Why decentralized social networking never makes it
upon2020.com
upon2020.com
Actually, trending was implemented by Gargron, but user outcry against it basically caused him to scrap that feature.
Edit: https://github.com/tootsuite/mastodon/issues/7702
Here's a major user feature for why people switch to Mastodon and Pleroma: better tools for users and admins to manage spam and stop harassment at an individual and community level.
Edit2: It's pretty funny that "Trending" and "Mastodon" got mentioned here, as I forgot that in the outcry this was one of the pieces of feedback:
> [...] maybe you should not be defining the roadmap based on Hacker News thinkpieces.
If you listen to the loud fringe, you'll end up with a product only the fringe wants to use.
Is that a problem? I mean, my first thought when I hear this is, "yeah, broad adoption is totally necessary." But, then I think about it more.
- Is it a problem for indie developers to build games that appeal to niche audiences?
- Is it a problem for software developers to build CLI tools or advanced software that solves very specific problems at the expense of user friendliness?
- Is it a problem for communities to be built around specific problems and for them to adopt very specific, focused codes of conduct or moderation?
Maybe it's OK to have a bunch of small, separate decentralized networks instead of one big decentralized one? With something like Facebook, sure, the point is to have everyone in one place. But one of the big points of systems like Mastodon is that small communities are better than big ones. Small communities are easier to moderate, more flexible to adapt to novel situations, and faster to pivot when things go wrong. Having niche adoption in that scenario could be kind of a feature, in the same way that having niche adoption for a game probably means you're getting something more focused and custom tailored to the people who want it.
It might be that not every social network needs to appeal to the overwhelming public at large. It mind end up being completely acceptable to have a social network that's very specific to one or two communities.
In that scenario, the death of Facebook might not be, "here's a replacement, everyone get off." It might be death from a thousand cuts; "here's a dozen replacements, there's probably at least one you like more."
Edit: that said, there's no reason that an obscure subculture adopting a different social networking technology couldn't lead to that tech becoming universal. Facebook rose through a certain high-prestige subculture adopting it but with the Internet tending to creating different association patterns, high-prestige/"early adopter" may not, in fact, be the leaders of adoption in the future.
I was on Facebook for the network effect - then I got into indie game development, and all of the other game developers and gaming press are on Twitter. So I dropped off of Facebook and moved to Twitter. I still pull up Facebook occasionally to cross-post something to friends, but it's a different tool for a different purpose.
In theory, nobody wants to join multiple social networks. In practice, I don't see people actually all that worked up about logging into multiple sites. My siblings jumped to Snapchat with no problem even when they were still using Facebook. When I started moving away from Facebook I told them to start emailing me - also wasn't a problem, they handled it fine. I run a small Discord server specifically for people who want to talk to me about the stuff I build. It's a really tiny community, but it's still a useful enough tool that I'll probably never build a Facebook page for anything I release.
So those are maybe 3 or 4 tools that have successfully migrated communities off of Facebook. There are maybe 1 or 2 communities left that I have to take care of, and then I could just flat out delete my account. Certainly I'm only checking Facebook maybe 2 or 3 times a week now because of a few stubborn holdouts I haven't convinced to move off yet.
Network effect is definitely a thing - but it's probably not all or nothing. It might be fine to replace Facebook for only for one or two communities. It might be fine to say, "Okay, all of my college friends are on X, but all of my professional contacts are on Y, and my hobby model train community is on Z." And multiple small, federated social networks are likely to be better at serving small communities than something like Twitter or Facebook is.
How necessary is it to you that there be one service that you use to contact literally everyone? I was on the side of "pretty necessary," but the more that I think about it, the more I realize that distributing my contacts across multiple services is not actually all that much of a problem for me. In fact, it's often quite helpful, since it allows me to tailor content more.
Sure, if one has one reason to be on a network, getting off it is fairly easy. But if one has IRL professional-collegues, friends, family, and multiple special interests with friends, then it's hard for all those to switch.
"How necessary is it to you that there be one service that you use to contact literally everyone?"
Well, you want to be able to be-contacted-by a variety of people. Facebook can wind-up indeed just one of several of the networks the average person is on but that doesn't it won't remain the largest and quite possibly not much smaller than today.
Most people are probably part of several communities on Facebook. There's probably not one completely homogeneous demographic. If your IRL professional colleagues are on it, maybe someday they switch to something that's really good at work communication and that doesn't have stupid app spam because it's built specifically for professionals. Maybe your friends and family switch to something that does closed circles and off-the-cuff communication better like Snapchat.
They can go to separate platforms, and you'll just sign up for two services. Those communities (usually) don't need to overlap. So what people call the network effect of Facebook may really be many different, small, isolated network effects, each of which can be individually attacked by separate services.
If that's the case, you don't need a Facebook killer. It's fine to have six or seven completely different Facebook killers that each eat just a few limbs off of it and then go off and do their own things.
I think tech people like to imagine there are users who are very distinct in ideas and taste to them. I'm less confident.
I suspect you may be subtly prompting them somehow, like by making it clear that you find it creepy. Having an opinion like "oh I don't like Facebook ads" is basically free because nobody will expect you to actually follow up on it and stop using Facebook. It's a quick bit of echo chamber virtue signalling that sounds better to people's ears than, "actually I quite like the ads on Facebook" which everyone knows from reading the press isn't the "correct" opinion to hold.
After all people have been saying this for years, and it never affected Facebook's revenue or user numbers before. I don't think their recent bad results are due to a sudden sea change in how people perceive ads.
It's also possible nobody is actually creeped out - but honestly, it just needs to be a sizeable portion of journalists and politicians before you start having legistlative problems. How many people were genuinely worried about nuclear power, for instance?
I pick email: https://hackernoon.com/email-re-skinned-as-a-social-network-...
The ideal would be a "feed" as well as an ordinary list of messages.
Also, it would be nice to have this sort of thing run either on a server or a client.
Our company (qbix.com) made $1MM selling apps for iOS and Mac, and reinvested most our profits over 7 years into building just such an open source decentralized social network. Here is a video about it:
https://m.youtube.com/watch?v=pZ1O_gmPneI
You can find the entire general purpose platform here: https://github.com/Qbix/Platform
And even though we had done all this, we have not yet started to popularize it. We wanted to get the kinks out first.
Examples of stuff you should have at a minimum: user accounts, onboarding, pub-sub, transactional emails and sms, notifications on all platforms including web push, real time updates via web sockets, privacy and access control to content, multi user collaboration, one click invitations, easy installer and updater scripts, caching resources locally, indexing and relationships across domains, authentication across domains (github.com/Qbix/auth) with seamless experience across domains and more.
And then you should probably also integrate with local address books, support uploading and sync from gmail and other sources, support payments on all platforms eg applepay and androidpay, and so on.
I have met teams working on other projects including MaidSAFE and Solid by Tim Berners-Lee. The missing piece is all the user-friendly apps and components. It’s a bit like when the Mac and Windows came out, personal computers finally got much more massive adoption. Or like when the Web became good enough to compete with AOL. It takes time for open source to catch up to proprietary software, but once it does, it is a force to be reckoned with.
PS: If anyone wants to join our project to either contribute (design, development in PHP or JS, or grow the community) get in touch by emailing greg+hn @at@ qbix.com
They don't hate said things, they just don't care about them as much as finding interesting people to follow/groups to be a part of, joining in an active community, finding information about things they're having trouble with, being entertained, etc.
But most decentralised systems seem to be marketed on the assumption the decentralisation side (and the philosophical aims of the service) are the only important thing, and hence don't really appeal to those outside of the tech crowd and those with a desparate need for avoiding censorship.
Secondly, the UI is often worse than the centralised or commercial alternatives, with very few systems (Mastodon arguably being of the exceptions) looking or feeling as modern or responsive.
Finally, plain old network effects have an effect too. The more people who use your service, the more popular it'll become, since more people they know will be encouraged to join in.
But to achieve that, you usually need to put time and money into marketing your product and attracting as many early users as you can. That's hard work, and it's not fun for a developer.
What is fun for a developer is to play around with programming languages and data structures and protocols and frameworks and all the new shiny stuff they saw on Hacker News. So all the effort goes towards making a technically fancy system, whereas none of it goes towards actually getting people to use said system, or getting the word out in general. Get away from the code editor and get into community building.
(Though to be fair, that last one is also a problem with most new social networking sites in general, decentralised or not. Every modern Reddit or Twitter alternative feels like a programmer's fun pet project rather than a service they're marketing to people)
Either way, I think that covers it. The lack of certain features like trending content doesn't matter as much as how these systems don't market themselves correctly and put little work into the things people actually care about.
>But most decentralized systems seem to be marketed on the assumption the decentralisation side (and the philosophical aims of the service) are the only important thing, and hence don't really appeal to those outside of the tech crowd and those with a desparate need for avoiding censorship.
I think it used to be that way but (at least from what I've been hearing) a lot of non-tech "normal" people are starting to realize the problem with censorship. They might not see that it's a symptom of centralization yet but they definitely are starting to care.
Once upon a time there was Friendster, Orkut, MySpace, Facebook... Digitizing your connections and friends was a new and exciting concept. Now everyone has been connected and we've seen it's not always a great thing. I don't think it's interesting enough of an experience to replicate rebuilding a network elsewhere at this point.
However,
"The more you tighten your grip, the more star systems will slip through your fingers"
Ie. the more oppressive and tyrannical Facebook (and Twitter, and the US government) becomes, the better the alternatives become. There's a reason the shit Snowden uncovered was meant to be kept secret. It wakes people up.
As in blockchain?
What I'm talking about is something like what happened with Retroshare. They created a friend-to-friend social graph that could facilitate file-sharing, chat, etc. Then the users quickly figured out that bootstrapping an F2F network is extremely difficult unless you are in the same physical proximity as your peers (e.g., a college dorm). Otherwise there is no easy way to do key exchange.
So the idea was introduced that you could have these bootstrap servers (not sure what they called them) so that all the new users who couldn't download anything could use them to do a key exchange with at least one other random peer, thus entering the network that way.
But such a user is no longer building an F2F network. Instead it is some kind of friend roulette network where you trust random nodes only because the software would be unusable otherwise. Worse-- at the time Retroshare data wasn't onion-routed so the user was shuttling data unencrypted data among untrustworthy nodes, and encouraged by other users to shuttle share data for friends-of-friends because that would presumably improve your connectivity.
At that point everyone except little LAN F2F islands would have only been slightly worse off with a Facebook-style app where Retroshare devs surveilled the network and spent all their time denying access to whatever they considered bad actors.
It goes without saying that FLOSS devs would never build a Facebook-style app. But there is at least one other piece decentralized app I know that did essentially the same bootstrap roulette as Retroshare.
The only way that can happen is if users and devs feel like they are doing something righteous by using such a piece of software, rather than just leveraging a tool to do a job.
Imagine you are the “resposible adult” who takes care of an elderly relative, or a child, with a complex, chronic disease. You are probably middle-aged, female, family-minded, and have your hands full. Other than juggling your family and your job, you also need to find the right specialists for young Bob, organize rides to and from treatment, get second options, research conditions, get prescriptions filled, supervise that Bob actually takes them on time, keep all the other relatives updated, draft them for particular task (“can you take him to the physical therapist on Tuesday”), listen in to some on-line self-help groups, being a Mom of course etc. etc."
I was just in this situation taking care of my mom after her stroke and I don't think this kind of service is very compelling. Most of mom's friends don't know much about computers in general. Some don't even own a computer. taking some time to make phone calls, text messages, email, and letters is well worth it. It absolutely has to be done personally (well I think so at least).
Yes there are heavy mathematics behind the structure of both types of network, and it's true that without centralization the network itself can't reach a certain limit. But this is a double-edged sword, this is only true to the whole system's perspective and it's completely different on the node (user) level graph-theoritically. Also, the article forces certain needs on the users. There are lots of us who DON'T want to magically find anyone or be found magically with a keyword, nor interested in "trends" - this is heavily misleading for we still don't have proper metrics for it and won't have one for a while because the problem implies building fault-tolerant reputation networks (that are decentralized by the way)
It's a matter of time.
Of course, I don't see how decentralization helps with privacy though - instead of trusting just one vendor, you end up trusting several vendors (either the service providers in the case of federation, or the app writers in the case of full decentralization) assuming people you communicate with are using a varied set of vendors (i.e. you don't end up with a dominant vendor like gmail).
1. Mastodon is closer to Twitter than FB. Things are public by default. Users even have an RSS feed.
2. Get off targeted-ad-driven platforms and the Cambridge Analytica business model goes away and this stops being an issue.
I don't know if sock puppets can be eliminated on any large platform.
Point 2 is only true if the data has no other use. Political parties might find a lot of value in discovering Mastodon users whose posting habits indicate that they are currently voting one way but are disatisfied with that party. Insurance companies might also be interested about users being part of groups for people with certain mental or genetic diseases. I can imagine a sort of "reverse glassdoor" site where employers can pay to get a complete profile of everything an applicant has ever done or said online. The CA business model works very seamlessly with ad targeting, but that is by far not the only use for such data.
People could argue that Facebook effectively has a monopoly on social media so that is a "centralised" source since you have to go to Facebook to access it.
Yet pockets exist (e.g. HN) outside of Facebook, ergo decentralised.
I do agree though, it is not distributed which has a different meaning.
Checkout dat project, ipfs et al for that.
Next question.
(My prior assumption was that IPv6's lack of strict need for NATs means that in an all-IPv6 world it is easier for anyone to run a server, versus today. But I don't specialize in networking, so I'm all ears for other takes on this.)
- Decentralized YouTube (https://d.tube/) is almost in the Top 10K most popular sites in the USA (https://www.alexa.com/siteinfo/d.tube).
- Decentralized Reddit (https://notabug.io/ , https://blubit.space/ , etc.) has ~1K daily uniques.
Both sites are built using our https://github.com/amark/gun ! Soon they'll even integrate same private/public keys so you can reuse your same account across many different dApps.
This will only strengthen the network effect, and make decentralized apps take off even more.
Also note, Mastodon is federated, you still "belong" to a server.
I'm not sure that makes the point I assume you are trying to make.
Centralizing that network is how you profit.
Humans are greedy.
I am sure some other technologies will come up but most of those solutions are trying too hard to be the next thing that replaces all the previous technologies.
Centralized solutions are not impeded by moral factors; they can optimize for user acquisition and profitability instead - They can corrupt the government such that new regulations work to their advantage. They can promise key external decision makers lucrative future contracts or positions in exchange for favors (e.g. onboarding lucrative customers)...