I have been shadow-banned on an app (Tinder) and I cannot get around it. I have been doing some background research into how they are able to identify me and have come across device fingerprinting (I confirmed this through fiddler and seeing a device fingerprint being sent to their API) and perceptual hashing (to be able to identify you based on your old photos that you cropped a few pixels off to try fool any primitive hash algorithms etc)

I wrote a previous post on fingerprinting here.

I got a new email address, new phone number, deleted the native app then rebooted my router to get a new ip, used Firefox for mobile, but the same photos and no luck, still shadowbanned. I repeated the process using a different browser (would presumably have a different fingerprint) and a friend's photos and also no luck.

I don't believe perceptual hashing or facial recognition to ban users is feasible at their scale, as people use the same group photos etc. in their profiles. Also they must have a few billion photos, so there will be a lot of similar photos of people posing at famous landmarks etc. They have 200m users so 99% accuracy would still cause a support nightmare?

There is a similar report in the wild of the situation here.

How is it possible that they are still able to identify me?

  • 3
    I'm not really sure what you are asking. There is lots of noise in your question about being shadow-banned and why this could be but ultimately you ask how feasible perceptual hashing would be to identify users when they use similar photos. But why do you focus on perceptual hashing? There are other ways to identify users based on photos - for example face recognition is not done with perceptual hashing. – Steffen Ullrich Feb 2 '19 at 20:20