Comment by carapace | Hacker News Reader

Comment by carapace | Hacker News Reader

Parent Full thread

> But the reality is that no matter how much money and manpower (plus less-than-perfect “artificial intelligence”) Facebook throws at curating hateful or illegal content on its services, and no matter how well-meaning Facebook’s intentions are, a user base edging toward 3 billion people is always going to generate hundreds of thousands, and perhaps millions, of false positives every year.

In re: FB, there was an FB engineer saying the other day that it was impractical to have humans review every post.

I almost responded (but refrained out of a sense of not being snarky) that the word "impractical" here actually means "uneconomical".

No one's forcing Facebook to be what they are.

> no matter how well-meaning Facebook’s intentions are

They want money at the end of the day.

I know I'm beating a dead horse here, y'all know this already, but the service is free to entice the "users" who are then sold to the ad/marketing folks, the REAL users of FB.

It's possible to set up a non-exploitative social network (and indeed, many people are doing just that) but FB/Twitter/etc. have sucked the air out of the room and normal people do not care that they are digital cattle. They just don't, not in any effective operational way.

If somehow FB shut down tomorrow everything their servers do could be done by other servers (the data would be a problem, archive it, yeah, but then who gets to access what? That's just more data though, eh?) FB doesn't fundamentally add anything to the underlying capabilities of the Internet, eh? There are no "magic angel babies" working there.

I think if you somehow went back in time to, say, the 1950's and tried to tell people about FB they just wouldn't believe it. They wouldn't believe that so many hundreds of millions of people would voluntarily give a private company every detail of their private lives (more than the Stasi worked hard to collect) for no more than the Internet dressed up in a crappy UI.

But people do crazy things as long as everyone else is doing it too and there we are.

I don't really have anything constructive to say. Sorry.

> as long as everyone else is doing it too

That right there never ceases to amaze me. How everybody just willingly joins knowing they’re hurting themselves but out of the need to be in the loop with everybody else.

I’d say you hit the nail on the head. Crazy things indeed.

TLDR: $6.25M is all it would cost to implement Orwellian censorship across Facebook.

TFA says there are "hundreds of thousands, and perhaps millions, of false positives every year." Let us assume there are a million false positives that require human intervention every year. There are 50 work weeks in the year, so to resolve 1,000,000 false positives over 50 weeks requires humans to resolve an average of 20000 per week, or 4000 per day, or 500 per hour, or 125 every 15 minutes.

Therefore, a team of 125 people could theoretically solve this problem by resolving 1 false positive every 15 minutes. At $50,000 US dollar pay scale, Facebook would need to pay out $6.25 million dollars to solve this.

Therefore, given "a user base edging toward 3 billion people is always going to generate hundreds of thousands, and perhaps millions, of false positives every year," it would cost FB ~$6.25M to hire a team of 125 people to resolve the issue, per million false positives.

That seems economical for a multi-billion dollar company.

Facebook already employs 15,000 content moderators, both directly and via contractors. By many accounts, they're overworked and don't get enough time to deal emotionally with what they have to see [1] -- and they only review a fraction of the content on Facebook. 125 people would clearly be many orders of magnitude too small to review everything.

The "hundreds of thousands, and perhaps millions, of false positives" quote is referring to the rate after human moderation (and is also just speculation by the author).

I think people tend to underestimate the scale that would be required to solve this problem because they tend to underestimate just how terrible people are.

[1] https://www.theverge.com/2019/6/19/18681845/facebook-moderat...

A lot of those false positives are going to be things that look very much like a true positive to an underpaid, overworked moderator. I don't think 15 minutes would cut it. Actually, I'm not convinced that any amount of time would, because there are too many occasions when the distinction between hate speech that must be banned and social justice speech that only a bigot would ban rests entirely on the social standing of the person making it within the appropriate circles.

Also, I strongly suspect that Google, Facebook, etc are already applying this kind of manual scrutiny to potential false positives behind the scenes - as you say, it's relatively cheap - and this explains a lot of the problems people have been having with their moderation decisions and why appealing decisions so rarely works out.

It will not be only the ones whose post was removed as a false positive asking for a manual review.

>I almost responded (but refrained out of a sense of not being snarky) that the word "impractical" here actually means "uneconomical".

Well yes, that is the definition of unpractical more or less. (Have you ever heard anyone say impractical and not just mean ecomomically unviable?)

Reply on news.ycombinator.com