If only 0.01% of toxic content by terms of volume is receiving disproportionate attention and focus, not only permissible but actively encouraged by complex opaque algorithms, then it is certainly a much bigger problem then it's meagre character count would otherwise suggest.
To me, this needs to be split into two use cases. The fact that .01% of content gets viewed a lot more than average is not a bad thing. Much of user content is junk with a heavy dose of ads for a seasoning. It is perfectly reasonable that original or thoughtful posts, even if unpleasant or provocative, get much higher traffic. The problem is the algorithmic reinforcement that encourages those making short-term sensationalist statements.
Let's still allow folks to voice their opinions, whichever they are, and advertise them by the digital age equivalents to "word of mouth". Just keep it human, not algorithm driven. My 2c.
Without those spontaneous user-composed opinions and anecdotes, the platform would quickly become (or perhaps already has become) plastic and artificial. As anyone having to relate information through a bureaucratic communications department has certainly experienced, even the most unique, interesting and useful story in the world stands little chance against an eager team of copy editors and PR representatives.
My presumption is that by far the single biggest signal on whether to display a post on the FB feed to more people is how many people like and share. That's a people problem at the core, not an algorithm problem.
It's still an algorithm problem at its core, because the algorithm is what determines what the "single biggest signal" it will use for amplification is. There's no reason why maximum shareability should be the main factor, it could be something else like some proxy for erudition, for instance.
Simply: it’s all content and it’s impact depends on the people who take the information in and then apply their beliefs and understanding to it.
Some content is simple - recipes for example.
Others are impossible to parse: “they want to reduce hate speech” translates as “they want to remove conservative/right wing speech” (this actually happened in a conversation yesterday. Suggesting that the term hate speech is being transformed/co opted)
Engagement in contrast, is simple, reliable and measurable.
I don't think that's true, it's just that the proxies are imperfect. But the specific alternative metric wasn't that important to my point.
> Engagement in contrast, is simple, reliable and measurable.
That's the root of the problem: the algorithms have been designed to use "simple, reliable, and [easily] measurable" things as their whole input, and then narrowly A/B tested to maximize some other easily measurable output. Looked at holistically, it's too simple to do an acceptable job.
If you took a similar attitude, and applied it to create an algorithm for another discipline, like medicine for instance, it'd be pretty clear it'd be a disaster.
However, this shows that the root problem really is human. You can tell because you are suggesting designing an algorithm to fix a pre-existing problem. Without algorithms of any sort, this problem is there.
I don’t think it can be solved by algorithm.
I think we can all agree that quantity isn't the right metric for this - whether that's number of posts or amount of traffic on a given post. What we're really trying to do is gauge what the conversation is.
Having said that, there are clear distinguishable problems we can point to that have an interesting quantifier. Anonymous 'Campaign' groups run by individuals that have previously been associated with Brexit campaigning are currently spending millions on Brexit advertising. There is no indication of where the money comes from to pay for these (we know it's not from the people who are associated with the page because they're just staffers).
In that way I agree with you - the thing to worry about is where the ad dollars are spent, but I don't think that's the only thing we should be worrying about.
The 1% Rule says that in any online community, 1% of participants create content while 99% lurk [1].
I'll propose Panarky's Corollary that 90% of the crap is produced by 0.1% of the participants and then shared, liked and amplified by 90% of participants.
These 0.1% pathological participants are corporations advertising, governments propagandizing, zealots proselytizing, and trolls antagonizing.
Facebook can see these patterns better than anyone, and they could eliminate or de-rank the material from the 0.1% if they wanted to.
But the 0.1% is disproportionately incendiary and therefore profitable, so Facebook has a very strong incentive to look the other way.
[0] https://en.wikipedia.org/wiki/Sturgeon%27s_law
[1] https://en.wikipedia.org/wiki/1%25_rule_(Internet_culture)
That's not my starting assumption.
I think Facebook has a problem that its typical so-called "benign" users are creating less and less content, while malignant content-makers are not discouraged.
Not really. Facebook put it there for me. Time and again it shows me items like "[Friend] replied to a comment on [toxic thing]", where the thing gets a prominent preview etc. I haven't seen a real update from [Friend] in my feed for months, yet there are plenty when I visit their profile. Even if the friend is fighting the good fight and has commented to correct some misconceptions, it's clear that this particular action was selected for my feed because it's related to a controversial item.
If it weren't for comments from non-friends, Facebook would have no toxic content for me at all.
Facebook was great until it started inserting posts from people I didn't explicit whitelist as friends, and I'm including news articles in that category.
https://www.thisisinsider.com/social-media-link-depression-l...
https://www.cbsnews.com/news/rise-in-suicide-and-social-medi...
You mean like the rumour-mongering, low-effort politics etc. in your standard tabloid?
The whole point of tolerance is to say "I don't agree with that, but I respect your right to do your own thing."
This is advocating the exact opposite. Under this philosophy I can not longer be tolerant, because anything I "permit" I will be accused of promoting.
This seems like a recipe for a horrible, horrible world.