3 ms·
I hear this argument all the time. What I never hear is anything that actually helps to judge the trade-off. How often is a flag incorrect? How often is a legit
by andreareina 5y ago
I hear this argument all the time. What I never hear is anything that actually helps to judge the trade-off. How often is a flag incorrect? How often is a legitimate user able to recover from the flag and how much effort does it take? I notice you say your flags have only been often temporary, i.e. sometimes permanent.
Are we really ok with throwing away due process because it might help the criminals? Sure nobody needs that game but access to money services is a bit different. And what you also never hear about is being refunded for all the content paid for in the blocked account.
- 0xy 5y agoThe justice system has thousands of false positives and throws innocent people in jail for decades sometimes. As a whole, society accepts some level of error margin in every facet of society. After working in risk and fraud, I've seen some truly horrifying stuff. An elderly woman threatened physically if she didn't let the attacker drain her bank account. Huge scams for thousands of dollars. Organized crime rings fleecing millions of dollars. Identity theft. On and on. I would wager that society would accept some level of errors in exchange for drastically reduced criminal activity that actually hurts people. Legitimate people getting blocked inadvertently are rare, and often of the stories you hear on HN and Reddit and other places, a good portion of these "innocent victims" are actually doing sketchy stuff when you dig into it. Justice isn't and cannot be an exact science. Society agrees and accepts the risk.
- andreareina 5y agoI'm not asking for exact numbers. But let's have some numbers, because otherwise how can we agree to something when we don't know what we're agreeing to?
- 0xy 5y agoAccording to some studies, up to 1% of prisoners in the U.S. are innocent. That would mean 23,000 people wrongfully imprisoned. In my experience, false positives in fraud and risk models are lower than 1%.