[coderintherye]

I think with fraud detection, there is sort of this ingrained idea that you can't tell the person in question what the suspected fraud is with the hypothesis that it will allow fraudsters to figure out a way around your detection.

I don't know that this hypothesis is true though and it certainly harms your real users when there are false positives. Fraud is certainly a big problem for platforms, so it's not like they can just ignore, but at the same time surely can do better than treating your long-time users to a ban with no explanation and no warning.

[martin_a]

Feels somewhat like security through obfuscation and I think fraudsters will find a way around anyway.