Engineering Ethics Blog: Facebook’s Dilemma


 

This week’s New Yorker
carried an article by Andrew Marantz whose main thrust was that Facebook is not
doing a good job of moderating its content. 
The result is that all sorts of people and groups that, in the view of
many experts the reporter interviewed, should not be able to use the electronic
megaphone of Facebook, are allowed to do so. 
The list of such offenders is long: 
the white-nationalist group Britain First; Jair Bolsonaro, “an
autocratic Brazilian politician”; and of course, the President of the
United States, Donald Trump. 

 

Facebook has an estimated
15,000 content moderators working around the world, constantly monitoring what
its users post and taking down material that violates what the company calls
its Implementation Standards.  Some
decisions are easy:  you aren’t allowed
to post a picture of a baby smoking a cigarette, for example.  But others are harder, especially when the
people doing the posting are prominent figures who are likely to generate lots
of eye-time and thus advertising revenue for the company. 

 

The key to the dilemma that
Facebook faces was expressed by former content moderator Chris Gray, who wrote
a long memo to Facebook CEO Mark Zuckerberg shortly after leaving the company.  He accused Facebook of not being committed to
content moderation and said, “There is no leadership, no clear moral
compass.”

 

Technology has allowed Facebook
to achieve what in principle looks like a very good thing:  in the words of its stated mission, “bring
the world closer together.” 
Unfortunately, when you get closer to some people, you wish you
hadn’t.  And while Zuckerberg is an unquestioned
genius when it comes to extracting billions from a basically simple idea, he
and his firm sometimes seem to have an oddly immature notion of human nature.

 

Author Marantz thinks that
Facebook has never had a principled concern about the problem of dangerous
content.  Instead, what motivates Facebook
to take down posts is not the content itself, but bad publicity about the
content.  And indeed, this hypothesis
seems to fit the data pretty well.  Although
the wacko-extremist group billing itself QAnon has been in the news for months,
Facebook allowed its presence up until only last week, when public pressure on
the company mounted to an apparently intolerable level. 

 

Facebook is a global company
operating in a bewildering number of cultures, languages, and legal
environments.  It may be instructive to
imagine a pair of extreme alternatives that Facebook might choose to take instead
of its present muddle of Implementation Standards, which makes nobody happy,
including the people it bans. 

 

One alternative is to proclaim
itself a common carrier, open to absolutely any content whatsoever, and attempt
to hide behind the shelter of Section 230 of the Communications Decency Act of
1996.  That act gives fairly broad
protection to social-media companies from being held liable for what users
post.  If you had a complaint about what
you saw on Facebook under this regime, Facebook would tell you to go sue the
person who posted it. 

 

The problem with this
approach is that, unlike a true common carrier like the old Ma Bell, which
couldn’t be sued for what people happened to say over the telephone network,
Facebook makes more money from postings that attract more attention, whether or
not the attention is directed at something helpful or harmful.  So no matter how hard they tried to say it
wasn’t their problem, the world would know that by allowing neo-Nazis, pornographers,
QAnon clones, terrorists, and whatever other horrors would come flocking onto
an unmoderated Facebook, the company would be profiting thereby.  It is impossible to keep one’s publicity
skirts clean in such a circumstance.

 

The other extreme Facebook
could try is to drop the pretense of being a common carrier altogether, and
start acting like an old-fashioned newspaper, or probably more like thousands
of newspapers.  A twentieth-century
newspaper had character:  you knew pretty
much the general kinds of stuff you would see in it, what point of view it took
on a variety of questions, and what range of material you would be likely to
see both in the editorial and the advertising sections.  If you didn’t like the character one paper presented,
you could always buy its competing paper, as up to the 1960s at least, most
major metropolitan areas in the U. S. supported at least two dailies. 

 

The closest thing the
old-fashioned newspaper had to what is now Facebook was the letters-to-the-editor
section.  Nobody had a “right”
to have their letter published.  You sent
your letter in, and if the editors decided it was worth publishing, they ran
it.  But it was carefully selected for
content and mass appeal.  And not just
anything got in.

 

Wait a minute, you say.  Where in the world would Facebook get the
dozens of thousands of editors they’d need to pass on absolutely everything
that gets published?  Well, I can’t
answer all your questions, but I will present one exhibit as an example:  Wikipedia. 
Here is a high-quality dynamically updated encyclopedia with almost no
infrastructure, subsisting on the work of thousands of volunteers.  No, it doesn’t make money, but that’s not the
point.  My point is only that instead of
paying a few thousand contract workers to subject themselves to the psychological
tortures of the damned in culling out what Zuckerberg doesn’t want to show up, go
at it from the other end. 

 

Start by saying that nobody
gets to post on Facebook unless one of our editors has passed judgment on
it.  When the nutcases and terrorists of
the world see their chances of posting dwindling to zero reliably, they’ll find
some other Internet-based way to cause trouble, never fear.  But Zuckerberg will be able to sleep at night
knowing that instead of paying thousands of people to pull weeds all the time,
he’s started with a nice sterile garden and can plant only the flowers and
vegetables he wants to.  And he’d still
be able to make money.

 

The basic problem Facebook
faces is that they are trying to be moral with close to zero consensus on what
moral is.  At least if the company was
divided up into lots of little domains, each with its clearly stated and
enforced standards, you would know more or less what to expect when you logged
into it, or rather, them. 

 

Sources:  The article
“Explicit Content” by Andrew Marantz appeared on pp. 20-27 of the
Oct. 19, 2020 issue of The New Yorker.



Source link