(Keep in mind after this, Facebook hired 20,000+ content moderators at great expense who they still employ to monitor for stuff like this).
(Keep in mind after this, Facebook hired 20,000+ content moderators at great expense who they still employ to monitor for stuff like this).
According to the article, Facebook did make the deliberate decision not to dedicate resources to Myanmar and other non-western countries, so I don't think it's negligence. "Greed" is more appropriate.
> Guy Rosen, Facebook’s VP of Integrity, told Sophie Zhang that the only coordinated fake networks Facebook would take down were the ones that affected the US, Western Europe, and “foreign adversaries.”
There's no obligation to operate everywhere. Instead, as discussed in previous parts, they invested heavily to get a monopoly in third world countries.
But they didn't. The situation was as you put it mostly unmonitored. In other words, the spread of genocidal content was organic. This is the equivalent of blaming the phone company or the ISP for what the users do with the communication channels.
I promise you if they tried that people would complain even harder.
No one expect Facebook to remove all misinformation/hate speech/etc. That is unreasonable. But they have literally 1 data scientist to monitor the whole non-western world.
In a year where their annual profit is $23.9bn.
Assuming 1 data scientist is paid $500k/year, surely it is sensible to dedicate $20mn, less than 0.1% of their annual profit, to hire 40 more such data scientist and literally increase their effectiveness on this topic by 4000%? At the very least, this is what I expect from a company that actually have a speck of moral.
> A strategic response manager told me that the world outside the US/Europe was basically like the wild west with me as the part-time dictator in my spare time. He considered that to be a positive development because to his knowledge it wasn’t covered by anyone before he learned of the work I was doing.
The integrity org, back when I was there, had thousands of folks and was aggressively hiring. They were not a lone island and were also supported by other orgs like data infra, FAIR, security, privacy, product teams and more.
Site integrity, Sophie’s team, was a small piece of the work done there. And they, like any team, relied heavily on other teams. You have all this unbelievable tooling internally, and internal teams are incentivized to get you to use it.
These issues were not from want of trying. The reality is that problems like this at scale are incredibly difficult. In my opinion, Frances trivialized a lot of this in her whistleblowing, but it makes a good news story so what can you do.
It pains me when people trivialize it as “just do the machine learnings to fix it all” or “if these tech companies actually gave a damn this wouldn’t be an issue”. It’s an incredibly hard problem, and much like security it will never be “solved”.
That doesn’t excuse when technology has a bad effect on the world. We need to be better, and for the most part as a society we are trying hard as hell. It’s also why the work is interesting and impactful. More should get into it.
I talked to FB people (recruiters, managers) in the site integrity team back in 2018. Admittedly I had some reservations up front, but during the on-site day I made up my mind to never even consider FB as a potential employer again. During a chat session between interviews, I asked about the prospect of doing proactive education - essentially, detecting users who had been caught in the influence operations and then surfacing them a note that they had been targeted by such activities, so that they could themselves make educated decisions.
The senior manager I was talking with at the time was visibly taken aback by the very idea. "We don't do that!"
From that experience I drew the inference that FB are fundamentally, as an organisation, incapable of doing the right thing. I suspect it's less about the cost, and more about the prospect of openly accepting accountability for what their platform is really used for.
Insisting that it's hard and requires advanced ML is complete BS. Self delusion by FB'ers to give themselves and excuse for their complicity.
If an airline doesn’t have enough pilots to service 20 routes they reduce their routes, not let passengers fly the plane.
Look at fsociety's comment in this same thread to see just how much misinformation is out there (which he helpfully corrects)
It just so happens a lot of misinformation is anti-corporate since many people want to believe every bad thing they hear about them. So I end up being one of the few commenters who say, actually, the big company did try here.
This is important because if every big company is supposed to be equally bad, there's no reputational penalty for the actually bad ones.
The point people are making here is that Facebook shouldn't have done the impossible and created a perfect tech solution to moderate content, they should not have operated in these countries at all if they couldn't do it safely. They did not even try, instead they ignored internal reports on these problems for years and actively decided to make the problem worse with their single focus of growth. And for as you put it, only a few $ in advertisements, to fuel the fires of genocide.
Of course it's greed, and of course it's reasonable. Apple is directly responsible for the work conditions of its workers and the contractors they use, absolutely. If they can't be responsible then they can close shop. There is no obligation for them to have a business.
They did certainly care about the growth of daily active users even if it meant they expanded to markets where they couldn't responsibly moderate their platform.
A company as big as Facebook is going to have different groups with different objectives that often do not see eye to eye or even know about each other. A company as profitable as Facebook is going to have projects that are motivated by things other than profits.
There is most likely a team at Facebook whose goal it is to get as many users as possible in developing markets. These markets do not generate profits from online ads. Facebook will never disclose its cost per user but in general online ads only make money from affluent users. I have heard that even mid income countries like Mexico or Romania are not profitable for ads. A place where people make an average of $3 a day is wildly unprofitable for Facebook and even the most optimistic development projections will not make them profitable for several decades. The OP is spot on that the amount of money that Facebook will ever receive from Myanmar is negligible.
Based on conversations with people at Facebook I generally get the sense that people there (especially at the top) have a strong sense of mission and a belief that their product makes the world a better place (I disagree with this but that doesn't change the fact). I would guess that these the project tasked with increasing users in developing countries is viewed more as philanthropy than business.
Part II of this series addresses this stat. During the peak genocidal years it seems there were only a handful of mods assigned to Myanmar, and none of them lived in Myanmar. To this day it sounds like there still may not be more than a hundred people moderating Myanmar content. Which is inline with the 20K number: 195 countries * 100 moderators per country = 19.5K mods total
I guess my main idea here is that when I look at the back-of-envelope numbers it's easy to see how even 100 moderators would have trouble keeping up with the content that millions of people produce. So you really do need those automated moderation tools to be effective.
Can someone please continue the thought experiment by estimating how much content each mod can review in a day, then extrapolating up to a week or month, and comparing to an estimate of how much content was actually posted on Facebook in the similar timeframe? I believe Part I or II provides an estimate like that.