Decentralization is the only way to systemically fight social media disinformation
#Russia 's information warfare in the #US didn't stop after Trump was elected. And US social network giants have failed to address it in the last 4 years. Partly because they directly benefit from content engagement, partly because as centralized platforms with billions of users, they naturally offer a bigger incentive to spread misinformation than any effort they are willing to mount to fight it.
The real solution, of course, is breaking the system into human-sized parts, through #decentralization. Even the biggest english-speaking #Fediverse instance has only 400,000 users, which makes it a mediocre target for misinformation with no opportunity for paid reach towards the nearly 3 million #Fediverse users.
You too can join the Fediverse and reduce paid agitators' incentive: https://fediverse.party/en/fediverse/
Hypolite Petovan
Despite the article's conclusion coming out of and going nowhere, I still stand by my opinion: a decentralized model makes it structurally harder to spread information and misinformation alike, which would make the IRA's job naturally harder even before factoring in human moderation.
Alexander
Fediverse has less problems with these hazards (including many others - spam being most obvious) because few people are using it, not because it has some architectural advantage. In my opinion Fediverse in its current form is even more vulnerable.
One may argue that solution is in fragmentation and isolation but so far people don't want it. They want to be connected.
Hypolite Petovan
Alexander
It is pretty much like email and nothing gets more spam, fraud and other malicious interference than email.
2. There isn't any identity verification or fraud control and it is very hard to combine them with anonymity and privacy. Big centralized services can just shove it: "No valid phone number? No Twitter for you! And no, we don't like this number either..."
3. Free services are simply lacking administrative controls because the first spam/scam trolling epidemic is yet to come. There are no spam filters, there is almost zero monitoring over instance content, no remote identity verification before the first contact happens (as I think, maybe I am wrong?), it is very easy to spoof users... In other words free platforms are just underdeveloped in this regard.
By the way malicious actors are already trying the water. So far it doesn't seem an... show more
It is pretty much like email and nothing gets more spam, fraud and other malicious interference than email.
2. There isn't any identity verification or fraud control and it is very hard to combine them with anonymity and privacy. Big centralized services can just shove it: "No valid phone number? No Twitter for you! And no, we don't like this number either..."
3. Free services are simply lacking administrative controls because the first spam/scam trolling epidemic is yet to come. There are no spam filters, there is almost zero monitoring over instance content, no remote identity verification before the first contact happens (as I think, maybe I am wrong?), it is very easy to spoof users... In other words free platforms are just underdeveloped in this regard.
By the way malicious actors are already trying the water. So far it doesn't seem an appealing target but only because it is Uncatchable Joe. It can change any moment now.
Hypolite Petovan
Again, what makes Twitter and Facebook so appealing to state-actors is the scale and the easiness of the reach. On Twitter, you can drive by popular figure accounts within minutes of creating your account without any prior research, and your reply will appear in front of the eyes of the thousands of people who will open the same tweet. On Facebook you can simply ask Facebook to pinpoint vulnerable people and show them an egregious political ad that will never be public. None of those behaviors related to Russia's interference in the 2016 US presidential elections are technically possible on the Fediverse.
I also believe that the lack of identity verification on the Fediverse means that it is entirely peer-based. There's no risk of... show more
Again, what makes Twitter and Facebook so appealing to state-actors is the scale and the easiness of the reach. On Twitter, you can drive by popular figure accounts within minutes of creating your account without any prior research, and your reply will appear in front of the eyes of the thousands of people who will open the same tweet. On Facebook you can simply ask Facebook to pinpoint vulnerable people and show them an egregious political ad that will never be public. None of those behaviors related to Russia's interference in the 2016 US presidential elections are technically possible on the Fediverse.
I also believe that the lack of identity verification on the Fediverse means that it is entirely peer-based. There's no risk of a "blue checkmark" verified account being hacked because there's no central authority to deliver an identity verification that could be hijacked. Each user has their own circle of trust that can be close or wide, but there's no way to know in advance and there's no surefire way of making it through as simple a subterfuge as hacking a blue checkmark account and replacing the display name.
And Mastodon, especially through Mastodon.social, may be the spearhead of comprehensive moderation and administrative controls because of its magnitude on the Fediverse. This is something that can be improved without depending on any bottom line.
Alexander
Fediverse currently lacks both. No technical measures in place and many users are just not ready psychologically to handle open web, they still expect it to be Facebook-like where everyone is supposed to follow some Code of Conduct and the police is on call to get everyone who doesn't.
I do believe both can be created for decentralized networks. It has to happen at some point but right now I don't see it happening.
Peer validation exists on any social network. However if it doesn't work on Twitter, it won't work on Fediverse for the same reason. Viral reshares do happen here and they propagate just as well.... show more
Fediverse currently lacks both. No technical measures in place and many users are just not ready psychologically to handle open web, they still expect it to be Facebook-like where everyone is supposed to follow some Code of Conduct and the police is on call to get everyone who doesn't.
I do believe both can be created for decentralized networks. It has to happen at some point but right now I don't see it happening.
Peer validation exists on any social network. However if it doesn't work on Twitter, it won't work on Fediverse for the same reason. Viral reshares do happen here and they propagate just as well. The only difference is the absence of central control which can at some point (hopefully early) intervene and squash malicious influence network-wide. This is generally good thing given how central authorities can't be trusted but it also makes trolling easier and shifts responsibility to fend it off to end users.
I don't know how Mastodon does moderation. How are you supposed to moderate incoming federated traffic? How do you flag objectionable content? Do you end up banning entire instances? What if "attacker" just creates multiple throwaway instances? What if evil guys are hiding among civilians, e.g. malicious traffic comes from big and generally okay node?
As you surely know even Facebook struggles with moderation even though they are so much better equipped. This is not an easy question.
Iron Bug
Hypolite Petovan
Volunteer-run instances still have to tackle the hard moderation problem, but it won't interfere with their inexistent core business model, bottom line and stakeholder demands.
Alexander
In my opinion it is not like that. Unless most instances approve new registrations carefully there is same danger of malicious actors registering throwaway accounts like they do on centralized networks (and perhaps it gets even easier as decentralized platforms don't have privacy-invasive spam/fraud control such as mobile number verification, session tracking and such). Plus there are additional venues based on ability to create new instances.
The best case model is a network of smallish instances with core members knowing everyone within each node and selectively federating with everyone but I am not sure if this is scalable for real mass use. I mean look at some of your own posts from Twitter :) One has 24K intera... show more This is pretty awesome, seriously.
In my opinion it is not like that. Unless most instances approve new registrations carefully there is same danger of malicious actors registering throwaway accounts like they do on centralized networks (and perhaps it gets even easier as decentralized platforms don't have privacy-invasive spam/fraud control such as mobile number verification, session tracking and such). Plus there are additional venues based on ability to create new instances.
The best case model is a network of smallish instances with core members knowing everyone within each node and selectively federating with everyone but I am not sure if this is scalable for real mass use. I mean look at some of your own posts from Twitter :) One has 24K interactions, another one has 60K interactions... This is the kind of participation people expect. This is what Fediverse will have to cope with if it becomes similarly popular.
@Paul Sutton, true, this is most manageable scenario. But the question is if people want it. So far I see how smaller discussion boards and other platforms are all shrinking in favor of Facebook, Stack Overflow and other giants. People want to be with the crowd.
I run a discussion board myself. It used to have rather unique dog breeders community and was very active but it is quiet out there for a couple of years now. Everyone is on Facebook.
Hypolite Petovan
All of these are small obstacles and hindrances that aren't going to prevent all spam on the Fediverse, but it's going to make it innately more expensive at comparable reach versus Twitter or Facebook. Not because of any expensive moderation policies, just because of the nature of the system. More expensive means that it will be less prevalent, less prevalence means that it will be less expensive to manage.
And yes, I agree that the experience of Fediverse is different from Twitter or Facebook in terms of instant gratification and engagement numbers, and I believe it's... show more
All of these are small obstacles and hindrances that aren't going to prevent all spam on the Fediverse, but it's going to make it innately more expensive at comparable reach versus Twitter or Facebook. Not because of any expensive moderation policies, just because of the nature of the system. More expensive means that it will be less prevalent, less prevalence means that it will be less expensive to manage.
And yes, I agree that the experience of Fediverse is different from Twitter or Facebook in terms of instant gratification and engagement numbers, and I believe it's a good thing. I had to work to make my experience on Facebook and Twitter better because the base experience was awful to me. On the other hand, I believe Fediverse software is kindly by design and by default.
The Fediverse will never be appealing to people enjoying themselves on Twitter or Facebook, and that's for the better. It further participate the audience fragmentation making social media influence campaign more expensive.
Hypolite Petovan
Alexander
Twitter audience is about 100x larger. So if we hypothetically assume equal public interest in Fediverse we can assume largest instances will have audiences of million or two which isn't bad.
But I don't think it should be estimated by instance audience numbers however. The effect mainly depends on viral propagation. If you post something on Twitter but don't have enough followers (or luck) it is unlikely to hit large chunk of its audience on its own. So you will use some tactic like one described in the article - and it works on decentralized network just as well provided instances are interconnected.
I do agree that Fediverse is a different place with different atmosphere (although after exploring some of the larger Mastodon instances I am not so sure about them) and perhaps it is our strongest suit :)... show more A couple points here.
Twitter audience is about 100x larger. So if we hypothetically assume equal public interest in Fediverse we can assume largest instances will have audiences of million or two which isn't bad.
But I don't think it should be estimated by instance audience numbers however. The effect mainly depends on viral propagation. If you post something on Twitter but don't have enough followers (or luck) it is unlikely to hit large chunk of its audience on its own. So you will use some tactic like one described in the article - and it works on decentralized network just as well provided instances are interconnected.
I do agree that Fediverse is a different place with different atmosphere (although after exploring some of the larger Mastodon instances I am not so sure about them) and perhaps it is our strongest suit :) I am just not sure that it is software and not people. I don't really believe in technological fixes for social problems but maybe I am mistaken here :)
Hypolite Petovan
And I don't believe Fediverse instances will keep growing vertically (number of accounts per instance) as much as horizontally (more instances). For every 100,000 users mastodon.social will get, how many more spread on a myriad of new instances?
So yeah, some tactics will still work, but not all, or not with the same impact, which may make them obsolete.
And yeah, I'm with you, technology won't solve social problems, but it can mitigates the problem it introduced. We were never supposed to be available to each other 24/7, and this is what Twitter and Facebook provide and promise to businesses and political parties alike.