Hacker News Reader

Comment by kensai | Hacker News Reader

3000 persons on top of the current 4500 is a big addition. If all these persons are dedicated to the prompt elaboration of complains and violations of TOS, it might indeed make the difference.

I don't ask if it is economically viable, I guess he knows what he is doing. Facebook is not losing money anytime soon.

View on HN

That's 7500 people whom will be "forced" as much as free employment can constitute force to do nothing but look at questionable often offensive content.

What a funny way our economy has evolved. 100 years ago many of these people might have worked in farms, 50 years ago they might have worked in factories. Today they sit in an office looking at ostensibly offensive material that facebook deems the rest of society should not be exposed to.

So? We have 1000s in the bay are building the next chat bot, cat pic ai engine etc.

Equating these two classes of activity is not even wrong.

I think it will soon enough be possible for machine-learning solutions in quality control of posts and checking of offensive content. We are not there yet, but the inroads are significant.

Well that's a big problem. Once you mechanize it, then there isn't even anyone to appeal to when you are subject to an unjust outcome. I wrote at length on this post above, but do we really want computers deciding who is or is not allowed to take their clothes off?

Fair point. But at least it will be objective. Applied to everyone with the same algorithm. And yes, the algorithm can get improved.

Makes me think of some semi-dystopian world like The Giver.

Reminds me of the Buckminster Fuller quote:

>We keep inventing jobs because of this false idea that everybody has to be employed at some kind of drudgery because, according to Malthusian Darwinian theory he must justify his right to exist. So we have inspectors of inspectors and people making instruments for inspectors to inspect inspectors.

It's a communication system used by billions, with a ratio of one moderator to hundreds of thousands.

Factory worker or a farmer would never produce anything for so many people.

OK, 7500 people dealing with (according to the article) millions of reports every week. So let's assume 2M reports, the smallest number that counts as plural "millions". Assuming a 40 hour week and no breaks, this give FB 300K man-hours to handle 2M reports, meaning each person has an average of at most 9 minutes to view the content, investigate, make a judgment and take an action. Anybody who's worked in such a role: Is that enough time?

The question is what '2M reports' means. If that means 2M reports are made by users, but they can be made on the same page/post/user, then they can be easily grouped. So, if you get 500 reports on a single post, the reports can be investigated in bulk and that would make the number of investigations smaller. I believe this is the case.

Nine minutes per piece of content would be _significantly_ more than any other company I've seen has given to reporting of user-generated content. I would also agree that unless it's a video, nine minutes is enough time.

This is a tremendous amount of time per post. A 6-7post/hr throughput is incredibly slow, unless their guidelines require they produce specific, rather than generated, feedback per post.

However, the actual number of people reviewing is unlikely to be 7500. A decent percentage of that headcount will be dedicated management, HR, analytics and administrative staff.

As another poster pointed out, you're assuming the reports are always unique, where in reality the number of unique instances/articles facebook has to investigate is going to be less than the number of times people report something.

Also 2 million isn't "the smallest number that counts as plural millions" - anything over 1 million is considered "millions".

Lastly you seem to be assuming that this is 100% manual and that each report needs to be treated as seriously as each other report. I'd imagine that at the very least these reports will be prioritized and if something is reported once by a user who reports 100 things a week, it will probably be ignored - giving more time for higher priority things.

What worries me is how these gatekeepers will be chosen. Will there be criteria that makes sure they aren't just applying their own biases? Will they come from all sides of the political spectrum, or will one political leaning be better represented? Will political content be reviewed by more than one reviewer?

I can't find it anymore, but there was a leaked pdf of their review guidelines. It was about 50 powerpoint slides and it's certainly more specific than "delete what you don't like". They had quite a few examples, usually in pairs (one to delete, one not to) to highlight the demarcation. I seem to remember stuff like "Sean Connery is stupid" (allowed) vs. "Sean Connery is stupid, like all scots" (delete, because it insults him based on his nationality).

(Not sure about the specific example–"stupid" may be too weak an insult)

They also devoted a surprising amount of pages for when to delete pictures of people urinating. Must be a much bigger problem than I would have thought.

Randomly. They'll just be the content janitors, and I doubt they will be subjected to anything more than the most rudimentary ethical screening or training.

I won't be a bit surprised if there's a systematic effort to encourage applications from within socially conservative entities such as religious groups, and going to be carefully monitoring their corner of the web for such.

Biased "junior level inquisitors" might not be all that bad as long as they represent a spectrum of biases and are reliably kept from specializing on opinion bubbles relevant to their personal bias. In any case, Facebook would be well advised to put some greying neckbeards in charge who remember everything about /. metamoderation.

He is keenly aware (cf the privacy bugs fiascos of mid 2010) that trust in the system is hard to build up and easy to lose. It takes enormous sincere and public effort to pull out of loss-of-trust spirals.

You can count on the positions being low-paid in developing countries. 12 hours a day at $1 an hour maybe.

Isn't there a lot of cultural and language nuance in determining what's objectionable though? I can't see that working out well.

The majority of them should be from developing countries, since that's where most of FB users are.

Oh great, like I want my feed curated by someone from the other side of the world with radically different sensibilities. What could possibly go wrong?

The current ones are in Palo Alto and pay $17-20/hr. They are staffed by vendors Accenture/Collabera - obviously with very few benefits compared to facebook. http://apps.collabera.com/jobs/57642-Content-Review-Analyst-...

Reply on news.ycombinator.com