It should be even more evident that you can't judge whether they're doing a good job by looking at just the numerator. You need to know the denominator as well. Or rather, the denominators.
In a simplified model there are two groups of users {good,bad} and two outcomes {suspended,not suspended}. You're saying that success can be judged by whether there are any people claiming to be good (though you have no idea of whether that's true, they're just claiming that on the internet) ending in the {claims-to-be-good, suspended} bucket.
But actually to judge whether they're doing a good job, you'd need to look at the {good, not suspended}, {bad, suspended} and {bad, not suspended} buckets too.
The first one is a baseline. Obviously if you've got a billion users, all your numbers are going to be 1000x higher than a somebody with a million users just due to the higher number of users. The number of internet complaints will be 1000x higher too. But the actual harm to the average user from the mistakes is the same.
The second bucket are the successes, and they are going to be totally invisible to everyone not working on the problem. Not only to the random HN commenters, but to the random bigco employee too. They literally can't judge the success. The only visibility will be if that number is too low, since it obviously means the third bucket is too high. And that will be visible as the platform being overrun by spam and scams.
Now, I understand that your view is that it must be completely impossible for a possibly good user to lose an account. That's just the kind of thing that people who don't understand the problem space would say at an all hands open mike, and then get ignored because their view is just so detached from reality. It's not even a matter of resources; even if you threw infinite resources at the problem, what you'd end up with is a worse experience in the aggregate.
You'd have scammers reinstated, and continue scamming more people. You'd have accounts be hijacked because the scammers are going to be better at social engineering their way into accounts than the real users will be at social engineering their way to account recovery.
It's all tradeoffs, and absolutist statements about how it's unacceptable for even a single good user to suffer any harm are just as unrealistic as absolutist statements about how even a single piece of spam can't make it through. The best you can do is try to find the best place in the tradeoff space.
In a simplified model there are two groups of users {good,bad} and two outcomes {suspended,not suspended}. You're saying that success can be judged by whether there are any people claiming to be good (though you have no idea of whether that's true, they're just claiming that on the internet) ending in the {claims-to-be-good, suspended} bucket.
But actually to judge whether they're doing a good job, you'd need to look at the {good, not suspended}, {bad, suspended} and {bad, not suspended} buckets too.
The first one is a baseline. Obviously if you've got a billion users, all your numbers are going to be 1000x higher than a somebody with a million users just due to the higher number of users. The number of internet complaints will be 1000x higher too. But the actual harm to the average user from the mistakes is the same.
The second bucket are the successes, and they are going to be totally invisible to everyone not working on the problem. Not only to the random HN commenters, but to the random bigco employee too. They literally can't judge the success. The only visibility will be if that number is too low, since it obviously means the third bucket is too high. And that will be visible as the platform being overrun by spam and scams.
Now, I understand that your view is that it must be completely impossible for a possibly good user to lose an account. That's just the kind of thing that people who don't understand the problem space would say at an all hands open mike, and then get ignored because their view is just so detached from reality. It's not even a matter of resources; even if you threw infinite resources at the problem, what you'd end up with is a worse experience in the aggregate.
You'd have scammers reinstated, and continue scamming more people. You'd have accounts be hijacked because the scammers are going to be better at social engineering their way into accounts than the real users will be at social engineering their way to account recovery.
It's all tradeoffs, and absolutist statements about how it's unacceptable for even a single good user to suffer any harm are just as unrealistic as absolutist statements about how even a single piece of spam can't make it through. The best you can do is try to find the best place in the tradeoff space.