4 ms·
I mean, they have _millions_ of users that have never had a problem. It's far more likely this is a mistake than any kind of malicious act, just given the sheer
by b4ux1t3 9y ago
I mean, they have _millions_ of users that have never had a problem. It's far more likely this is a mistake than any kind of malicious act, just given the sheer number of customers they have.
- nxc18 9y agoWe've had multiple people prominent enough to make the front page of hacker news in the last few months. Maybe we should get some statistics on how many people are prominent enough and controversial enough to land on the front page of hn, then work back to how many of their millions of customers are actually affected. Its probably a lot more than just 1.
- mirimir 9y agoYes, I suspect so. What it takes, arguably, is being just controversial enough to attract complaints and/or attention from algorithms. But not famous enough for a ban to attract huge attention in mainstream media.
- glitcher 9y agoWell there is now at least one highly motivated statistician out there to analyze the data you suggest. But how do you get the data to him?
- mirimir 9y agoI doubt that this is a mistake, exactly. It's likely intentional, albeit through unreliable automation. I mean, from Google's perspective, false positives are arguably far less problematic than false negatives. And I'm sure that it happens more than we know. Given that most affected users have no way to get attention. Edit: OK, so I'm dyslexic. False negatives are arguably more problematic, because they attract more attention.
- Turing_Machine 9y agoIf it only happens to one out of millions, why can't they have an actual responsive human being to respond to queries? "We have so many zillion customers that we're forced to use bots for customer service" and "Only one out of a zillion customers has this kind of problem" would seem to be mutually exclusive.
- mirimir 9y agoWell, I've read that Google customer service is quite responsive, for paid accounts. And regarding non-paid accounts, determining algorithmically which queries merit human replies is arguably just as difficult as determining which accounts have violated ToS.