I am misinterpreting what you wrote.

In response to your meta question, I found it weird that you specifically asked if the parent felt bad about the 4000 men but not the 34000 women. Though I didn't downvote.

Whoops, that was entirely unintentional. I think my eyes were drawn to the parallel 1000s on lines (2) and (3). I feel weird about missing the women and I apologize for the omission. Everyone's time is valuable.

On the internet, nobody knows you're a dog.

What were the results of your A/B tests??

I'll probably write a blog post about this in the near future. What do you think the results were?

Feel free to reach out to me privately (contact info in profile) and I'll share with you a quick digest of the results before the blog post gets published.

The imbalance in swiping effectively turns Tinder into Bumble. Women are really making the first move, which is great.

Nearly all of the data contradicts what you're saying.

Reply on news.ycombinator.com