A reasonably quick read on breaking up Facebook and Google, due to their ability to abuse their monopoly positions to dominate attention and their motivations to pump up “engagement” at all costs - even at the cost of direct harm to societies.
This paper discusses how we got here, and what we can do - which involves radically reducing the consolidated power of these two tech companies.
One of the better ideas in there:
MAKE FACEBOOK AND GOOGLE RESPONSIBLE FOR CONTENT. Communications networks should be forced to make a choice between being regulated as a communications facility or as a publisher. Requiring this decision means modifying Section 230 of the Communications Decency Act and Section 512 of the Digital Millennium Copyright Act so that large communications networks do not receive liability protection if they profit from advertising or targeted advertising. This modification would likely force Facebook and Google to change their business model. Amending Section 230 and Section 512 would also help restore a level playing field for publishers, who are legally responsible for the content they publish.
Force online platforms to “pick one” - either you get to sell ads on user-generated content and are responsible for what’s posted, or you get liability and can’t shove ads into user-generated content streams.
This, if actually enforced, has the potential to break some of the feedback loops that are currently responsible for a lot of the toxic effects of social media. Facebook, in particular, has no incentive to actually do more than the minimum required to deal with “fake news” and such (in the literal, “this was totally made up and is 100% false” variety of “fake news”), because that sort of content is very “engaging” - it shares well, goes wild, and leads to more people on their platform viewing ads. Sure, it’s literally made up, reinforces existing biases, etc, but… I mean, who cares about that? It sells ads!
YouTube is just as bad. From what I’ve heard, at least a few years back, their internal metrics were pretty straightforward: Hours watched. That graph needs to go “up and to the right.” Their goals, at some point in time, were to beat out network television for viewer-eyeball-hours. And that decision - focusing on hours watched - has lead to quite a bit of the human-toxic behavior of their algorithms. Despite what they might claim, there’s no way the recommendation engine knows anything about what it’s recommending. The closed captions are bad enough before you get to the state of pulling semantic meaning out of content, so the engine is almost certainly just A/B testing various things. “Someone who watched video 1234 is highly likely to sit through video 5678, and that increases hours watched, so we’ll recommend that!” Great for increasing viewer hours, except for the fact that you end up recommending all sorts of conspiracy crap and other nonsense. Turn someone onto a few conspiracy theories, they’re likely to look into all sorts of others - and as long as that’s on YouTube, well, what a win! A more engaged viewer! Sure, you’re filling their heads with lies and nonsense, but as long as they’re watching…
You’ll excuse me for not thinking this is the best set of ideas I’ve heard.
There’s obviously a long path between where we are and something more sane, but I’m just not happy with what the current behaviors are.
Eh. I’d say it’s just networking effect, and if anyone really cared, they’d use the federated types of social media.
Which… no one does, because it’s not the easy solution.
It’s not easy, partly because we’ve allowed “easy” to override everything else. Facebook is exceedingly easy to use. YouTube is exceedingly easy to use, to the point that it’ll auto-play down into weird rabbit holes (or at least used to).
I’m trying to come up with ways to restore older, distributed, smaller communities - this being one example here. But ideas are welcome!
The gating issue is hosting. In the great before, the ‘easy’ solution was usenet, federated and hosted by multiple parties. One generally paid for access (directly to a host or indirectly via a school or service provider) and censorship was limited to moderation and/or providers not carrying more controversial channels. Then came webboards which stopped federation (to keep the eyeballs on site) and the hosting model was ad or interest based. As these entities grew, they became targets for people who wanted control (for various reasons). Federation, obviously, isn’t in their best intersts.
The solution is to go back to federation. Platforms like mastadon or diaspora* seems to be winning the standardization wars, though a better version would be one hosted on a ‘freedom’ (tor, freenode, etc) platform that can be more redundant than a single provider. The second gating mechanism, then, becomes ‘ease of access’. A website is easy. Newsreaders, less so. Tor / freenode is hard.
Federation is about the only way. Lots of places, local control, same protocols so each place can talk to the others. (so d* and mastodon)
Let’s say you run $anarchistDiapora*node and I run $LEOnode … neither controls the other’s speech, and if things come to worst, we stop syndicating to each other. Same stack, middle nodes can talk to both (maybe, if they still connect to both nodes…)
Also solves the namespace problem somewhat. Instead of name poaching, since there’s no central or master node, User@biznode.com is the official source for that biz. Kinda like email in that regard.
I think one interesting question is, “Do we even need federated systems for social media?” Do I care what someone across the globe is doing/thinking, or is it just a distraction from things that matter more?
There’s certainly some value in federated systems - IM systems, for one, where anyone should be able to talk to anyone, though one may also have local pods of people (I run a Matrix server, those of us who talk on it don’t have traffic leaving to other servers). But if the problem is scale, I’m not sure federated solutions really solve that. Humans just don’t scale, and the internet, as a whole, has tried very, very hard, to scale humans like computers.
No, but it’s occasionally relevant to have a state agency, say IDOT post once, and it shows up in my stream along with other stuff. I can only handle so many streams at once. Now, this is a case for RSS syndication into a single blog stream… but that ship has also sailed long ago.
I don’t generally think government agencies should be using social media as a way to communicate, though I’m also pretty sure that ship has sailed.
They go where the users are, but yeah. Though gov should really do ‘all of the above’ for broadcast stuff like traveler information, etc.
I’d faintly rather see “This is the canonical government data stream, import how you wish.” Instead of ad-hoc posting on Facebook, Twitter, Instagram, etc.
No idea how to get from here to there.
I may mess with a Mastodon or Diaspora instance this weekend - I assume they’re not resource intensive for small numbers of users.
That’s the great thing about history: until we forget it we can always pull it back out if it becomes useful. Even gopher and usenet served excellent purposes and if for economic reasons large scale search engines finally did bite the dust we may well see the re-rise of indexes such as those again.
Great idea honestly. Then it doesn’t matter what format it is as long as it’s open, fairly easily accessible and parsable, and accurate and up to date. Instead of federation, one could combine the twitter and rss paradigms into one and end up with a subscription based non-federated but also non-centralized system for information broadcast and sharing.
Something else that came up in conversation: One of the problems (at least, as some see it) is that the private, targeted advertisement system is opaque and essentially impossible to audit. I can’t know what you’ve seen, you can’t know what I’ve seen, and we sure can’t trace back who is doing a lot of advertising and promotion. As long as they pay the FaceBucks, well. All is good.
As much as I think blockchain is a solution looking for a problem, for an immutable public record of something, it’s actually halfway useful - so one could envision an “Ads Blockchain” requirement for places that show personalized advertisement. It’s huge, annoying, and difficult? Well, one could go to showing everyone the same ads… which would also go a long ways to mitigating the “personalized engagement” evils of social media as currently implemented.
But the “Collect all the data on everyone, to customize their feed as much as possible, to show them the most personalized ads possible” model really needs to go.