The case for the decentralization of online forums
viksit.com
viksit.com
What's wrong with HN ? What's wrong with metafilter ? What's wrong with letsrun or doom9 ?
The problem is the mistaken notion that these can be big businesses.
If you can let go of the economics there is no technical - or usability - hurdle.
We know this because we're already doing it right here.
A billionaire is choosing to be a responsible owner/patron of this platform, but ultimately this platform answers to him.
HN is a "good king" platform, our king has invested in public works that we benefit from and we appreciate it.
Hacker news is good insofar as it accomplishes our kings goals which align with our own, but we still have a king, even if he is a benevolent one.
It doesn't take a billionaire, even if it happens to be the case.
It would be fascinating if they wished to publish the operating costs, but even without that, IIRC the hardware was quite minimal.
And I’d say HN was exceedingly lucky find these excellent moderators. Not sure you could replicate that success no matter how much cash you had to burn.
Don’t get me wrong, I think HN is great the way it is, I just don’t think paid moderation is a requirement
Many people learn that HN runs on a single dedicated server and figure it costs ~$1000/month in operational costs. But my point was just that this ignores the moderation, which is a crucial feature of this community.
Of course you could try to recruit volunteer moderators to do the same job. But finding and training and managing them is not easy. Nor is dealing with ideological or interpersonal conflicts. None of this stuff is easy, and just because volunteers can and have done this work, doesn’t mean that “just have volunteers do the moderation work” is a solution to successfully running a big online community.
In the age of people paying to belong to a community via Patreon, Discord, Twitch, etc. it shouldn’t be hard.
There’s a lot of road ahead in terms of learning how to develop transparent and Democratic online communities… but as the shitification of the internet becomes critical, it’s a necessity.
* someone needs to pay for it
* whoever is doing and managing that can turn bad and want more.
How do you solve that to be immune to the reddit/stackoverflow disruption ?
Even if it could be covered by donations it there is still problem of any inevitable management of it deciding to push it in direction different than community wishes.
> We are holding this discussion on a classic, centralized, web1.0 forum which - from my vantage point - is one of the most valuable and enriching forums currently online.
Not web 1.0. Usenet and mailing lists(mostly) died for that, as did many "traditional" forums.
> We know this because we're already doing it right here.
Ok, now look up who is paying for that.
This is centralized thinking. In a decentralized open protocol system, everyone pays for their own way, but it's very cheap at an individual level. There's no central "it".
> Not web 1.0. Usenet and mailing lists(mostly) died for that, as did many "traditional" forums.
I'm on several email groups where I've been a participant since the very early 90s. Every so often there's the "oh have you seen this new shiny thing, let's move to that!"
Every single one of those new shinies have disappeared, but the email group lives on.
Why? Because email is 100% decentralized, nobody owns it, there's nothing to own. It's an open protocol. So it lives on forever.
Also, I still read Usenet on most days, just like I've been doing since the late 80s. It's alive (if not very active) because nobody owns it.
I wish we could collectively learn these lessons. Centralized systems cannot exist for very long.
It is if you want to have central page for people to visit. Someone needs to pay for hosting that else you will have no users if everyone needs to host their bit
> I'm on several email groups where I've been a participant since the very early 90s. Every so often there's the "oh have you seen this new shiny thing, let's move to that!"
...do you know how mailing lists work ? It's a central server that distributes it to the members. Pretty light one, sure, but it is still something someone have to host and manage. There is no "distributed mailing list"
> Also, I still read Usenet on most days, just like I've been doing since the late 80s. It's alive (if not very active) because nobody owns it.
Central servers managed by people again. Yeah they are federated but still rely on a bunch of institutions to throw some servers at it.
There is no "central page" for Usenet.
> ...do you know how mailing lists work ?
HN isn't the place for this kind of discussion style. If you insist on asking, I have been running email servers personally and professionally for 33 years now and have worked in enterprise tech companies specializing in email. I have a resonably good idea of how email works.
> > Usenet
> Central servers managed by people again.
No. You can run your own Usenet node just as you can be your own email server. That's why these protocols are fully decentralized. You also can rely on someone else's Usenet server or email server, but that's your choice to make if you wish. You don't have to do that to participate in the ecosystem. That's the strength of decentralization.
>HN isn't the place for this kind of discussion style. If you insist on asking, I have been running email servers personally and professionally for 33 years now and have worked in enterprise tech companies specializing in email. I have a resonably good idea of how email works.
I asked how you think MAILING LISTS work, not how email works.
And you obviously have no idea how MAILING LISTS work
THAT is a fundamental and interesting (almost philosophical) question: should a forum/website/media FOLLOW "community wishes" (democracy-style) or FOLLOW a vision even without traction (benevolent or bureaucratic dictatorship style) ???
BOTH are good AND bad in a way or another...
* FOLLOWING COMMUNITY WISHES will ensure more and more people, but lower average quality (bazaar style) because everybody will talk about what interest him in the end... so hard to keep a focus. Moreover, community wishes is prone to forum "take over" by subgroups (think for example "extreme right" - for whatever that means, or hackers, or...)
* CHARTER will ensure that a forum has a single "topic" (even a niche) but the community won't grow as much, it need strong moderation, people will regularily clash/disagree and leave about the rules... but quality will be higher (more focused and specialized)
But there needs to be some push for a certain direction for the social space. Just maybe not "dictator for life" as is on pretty much every subreddit. I wonder if some kind of elected "democracy", with people voting having their vote weight tweaked by "contribution level" to the community. Still probably prone to some biases but at the very least not in the "lowest common denominator" direction.
Decentralize by making small, simple, isolated things. Not by making complicated interlinked things and trying to form a larger network. The larger network doesn’t necessarily have value.
> The people using your software, even if you own it and pay for it, have rights and will behave as if they have rights. And if you abrogate those rights, you’ll hear about it very quickly.
> That’s part of the problem that the John Hegel theory of community—“community leads to content, which leads to commerce”—never worked. Because lo and behold, no matter who came onto the Clairol chat boards, they sometimes wanted to talk about things that weren’t Clairol products.
> “But we paid for this! This is the Clairol site!” say the sponsors. Doesn’t matter. The users are there for one another. They may be there on hardware and software paid for by you, but the users are there for one another.
Forum members only have "rights" under some definition of rights where they grant you absolutely nothing.
Exactly that. A community communicating can't be the huge profit center that a corporation needs it to be, without it being ruined.
> We know this because we're already doing it right here.
Exactly.
Do you know the operating costs of HN? How can you judge it takes a billionaire to subsidize it?
They do retain at least two full time staff members though, so that’s going to be 10k/mo+ at the minimum.
I can't find it right now but I remember some of the infrastructure details have been posted in the past and it was quite minimal. It doesn't take much to run an efficient site like this one.
Hard to let go of concepts like FIDONet from decades ago. There was merit there, just not enough technology to do it right then.
If it's more of a protocol like IRC or Usenet, it could work very well for both small and large communities.
I think mastodons biggest mistake is pigeonholing a Twitter replacement.
To my mind HN is in that classic genre of single focus communities with a deep history of being the place to go to discuss topic X.
It fits in next to sites like flyertalk, woodweb, bodybuilding.com and a million more.
I think the web was definitely better when every topic had its own little community and if Reddit imploded tomorrow and that happened again I would be thrilled.
Why isn’t it good enough to have a whole bunch of idiosyncratic privately run discussion forum communities spread across many different websites and servers and so on?
We had that before Reddit and many examples still exist, it seems like it was closer to the ideal than Usenet ever was.
I think a ton of random different communities with quirks and different approaches is the best option of all.
HN isn't a business at all. It only loses money, and in return generates some sort of goodwill or authenticity for its owners, who are extraordinarily wealthy. All problems can be ignored if we can always rely on rich people to give us what we want.
In addition to Hacker News making them the most well known startup accelerator in the world, who people will apply to from all over the world, it also provides benefits for their incubees, as they get free advertising on Hacker News, and presumably this makes their investments more likely to succeed (although I'd love to see a comparison of the historic portfolios of various startup accelerators)
Weed need a decentralized equivalent to Usenet.
Nobody wanted to pony up. It's that simple.
Nowadays, a bunch of people have gigabit to their house. Colo is $400/mo for a cabinet and gigabit. 3U servers are under $500 and probably under $200 if you dig. I can go on.
You could resurrect Usenet easily, but nobody is doing it. Why? Because it's a pain in the ass. People forget that a lot of the early internet stuff relied on a bunch of people providing free labor (generally at universities) to keep it all afloat.
Usenet was decentralized, but not in the current sense of the term IMO. When I think of 'decentralized', and what I meant, was something offering anonymity and resistance to takedown orders, which are things it didn't have real protection against.
I think the free labor issue could be circumvented by users having their own killfiles, and a fediverse approach of only syncing with 'good' servers.
Groups was contemporaneous with Dejanews. But Dejanews was basically going under when Google bought them. If it hadn't been for Google, Dejanews would have gone poof and we wouldn't have had any archiving of Usenet stuff at all.
That is ... Usenet. That's exactly what Usenet is (not was, because it's still there, even if not many people use it. I still read it on many days.)
I'll note that everyone who has claimed to build a better email or better Usenet for the last 4 decades has invariably failed. So there's that.
In the short term, it's very easy to build a more sleek and fancy communications platforms. Hundreds have been built and they can be nice for a while. All of them eventually fail due to the inherent limitation of proprietary profit-seeking closed systems.
Of course I'm not claiming NNTP and Usenet are perfect. Sure we could tweak them for this and that improvements. But the fundamental principle of decentralized peer to peer distribution is the core strength. Without that, any solution will ultimately fail in the long run.
Whether the law is good or bad is another story but as long as that law is in force this sort of requirement is a mirage.
Server operators don't exist and live in a vacuum, when they get services blocked, servers seized and prison sentences imposed for not cooperating with lawfully issued orders (such as a subpoeana to identify a criminal using the service), then what? All that is happening already.
If you want to deal with unlawful orders of this kind or abuse by the authorities then the technology is the wrong battlefield to fight this. This has to be done in parliaments, not server rooms.
Take down orders are resisted by having it sufficiently decentralized. It may be possible to take down one server, but taking down every server (even trying to take down every NNTP server in today's world) is a significant effort.
Usenet is a perfectly acceptable decentralized system for sending messages to each other and the world.
Stand one up with the groups that I'm interested in and I'll point my copy of trn at it.
Usenet, I think, has a strong hierarchy structure rather than just being a 'web', although it is that as well. So if the takedown order is directed at a server in high enough position, it is essentially removes it from all of Usenet - and it's not just takedown orders, but actors with the ability to remove content unilaterally.
Usenet really isn't sufficient simply because of how old it is. We can do much better, and I believe there are already numerous superior systems that exist but simply have not been adopted yet, likely due to ease of use and/or lack of need, the former can be improved, and the latter will happen naturally.
There were some sites that had bigger pipes and more tolerance for storage requirements than others, and it was thus easier to peer to one of those big ones to route posts that would get out everywhere else faster - but there wasn't any hierarchy as such.
The hierarchy of unset was in the naming - not the organization of how things were routed nor for a "who controls that." There is no lumber cartel forcing spammers to print out things and buy paper.
However, if a site misbehaves often enough it gets shunned sufficiently through per user kill files and admins not accepting posts that originated or passed through some server.
If you were to take down a big server that lots of people routed through, they'd route around it or another would take its place. Taking down The Pirate Bay didn't stop bit torrent.
My memory was that smaller ISPs would peer from the larger ISPs, that's the type of hierarchy I was thinking of. Maybe not in design, but in practice.
http://usenet.trigofacile.com/hierarchies/index.py?status=pr... and https://web.archive.org/web/20220815151921/http://bofh.taron...
Peering to larger ISPs was more of a matter of convince and that is often part of a business arrangement for a company and a client.
Another use of usenet back in the day SGI had its bug tracker write out to news. A Usenet reader was one of the preferred interfaces to creating and commenting on bugs internally. Furthermore, exporting the bug tracking system to a partner was "simply" a matter of adding them to the feed and making sure that they didn't route bugs from their server to others.
The point is that the "little to big to bigger" was just convince - not code or enforced in any way.
Many university CS departments would peer with each other back in the day. Recall that news was originally across UUCP - the same mechanism that email was sent. If you wanted to send email to someone at another university you'd much rather have them as a peer so that it was just one hop rather than going some other route that could take days to get there. And so various hub and spokes evolved. Everything in the University of Wisconsin peered with Madison. Everything in Illinois peered with UIUC and Madison peered with UIUC. Note the word 'peer'.
And if UW Whitewater wanted to get their email to UIC faster, then they would peer too. Would UW Madison complain that it's not going through them? Not at all! It meant their phone bills would be less.
But the same thing may well happen again, and that's a kind of flaw. A truly decentralized system to combat the things we want it to combat should enforce or be resistant to any de facto hierarchies.
If you try to enforce this, you will get mastodon.social and a much of larger-smaller sites rather than hundreds of even smaller federated sites.
It becomes too expensive for the small sites to federate with all of them (you'll note that turn key hosted instances of mastodon often have limits to how many different sites they can federate with).
The famous quote of "The Net interprets censorship as damage and routes around it." is from Usenet. It took very concerted efforts in order to moderate or completely censor a badly behaving site.
I would contend that usenet did a very good job of handling its robustness of delivery that didn't involve spending too much time for a given site to try to peer everywhere.
Who would you issue a takedown order on a Usenet post? Remember it's peer to peer distribution. There is no central repository.
An analogy is to say email is centralized because may people use gmail. Yes many people do but that's in no way necessary. Everyone can run their own if they like. For the entire decade of the 90s through several employers, I never was part of an organization which even had a central email server! Everyone sent and received email directly to/from their personal workstation.
Reddit didn't come along until 2005, and the way I remember it is by then usenet--binary groups aside--was was pretty much dead.
Can't buy ads for a hot... single AI farms in your area.
It doesn't. In most cases, like Reddit or Twitter or Mastodon or whatever the issue isn't that something is centralized but that things actually do cost money to build and run.
Even decentralized architectures like Mastodon need that - and very few people actually want to pay for it. That is why the various efforts to monetize the content, lock down APIs and push in your face ads happen.
The VCs that were paying for Twitter or Reddit aren't willing to do that forever - and neither of the companies are actually making profit or even not making a loss. Whether that is because the business is poorly run or some other problem is secondary but unless they start to make money somehow, they will close down at some point. It is that simple.
The article - like most that spout these 'decentralization solves everything' (same like blockchain/crypto/web3/metaverse/etc.) mantras completely ignore this problem or handwave it away. As if a decentralized system ran on fairy dust and unicorn poop and didn't need to pay for servers, electricity, wages, etc. Sorry, but that's not how the world works, folks.
If you don't want to pay money for a service then that leaves the operator with either ads - or has to pay it out of their own pocket. No "defi" or "fedi" solves that - only makes things maybe more resilient when one operator goes bankrupt or rogue. But the rest still have to pay their expenses somehow.
Also the entire premise is BS - as if online forums were somehow centralized and everyone was prevented from running their own server and community, using their own rules (no "censorship" or "cancel culture"! Yay!) and money (a-ha!). Reddit, Twitter, Facebook etc. aren't the only places where one can have a discussion or post information.
I think this is a core assumption that I'm not sure I believe in.
Works for the public deserve public (distinct from government) funding. There are many things that say "if you don't give us money, we can't continue to exist."
"We need enough money to function" is a completely different sell than "we want to 1000x our investors investments."
Reddit's enshitification[1] is a function of needing to please investors. Many people are very happy with old.reddit.com, yet reddit paid for ux devs to implement dark patterns because that's what investors demanded.
I don't want to pay for signal, but I pay for it. I don't want to pay for Wikipedia, but I pay for it. I don't want to pay for my local radio stations, but I pay for them. I don't want to pay for my politicians to represent me, but I pay for that. I don't want to pay my taxes, but I've traveled enough to see what happens without government institutions.
I think people are willing to sacrifice money for purpose. Reddit and twitter were great free speech platforms before billionaire malfeasance. These platforms threatened authoritarians world wide. That is purpose I think people would financially sacrifice for in both external funding, and loss of income.
I think it is pretty clear that the next great social media platform will be a 501c3 and they will ask for money in the same way that radio stations do.
Maybe some people want that. Your "we" is putting in a lot of work.
It's almost as though the large-scale abandonment of old-skool small-scale, independent domain-specific web forums in favour of huge centralised platforms was a mistake...
(Just out of interest, can anyone give me an idea of the typical running costs of, say, a phpBB or vBulletin web forum with a couple of hundred users?)
I also run a BBS on a Raspberry Pi that plays host to a few hundred DAU, that engage in chat, forums, a MUCK, and the occasional file transfer which runs on my home internet connection.
I am sure someone will jump right in and tell me that I am doing it wrong and I can host everything on an AWS micro VPS for $4/month, but the amount that I pay is for my server. I also suspect someone will jump right in and tell me I am not allowed to host a server on my home internet connection.
But the SEO and marketing costs to get any kind of traction these days are in the hundreds of thousands of dollars. That’s the real reason we can’t have decentralized communities anymore, the big guys have formed oligopoly. They control the market and prevent any competition by owning user attention.
Wikipedia seems to be scaling fine, even to other languages (correct me if I'm wrong).
When people have more money to spend because they’re not paying half their take home pay on rent or a mortgage, they’re generally more amenable to paying for things.
The other half of the solution, is that most tech companies have got to stop trying to become ridiculous never ending growth machines like Facebook, and just become boring utility companies. No one needs Facebook to do all the shit it does - it just needs to be a digital yellow pages. No one needs reddit to be flashy - it just needs to function as a forum. Keep the price low (e.g an hour of minimum wage in whatever country you’re accessing from for a month for a no ad version) and you will get people signing up.
Increasing housing supply means decreasing homeowner capital and harming the major investment vehicle of a generation. That's politically untenable. Sadly, the pain will be felt by us and our children until we are able to exercise political power, which will require harming our parents retirements. Maybe the answer is renters unions, how many people can the police evict at once?
Young people don't understand their rent is high because they don't vote or exercise political power.
Go to any country in the world with an upper-middle class neighbourhood and ask them if it's alright to plop an extra condo or apartment building in the middle.
The only way you'll smoothly get permission is by providing huge, costly, benefits to the local households.
Otherwise it would involve a long drawn-out court battle, or some scheme with the local political potentates, pretty much everywhere where there are such neighbourhoods.
If supply goes up, existing homeowner value goes down, the amount landlords can ask goes down. I don't see any reason it is more complicated than that, the entrenched interest is obvious.
Doubling the housing in a neighborhood could halve peoples entire net worth.
So even if the average home prices slightly increased it would still be the same dynamic.
Every Fediverse instance I've used - and I've used quite a few - has been either so small that the total cost is a few dollars a month, or fiscally sustained by donations from users.
The problem is when people want to make a profit on their internet community. Then, no amount of community support is enough, because nobody wants to donate to your investors.
And that is somehow invalidating what I have said? Sure, if you have 10 users then your costs are going to be negligible and you can probably sustain it from your own pocket or donations.
But good luck running a larger/popular service like that. There are reasons why the more popular Mastodon servers had to stop accepting new users after the Twitter/Musk fiasco - it was simply not sustainable for the one-man operator crews.
Because it's a federated system, there is no need for instances larger than that (and in my opinion, they are a net negative for the network).
your point is well taken. in the article, one of the main reasons i say that while it’s important to not get caught in the crypto jargon - it’s also important to not ignore the technology paradigm behind it.
giving people an incentive to run a server by the populace that uses it is a well defined and well known way to run decentralized services. staking and validating are concepts that have gotten very sophisticated - but the end result of it doesn’t need to be a “crypto token” bag holders madness. it can just be USD or USDC.
one can imagine a world where server hosts stake a sum of money, and may lose it if communities that start there run amok in some way. the users of the system also stake money for privileges (say, like discord nitro - for nicer emojis and reactions).
this pool of money - if it can run discord - can run decentralized services.
better still - it allows for a pool of money to give those who do community work and engagement a way to get incentives.
lastly - rather than ads, businesses could give end users the cost of their CAC that would otherwise go to instagram or reddit.
Creating an account per decentralized unit is a high cost. Fighting abuse of a platform on the scale of decentralized units rather than the platform as a whole is a very high cost. How does a decentralized platforms perform against an adversarial attacker (spammers/nation states/etc)?
You have addressed the idea of centralization by talking about what it does well, but you have failed to address what decentralization does poorly. Namely, turning O(1) problems into O(N) problems.
How do decentralized organizations handle conflict/how do you prevent schism?
Email was very decentralized, but spam never got dealt with until email was nearly entirely centralized on gmail. So why could e-mail never successfully fight spam before centralization?
I don't think your post attempted to explain the forces that led us towards centralization in the first place and what the benefits of that centralization were.
I don't think decentralization is obviously better or worse than centralization, I think there are a series of tradeoffs that make different styles of platform better under different contexts. I think a "good king" centralized platform will beat a decentralized one until it doesn't (which is what we are experiencing right now), which means that during the reign of the "good king" the cost of decentralization is higher and therefore unappealing.
how much does HN cost to keep running, again? didn’t Stack Overflow notoriously run on hardware not that different from what some passionate devs keep in their home?
the hardware is cheap enough that the BoM cost angle is debatable today and will only be less true over time. the other costs (time commitments) fall into well-known buckets: moderation, and development. Reddit, Discord, etc are already moderated by unpaid users. lots of open source development already happens without money or an expectation of any return on capital.
you’re sort of implicitly denying that bbforums and such could ever be a dominant mode of online community, despite that at one point they were. if community aggregators (Reddit) is what displaced many of those independently-run forums, then today’s federation is something of an attempt to force that role back open and allow independently-run forums an even playfield.
What decentralization does is distribute this cost and complexity to more granular and manageable levels, so that instead of a single company having to carry the maintenance burden, a number of individuals or groups can do so instead. Of course, if the individual instance or whatever distributed unit grows too large itself, then you run into the same problem on a smaller scale. But it's more manageable, each instance can decide how it should be handled, and in practice most instances won't grow to have this problem.
That said, decentralization is not meant to solve the cost of running and maintaining infrastructure. That is an inescapable fact of the internet. What it does address is the problem of large companies controlling vast amounts of the internet, and getting to decide how information is shared online, and who gets to have a voice. Twitter and Reddit are prime examples of what we don't want to happen again, and there's no way to accomplish that with a centralized service.
A distributed, open-source, collaborative approach to forums could work. It did work in the past. Somewhere in between usenet and reddit was a plethora of small message boards run, frequently, on open source software. Many of the same people who ran those message boards now run reddit subs only because its easier and because users gravitate towards big, centralized places like reddit.
A return to a multitude of little message boards seems unlikely at this point, but migration to a monolithic and distributed open-source alternative to Reddit or Stackoverflow could happen. We just have to reach the point where enough people appreciate that there is a need for such a thing. A lot of people probably thought reddit was some kind of non-profit public service, but now they're going to have to wake up to the reality that Reddit is a business and they are now determined to make a profit even if it means driving their service into the ground. (From a business standpoint, a few years of actual profit leading to collapse is better than decades of producing no profits, especially in a sector where disruption could easily lead to collapse in that timeframe anyways.)
Moderation is a key problem to solve. Moderation is what keeps forums on-topic and free from spam, hate, and other trash, but it's also frequently abused to push bias or power trip. Existing social media sites typically do a poor job of providing transparent or accountable moderation. You typically don't know who the mods are, what they've done, what they're doing, nor do you have any way to challenge their actions in a meaningful way if you have a reason to do so.
Getting moderation up to professional standards without paying for it would seem like an obstacle to a free social media service, but today's social media sites don't pay their mods either. If you provide anonymity and don't demand accountability, it turns out people will just volunteer for the job! As such, a lack of an elegant solution to the moderation problem should not be viewed as something that blocks alternatives to the online forums we have now. They haven't addressed the problem either.
The replacement for reddit doesn't have to be perfect. It just has to be there.
However, there's two big things standing in the way:
1. The vast majority of the population - I'd say at least 99.99%, probably more - is only interested in jumping on the next big thing. People who liked and enjoyed being part of mailing list communities, handcraft html webrings, AIM, self-hosted forums, MUDs, etc. would probably be happy to go back to those if they were suddenly popular again. But when they went out of popularity in favor of current social media trends, everyone jumped ship. Everyone's following the crowd.
2. Online communities tend to not place any limits on output, so a few hyperonline individuals dominate the conversation (and in cases where moderation occurs, have methods to control the community). Hyperonline individuals don't tend to be the best socialized ones (people with healthy lives aren't spending all day on the internet).
With regard to #1
The vast majority of the population are seeking to fill some need when they get online. Its not about popularity, popularity only provides exposure.
Its whether that need is being satisfied, both at the time they choose to jump ship, and continuing on into the future. Also, almost everyone born after the 90s has been indoctrinated prior to the age of reason that they can go online and find anything they might need, addiction triggers and other psychological traps await. There is also, as always, the ever present coercion through concentration of business sectors into a few entities who play shell games and limit your agency and choice. In some places you can't apply for college anymore using paper, its all digital and digital systems break when they aren't properly designed for resilience; these result in less opportunities acting as a filter and coercively limiting those who fall into the implementation cracks with no other choice.
#2 is purely normative, and neglects abuse which may include jamming communications or something more subtle but still malign. This view is filled with supposition, and makes broad over-generalizations, its non-sequitur.
Decentralization doesn't tackle the "how do we long term maintain a social network" problem. No decentralized protocol does. Which means, ultimately, these servers are slowly being turned off by their owners in 10 years or whatever the case, and we're back at it with "how do we solve the current problem of online forums disappearing?"
Something like Facebook obviously costs a lot more to run, as it stores photos and video and also provides each user with an individually generated feed. In contrast HN is rather lightweight and basically serves mostly static cachable content to all users. I believe it was stated somewhere that the entire HN comfortably runs on a 64GB 4/8 bare metal server. On the extreme side I host my personal static web site on my home DSL connection fronted by Cloudflare, as the CDN does all the work and the bandwidth used is minimal.
To the people running these new communities the software cost is low as they're based off an open-source service of choice (Lemmy, Mastodon, etc.), and as we saw on Reddit moderation can be done by volunteers. Are the remaining costs for admins, hosting, etc. feasible for a non-profit with some user donations? And can we create more efficient platforms with a plainer style that will minimize the server and electricity costs?
90% of costs of places like Facebook are wages and various other expenses that you need too once you are running operation at a similar scale.
E.g. you do need to pay for a large team of lawyers because you are dealing with a ton of copyright takedowns, cease and desists, even lawsuits every single day.
The same with large team of moderators/content reviewers because in many countries people posting illicit content (with the definition of illicit being different by country!) may get your business blocked/shut down in that country. Including seizure of domain, any hw you may there, etc.
Add to that various regulatory compliance (e.g. the the GDPR once you are above certain size), engineers to run and maintain the systems, security team to deal with any attacks that happen 24/7, accountants, secretaries, even janitors and cleaners to keep the offices habitable.
Those are the actual costs. Not your puny server in a closet.
I pay for a certain amount of internet bandwidth - and depend on it! I work from home. Now I'd be willing to part with some amount of that, particularly in off-hours, to support a decentralized service - particularly a forum.
But I want to know what my commitment and loads are going to be upfront. I remember playing with IPFS and being extremely annoyed that just running the node ate something like 10mbps of my internet connection just by existing and doing nothing.
As far as I know none of the decentralized service ideas have a good answer as to how to spread the load or limit the load they'll impose on people hosting them - nor an answer as to how one can "help out" if they want (various "Raspberry Pi that takes care of it" proposals are not a bad idea).
But it is rather an exception than a rule, esp. at the scales we are talking about (e.g. Reddit or Twitter), with world-wide audiences with varying legal and financial regimes.
Look at newspapers. Look how many people actually bought subscriptions to popular services like Twitter, Instagram or Facebook - vs how many are using them for free.
* The founders take minimal amount of VC money to get things running
* There is no self-hosting of anything unnecessary
* Everything is hosted on-premise (not sure if this would cut costs in the long run)
* Serve the users the least amount of ads to make it worthwhile
So basically run a decent service that you can sustain for decades. You won't become an IPO billionaire, but maybe get a humble couple of millions.
Perhaps it can't be done, and enshittification is the only way. In that case, I don't see any other way than decentralisation run by donations to get us something decent.
1. Regarding the monetization scheme, nobody has ever gotten anything to really significantly work beyond advertising. Yes, there are some niche solutions that can support a dev team of maybe 1, but at the end of the day most people aren't willing to pay with anything except their time.
2. Social networks that don't have at least some level of top-down moderation always seem to turn into complete cesspools that fail. You could say reddit relies on their volunteer moderators, and that's totally true, but even these moderators must abide by rules, reddit has terminated numerous subreddits in the past for breaking the rules, etc.
I think people should fundamentally accept that humans can't "self organize" at a very large scale on the Internet. I think the reason for this is that anonymity (and I obviously see the irony in me posting about this) completely breaks our normal human systems of "checks and balances". I honestly think Mike Tyson said it best: "Social media made you all way too comfortable with disrespecting people and not getting punched in the face for it." There are just too few downsides to being a complete asshole or troll online, where in the real world there are natural consequences for acting that way.
2) Gotta quibble. I've seen (on reddit) plenty of top-down moderated cesspools. And several totally hand-off utopias too. I think the trick is 1) a subject matter that touches no nerves 2) a high enough bar for entry/respect that excludes the average riffraff.
Also we need a way to punch people in the face over the internet. Best I can think of is everybody gets a permanent dossier, publicly shared, that records all your good and bad deeds (and your judgment of good and bad too, for weighting). This makes assholish behavior muchbmore detrimental to a happy online social life.
re (2) - I wish we had a way to do subjects that touched no nerves on the internet.
that said - an identity system which keeps track of your activity (anonymously, pseudonymously, or real identity) is a great suggestion and something we've implemented in our protocol.
i've spoken to maybe 10 folks who are reddit moderators. every one of them does it for the fun - but if they could be directly rewarded with something more than "internet karma points" - they would take it, and feel good.
to your second point - 100% agreement there.
I mention the grassroots moderation in another reply on this thread, but the concept of a subreddit and the control of its creation is a very important one to consider. I don't think this problem has been solved, but I have some theories on how it could be done.
by making those who start a subreddit having to put a "stake" in the ecosystem after they hit a particular threshold - and having everyone who has a stake be able to vote on their future - seems like a reasonable way to do it.
Sharing common work is possible in decentralized forums, but only with somewhat standardized interfaces. Of which activity-pub and mastodon et al are sort of just on the starting path. Every forum choosing among a myriad of forum interfaces makes other subdomains of the same solution harder to coordinate and mature vs centralized.
this is a great point, thank you for posting it.
i've been researching this problem for a year now, and came across some interesting technical and social povs on moderation and network attacks.
- moderation: my pov is that large companies with teams of moderators (such as twitter or facebook) wont and cant scale to varied moderation policies across the world. so the task of moderation has to be boiled down to the communities themselves. the question then is - a) how do you prevent a moderator from getting power hungry, b) how do you incentivize them to do the work and c) how do you make this auditable and provide recourse mechanisms? for all of this, open access to an API to build the right tools is super important as you said. and having this done on a protocol is the only real way to scale it imo.
- network attacks: we've done a lot of work in approximate solutions for sybil attacks. using verifiable delay functions (VDFs) as a way to order messages, and having them stored on multiple locations (eg arweave) in append only ways gives a lot more resilience / provides a way towards solutions here.
These type of attacks apply broadly at many different levels, for example the minimum thresholds for amplification/de-amplification can be significantly lower than the 51% that's normally attributed by definition to these type of attacks.
Any decentralized moderation system would run into issues with a sybil attack distorting reflected appraisal within a community. Some benefits of such an attack might be dropping the post out of view/delisting it, using generative ai to create harassing/malign responses attacking volunteer psychology (where they stop contributing/volunteering when it costs them).
Useless/mindless generative chatter that is akin to jamming a communications channel, but indistinguishable by most people which in the absence of a clear signal causes isolation (thanks AI).
Some of these challenges also don't have a good answer because they are fundamental problems where a solution would break automata theory/determinism. So you are left with only a bad, potential liability from an approximate solution.
Most propaganda and political warfare focuses on attacking various parts of communications with the intent to either isolate, or destroy/debase personal identity, which is largely determined by reflected appraisal.
Once isolated, with their belief system destabilized it makes people vulnerable and does dangerous/weird things to people's psychology. This has been known for quite some time though hasn't gotten much attention outside certain academic circles, and by quite some time I mean roughly generally known between the 1920s-50s.
The main driving component was formalized under the Sapir-Whorf Hypothesis (1929), though it was in common use by governments prior to that.
Since most social media already has these issues (but they don't communicate about them in public regularly if at all), internally they do recognize these issues and take a Signals and Systems approach, but the result is as you've seen.
If you cant exceed parity vs. competitive solutions you wont be able to increase marketshare/market penetration.
Then there was spam. Everyone had to do their own spam filtering. Or ... pay an upstream provider to do it for you. In which case they're basically the centralized solution.
The reason Internet forums took off was because it was way easier to let one central group manager the spam and load and network and disk and.... instead of everyone doing it themselves.
https://discuss.flarum.org/d/32812-the-future-of-flarum-in-2...
one of the biggest challenges to this model (IMO) has been that there is no "subreddit" or "stack exchange" model for discovery, or for communities to build their own "spaces".
my post comes at this from the perspective of - what would it take to create a decentralized "reddit", complete with subreddits etc. your comment makes me realize that I can do better in addressing this point! ty!
What's really the difference between the two? Is it a question of degree, or in this example is it only decentralized if individuals in the network own their own content?
Theoretically you could have a centralized federated model where users own servers in a single data center!
The most extreme form of both decentralization and federation is a p2p model where each user owns their own data on their own computer.
I agree with the other thread that it’s best to understand the meaning behind these terms and not to rely on them as shorthand or gotchas, but we do need separate terms for these two concepts.
I've never thought of decentralization as a purely geographic concern - may very well be my own bias towards caring deeply about data ownership and viewing geolocation of servers as an implementation detail that I care little about
I associate arguments about the meaning of the word "decentralization" with pseudo-arguments about how nobody wants to run a server that require "decentralization" to mean that everybody has to run a server.
Agreed that the context of how a person intends to use each term is more important. I just wish we could have a consistent definition so it didn't require a full explantion each time.
But word spreads around pretty easily, and even though most forums are centralized, there are a lot of forums. Many centralized platforms = a decentralized platform, because people connected to multiple will share messages across channels.
The biggest issue is like, in the Reddit debacle, many communities are likely to be hurt or even destroyed. And it will be hard to re-form them and get near the same amount of popularity
The best we can hope is we can build AI-moderated content systems where the AI is subject to some form of self-governement by the community by e.g. voting among possible moderation models. Adding humans into the mix always invite uncomfortable gatekeeping and political issues
It doesn’t need to be. Forums can just be unfriendly to people who don’t meet the standards for discussion. They can do this overtly, through moderation that shoves out low value contributors, or more subtly through design choices and cultural factors that discourage low tier posting.
Technology truly moved far in the meanwhile. At the moment the only protocol that goes beyond expectations is Nostr.
Over there it doesn't matter where your data is hosted and there are zero blochains. The only thing that matters is your private key (identity) and with that you can always write wherever someone is willing to accept your writings.
Quite a novelty paradigm, worthy of a true sucessor to Usenet.
Past all email limitations, a simple mbox archives everything. It works offline, is decentralized, etc.
The first step to cut through all that is the ability to get everyone onto a common auth platform. But no one wants to do that because that means sharing user data, having to share the users with other possible platforms.
It is possible to actually solve the security issues technically, but that is likely a generation away. Until then, the fact that the servers AND the users computer are insecure lead almost inevitably to walled gardens, and avoiding innovation.
Thus there will be centralized locii of control for the foreseeable future. We can't democratize the way we use the internet back into a true commons until usable secure general purpose computing becomes widely available and adopted.
We had it (secure general purpose computing) in the early 1980s with floppy based computers and dialup. I thought we'd get it back by 2030, but I'm not as hopeful any more.
- identity - a common id system across the whole system that supports all existing means (think oauth based)
- a mechanism to sign your messages to prove that you truly were the poster (since there is no longer a central authority to validate that)
- a mechanism to store these messages and get them in real time (this is where activitypub can be super handy)
- a way to search / index these messages for retrieval
- a way to create a feed of activity that uses some algorithm for ranking
- a way to discover other forums / communities
so while activity pub serves well for one aspect, the other ones aren't quite there yet to be used in a decentralized fashion.
Many centralized forums these days have their own "narrative". It could be of a political or economical nature. For example, each moderator on Reddit is biased towards what should and should not be published on "their" sub-reddit.
Here on HN unconventional wisdom is not welcomed either. Posts get easily downvoted to the bottom and then visually suppressed, hinting other readers: "hey, look, this opinion is not important, feel free to either ignore it or downvote". This visual suppression is, imho, a subconscious manipulation. Why nudge readers toward a certain direction? Let everyone decide on their own.
The current standard is "agree with me or you’re a white supremacist tinfoil hat conspiracy theorist and a nazi". Alternative opinion is akin to a crime.
Twitter is moving in the right direction (unconventional wisdom). There are very few bots left (barely see any). Irrespective of your political spectrum, nobody's going to ban you. Everyone is welcome on the platform. Twitter is about to start sharing ad revenue with content creators, which is a fantastic move in my opinion.
Lack of moderation brings everything into a balance – a self-organized chaos which is the closest thing to what our everyday lives are.
Mastodon seems to be a good canary, to see how decentralization works. It’s already well under way.
But I’m just a scarred, traumatized old soldier, and my opinion probably doesn’t count for much.
Maybe we really do need another alt.tasteless. It would seem quaint, compared to today’s foolishness.
A decentralized network won't magically remove the need for moderation or make the job less onerous or more attractive to do. Only now the server/instance owner is not only paying for the hosting out of their own pocket, now they are (also legally in many cases!) responsible for the content moderation too!
What a win! That totally makes it super attractive to do.
edit: I think that the Community Notes function of nu twitter is a serious step in the right direction, and there should be more academic discussion about how processes like that could be implemented best, or even updated on the fly to match the changing desires of users.
Reddit and similar solved it through economics of scale.
I want people to create GitHub repositories for their blogs and just post to README.md in reverse chronological order.
Then email me your blog repository URL or reply here and I'll create a curated list of blogs.
my framework is,
- moderation needs two levels (top level and grassroots)
- top level - has to come from communities that are not in an "accepted" list which shouldn't be allowed to form. (the usual suspects like hate speech, CSAM). one way to think about is having community server owners "stake" some sort of investment and risk losing it in case communities on their servers get banned. this would happen for instance if you run a server for this protocol, and someone on your server creates a community against policies set by the protocol steering committee.
- grassroots level - moderators are those who actively contribute to the community, can be elected, have audits on their logs, and can also be incentivized for their work (eg, given a "salary" every month) through that staked fund we talked about earlier.
ultimately, no one is doing any of this free and we should assume that monetary incentives in some way that are aligned with the community's have to be factored in.
Even this becomes tricky because you first have to make a common list of banned content and define the lines between what each category of content is. Sure a huge majority of people would like to see CSAM on the list, but they definitely won't agree on what is/isn't CSAM in their opinion
I've recently been working through whether any level of content centralization is possible without the fundamental challenge of moderation. If anyone can follow me by consuming content I post to my own site then the main questions are around discoverability. If I'm posting content to a server and service that someone else is directly responsible for, even if federated, they have to consider moderation and the array of different content laws globally.
But today, I took another tack: I started wondering what it would actually _be_ like if you had such a technology. Skip the 'how', and just imagine there was a way to open an edit-text field and send your message to "the world", with no intervening centralized boogeyman waiting in the middle.
The thing that suddenly struck me is the question of how anybody would _find_ your message. Ok, you yeeted your thought out into cyberspace - who cares? Who's going to read it?
I find that in my own habits, the question of who's going to read my message has a LOT to do with how interested I am in sending it. I suddenly realized that this is the critical point. Something that Reddit has going for it is that using it has the subjective quality of going to a _place_ in the world - a fixed place that others can find. It's not just that your content is discoverable per se, but rather that it _will be discovered_, because it is posted in a known, frequented place that will have other eyeballs. Like Martin Luther's 95 Theses, _where_ you post most messages has a lot to do with how interesting they are. It's accessibility, yes, but also the practical question of whether people will actually exercise their ability to access it. If Martin Luther had nailed his theses to a random tree in the forest with complex directions on how to go find it, it just wouldn't have been the same. They would have been almost as easy to access (except perhaps for people who didn't find it easy to navigate the forest), but the fact the church door was an inherently centralized high-traffic zone WAS what made it interesting.
Perhaps this sounds a bit trite, but I've started to realize that we crave centralized places to converse because conversing isn't nearly as much fun without an audience. The Reddits and Twitters of the world are popular and interesting _because_ they are a centralization of our communications - a front door of the church we can pin things to. If you find yourself thinking this is very incomplete, give it a beat. For instance, like me, you might have some level of disdain for performative soap-box'ing on a platform like Twitter. You might object that "sure, a lot of Reddit was crappy, affected karma-farming, but I just used it to discuss analog audio equipment and ask tech support questions about Rust!". But even niche interests, tech support, NSFW subreddits - the are all fundamentally relying on the same principles of knowing that you will have an audience. You might want different audiences at different times, of course, but most of the fun is related to how close you can get to the upper bound of who you wish could look at your content. Maybe you don't really care that the whole WORLD can see your post, just that the entire world of ukulele enthusiasts can see it. But within that slice, the more the better. Also the more that you put those disparate audiences next to each together with a central nexus / map, the bigger overall audience you can expect to have due to cross-pollination.
I bet that if you're really honest with yourself about what your favorite places to discuss weird topics in the 90's were, you'll realize that even though it was some phpbb forum, there was some microcosm of centralization going on, if even just at the level of human affairs. If it was some esoteric hobby, then you were probably talking on the official forums of some organization that ran the biggest public events related to that hobby, or whatever. Whatever it was, there was _some_ germ of a winner of a popularity contest - some entity that had the visibility and recognition to centralize the discussion and make it a place where you might get some interesting eyes on your text.
I could be misunderstanding some of the fediverse, but it seems like there's an explicit intention to create these autonomous sub-communities and remove any kind of centralizing authority. But really, that top layer generally IS the value, and is how you create a sense of Place. You can make the infrastructure as decentralized as you want, of course, but if the discussions themselves don't feel like they're happening on the church doors, you really don't have anything a phpbb forum doesn't have. You've basically just got people sending messages out into the void, which is about as fun and interesting as talking to a customer support chatbot. So, at _some_ resolution, none of this is interesting unless you're giving people a way to post their content in THE place to discuss this topic. "A" place to discuss this topic is comparatively worthless.
Zooming way way out, I want to make a higher-level comment using an analogy. All of online experience is a lot like a party. There are many things that it's good to have for a party: food, music, drinks, entertainment, shuttle service - lots of things. Yet we can all think back to parties that forewent some or all of these things and still gave us fond memories. There are a few things, however, that your party _has_ to have, like, definitionally: a place to be, and great people that show up. A party without these things isn't a party. And if you organize all of the nice-to-haves, but skip the essentials, then you've just got something very sad... like a loud, well-catered, empty basement. Related - sometimes bigger parties are better, but at a certain point, things just end up crowded, and probably a lot of your favorite people have left anyway. Once somebody starts charging exorbitant rates at the door, things might be headed downhill.
The point of the analogy is that where we humans choose to gather online is a fluid and vexing thing. You're going to find it impossible to write hard-and-fast laws about how your new discussion space should work because all that really matters is people are finding a way to have fun. This party-like quality, more than anything else, accounts for the ephemeral nature of all online platforms, and the natural herd migrations from myspaces to facebooks to instagrams to snapchats to tiktoks, etc. And sure, AFTER the fact, you can look back and give a technical analysis of why Facebook succeeded. You can also look back on your party and point to specific things that created a fun situation. But this never transmutes into some kind of formula for future success, and if you think you have one, you're probably chasing last night's party. Are we so sure that holding the party at Brad's house is what made the party bad? Yeah, he started charging $5 a head, and a bunch of jerks showed up at the end. But I'm not so sure that means we should swear off house parties from now on. Like, I salute the open-mindedness of the guys trying to figure out a way to hold a party without a host... you know, as a thought experiment. But it would be cool if there was a place to hang out next Friday, too...