There's two things going on - technology, like you mention, but also scale.
Scale brings trust problems. Got a million people in your city? Even without technology, you probably don't know Bob. Bob can rip you off without realistically expecting repercussions, unlike in a small village where everyone depends on each other, and Bob would get kicked out and have to try to find somewhere else. So scale also breaks a lot of the value of permenant IDs, and things like outsourced trust (credit reports, etc) will never be perfect. And in a big society, even a 99% success rate would be A LOT of people getting fucked over.
Can this eyeball-thing fix that? Largely, no - any errors in the system that aren't identity theft will still be there. And potentially worse: even fancy technological systems are almost always vulnerable to technological exploits. And then if too much trust is placed in the validation mechanism, a bad actor could potentially go nuts.