49

Is It Time To Decentralize?

Disclaimer: Some may take this as a middle finger to SE and/or staff. I genuinely don't intend that. I think SE should want to decentralize, and could benefit from it. This is intended entirely to examine what I believe would be the best option for everyone, staff and community alike. I hope that comes through below, and apologize if I fail to be clear in that way.

There has been a lot happening recently - and not great stuff. From likely illegal content relicensing to the horrible treatment of Monica to questions about the TL being toxic to concerns about moderator review processes to GDPR issues - and even declines in sites as well as answer quality on sites. In short, serious concerns about network health - most of all the behavior of staff has led to a general sense of distrust, sparking what's been called a "crisis" by many users. Many users are now questioning why they are still here, raising money to help Monica because of harm caused by SE, and talking about labor laws. In fact, some people think SE should split into a for-profit and non-profit entity.

Is it time to make this a decentralized platform? Could a decentralized platform work? If so, what would it look like?

It seems to me like a lot of the frustration lately has come from relationships between staff and the community; a platform that didn't rely on the central control and development might allow for more options of future growth.

Improve this question
12
  • 20
    A lot of the issues here are not technical, they're social. There's a project or two that could for the basis for a perfectly good, though not distributed SE alternative, both in the pipeline, or existing, though needing updates.
    – Journeyman Geek
    Commented Nov 1, 2019 at 15:40
  • 4
    Putting the development/control of individual sites in the hands of the loudest minority is not ideal.
    – Kevin B
    Commented Nov 1, 2019 at 15:41
  • 2
    Not going to happen. SE needs "the free stuff" to add value to its commercial products. Who are you going to show job adds, if no-one visits your platform? Why would you want to use a teams product, if for all other answers you have to go elsewhere?
    – Luuklag
    Commented Nov 1, 2019 at 15:42
  • 4
    @JourneymanGeek "A lot of the issues here are not technical, they're social." And, more to the point, decentralized social systems simply do not work at any non-trivial scale. They've been tried. Over and over and over again, they've been tried. And they always fail; it's something that looks good on paper but just doesn't jibe well with human nature. So why try it again? Remember, the most dangerous words in the English language are "this time is different, I swear!"
    – Mason Wheeler
    Commented Nov 1, 2019 at 17:57
  • 4
    Can you include in your question some background info like: What problems do you think the sites are having that can be solved by decentralization/detachment? Advantages and disadvantages, notable examples like perhaps MathOverflow, statistics of improvement/degradation in various areas, etc? Like what makes you believe this is a beneficial move?
    – user1306322
    Commented Nov 1, 2019 at 21:32
  • 2
    "a platform that didn't rely on the central control and development might allow for more options of future growth." Or it might not. Without central control and development maybe nothing really gets done.
    – NoDataDumpNoContribution
    Commented Nov 1, 2019 at 21:42
  • Useful link: Comparison of Q&A software. The software is already available and several successful Q&A sites exist on specific topics, running free software and freely licensed.
    – Nemo
    Commented Jan 24, 2020 at 15:14
  • For whatever it is worth, I say yes. People thinking that the right response to the current situation is to trust another person who sets up a server are naive. If you can't trust Jeff or Joel circa a decade ago, what makes you think the person you trust today is any better? It may be worth a watch of FreedomBox and a review of Diaspora before jumping ahead with making the same mistake again.
    – HostileFork says dont trust SE
    Commented Jan 26, 2020 at 11:20
  • @MasonWheeler "They've been tried. Over and over and over again, they've been tried. And they always fail." Oh. You mean, like, the Internet? Maybe it failed at some level, but DARPA might have just been ahead of the curve. I'm old enough to have exchanged mail on a server where the sysop could break into the conversation when you're writing someone and say "hey, sure you want to write this and leave the message for that person when you hang up and they call in?" I'm lazy enough to want to preach gloom and doom for decentralization, but smart enough to know laziness is stopping it.
    – HostileFork says dont trust SE
    Commented Jan 26, 2020 at 12:27
  • @HostileFork Yes, I mean like the Internet. It's a prime example: they tried to build a decentralized system and failed to achieve it. I remember those days too, and they're long gone. Today, virtually all messaging goes through Google, Yahoo, or Hotmail on the email side and Facebook or Twitter on the social side. It's a shorter list of sites that aren't hosted on Amazon, Microsoft, or Google cloud servers than those that are. And so on. It's one thing to try to set up a decentralized system; it's another matter entirely to keep it that way. Today's Internet is massively centralized.
    – Mason Wheeler
    Commented Jan 27, 2020 at 15:01
  • 1
    @MasonWheeler And it's our fault for favoring convenience over what is correct.
    – HostileFork says dont trust SE
    Commented Jan 27, 2020 at 17:02
  • @HostileFork What's "correct" here? By what standard is something that's failed every single time it's been introduced, due to being fundamentally incompatible with human nature, more "correct" than that which actually works?
    – Mason Wheeler
    Commented Jan 27, 2020 at 17:07

6 Answers 6

Reset to default
57

I get the impression that most of the sites on the Stack Exchange network are seen as a net drain rather than a net good by the company. We're too small to drive significant traffic, often too small to be attractive even for ads, and yet we chew up a lot of employee time and company resources. It might serve everybody better to allow sites to depart gracefully while preserving the core functionality.

In other words, it seems to be in SE's best interests to support communities that want to depart. What would that support look like? The biggest barrier to any "rebuild elsewhere" effort is the "rebuild" part, writing or adapting the replacement software. It would be better if a friendly agreement between a community and SE could lead to the community running with SE's software, but on its own servers (which that community would have to provide).

SE has a way to do that. They sell that solution to large companies with deep pockets. The small communities that SE doesn't draw value from don't have deep pockets and can't pay for SO Enterprise. But the fact that the product exists suggests that something would be possible here.

What would it take for an interested community to gain access to the SE software to run outside of the SE network? Is it in everyone's best interest to allow that path to exist, just as SE 1.0 once did? Would SE grant a perpetual (or at least long-term) license, free or inexpensive, and wish a community the best?

Of course not being part of the network means not being tied into network-wide features like global login, featured meta posts, and HNQ. These are among the things a community would have to discuss. Login, at least, should be a solvable problem somehow, because SO Enterprise and SO Teams users need to be able to log in too, but I don't know if those are tied to corporate LDAP or some such (which would be a problem). We've now reached the limits of my knowledge in this area, but what's needed first is the higher-level discussion, not a deep dive into details. Would the expedient, mutually-beneficial path to separation be available to an interested community?

Improve this answer
18
  • 1
    SO enterprise uses a standard SSO system as an option. I can't remember which one on account of it being past midnight, but running your own is something I was looking at since I really don't want to be tied to google or facebook, and want to arbitrarily create and remotely remove logins/credentials. Used to be able to do it with openid ._. Will update if I remember what it was
    – Journeyman Geek
    Commented Nov 1, 2019 at 16:25
  • 2
    we chew up a lot of employee time and company resources We shouldn't, IMO.
    – ChrisW
    Commented Nov 1, 2019 at 16:31
  • 2
    @JourneymanGeek thanks for the info. It seems prudent to support Google and Facebook (sigh), both because people generally expect them and to facilitate migration, but also having another option (like SE does with email-based sign-ins) would definitely be good.
    – Monica Cellio
    Commented Nov 1, 2019 at 16:43
  • Could you clarify which parts you are envisioning splitting off, and which not?
    – aparente001
    Commented Nov 1, 2019 at 17:47
  • 3
    @aparente001 there are a few sites (trilogy and a handful of others) that SE presumably cares about keeping on their network as part of their brand and/or a source of ad revenue. I'm suggesting that any site not on that list that wants to leave be helped to do so in a way that minimizes disruption. That's a better outcome than people setting up an alternate site and the community fragmenting, so it's better for the community, and if the small sites are the burden they seem to be, it's better for SE.
    – Monica Cellio
    Commented Nov 1, 2019 at 17:50
  • 2
    I had to look it up! I found out trilogy = StackOverflow, ServerFault and SuperUser. Sigh. I just wish they would talk to you. // Any idea of the budget needed for the sort of split you started to sketch out in this answer? // Related post, hope you'll take a look: meta.stackexchange.com/q/337300/287826
    – aparente001
    Commented Nov 1, 2019 at 18:00
  • 2
    @aparente001: There is a Stack Exchange glossary (but it has the same problem of discoverability). And my profile page (but again the same problem).
    – This_is_NOT_a_forum
    Commented Nov 1, 2019 at 18:58
  • @ChrisW even in the best of situations, users generate work for the site owners. That why SO has... uhm.. had.. several hundred moderators to decentralise that burden too.
    – gbjbaanb
    Commented Nov 1, 2019 at 21:40
  • 1
    @PeterMortensen - Sorry, what's VC?
    – aparente001
    Commented Nov 1, 2019 at 22:26
  • 8
    -1, because you are 100% wrong. 'The biggest barrier to any "rebuild elsewhere" effort' is NOT in any way, shape or form, the software, or the unified login. It's two things first and foremost: network effects; and SEO Google search position. Both of those have been generated with the volunteer work of people contributing to SE (mostly SO); and cannot be recreated at scale. I (or well, a talented small team of developers) can easily re-create SE, or Twitter, or Facebook software, especially for smaller scale. It's re-creating the network effect that is near impossible.
    – DVK
    Commented Nov 2, 2019 at 15:15
  • 2
    I think the most important thing that decentralization could offer is being independent of the CoC.
    – Jeff Darwood
    Commented Nov 2, 2019 at 17:00
  • 8
    @DVK: If it's a divorce on amicable terms, there's no reason why the departing sites should lose their SEO position. SE can put in an HTTP forward, and Google understands that.
    – MSalters - reinstate Monica
    Commented Nov 3, 2019 at 0:00
  • 6
    That's why I talked about it as migration, @DVK. If a group splits off while SE continues to host a site, the new site loses. If SE assists a community in relocating, however, there's no search-engine competition. Some communities don't really benefit from the network effect and might be happy to go elsewhere -- the more niche the topic, the less benefit there is from HNQ. The largest challenge is people (the community), but I don't think the software is as trivial as you do.
    – Monica Cellio
    Commented Nov 3, 2019 at 1:07
  • 3
    @JeffDarwood more broadly, self-governance. A community can set rules that make sense for that community.
    – Monica Cellio
    Commented Nov 3, 2019 at 1:10
  • 3
    @DVK let me try again: my proposal is that if the Unicorns site moves, there will no longer be a unicorns.stackexchange.com site to be a search target. I'm suggesting that SE assist sites it doesn't value anyway in moving, not cloning. Yes the new site will have to start over with SEO, but SE wouldn't be competing.
    – Monica Cellio
    Commented Nov 3, 2019 at 15:13
11

Is it time to make this a decentralized platform?

Un-answerable objectively. Especially since you have not defined what a "decentralized platform" actually means.

Could a decentralized platform work?

In most meanings of the word, the answer is unfortinately "no, it could not work" - but NOT for obvious reasons (stated in other answers), and nothing to do with it being a decentralized platform - ANY new platform would not work.

The simple reason can be boiled down to "It would not be THE SE/SO brand and network". More specifically, "The SE/SO" means two things that present a near-insurmountable moat/barrier:

1. Network effects/existing user base/brand

SO has gradually over the time attracted millions of users and hundreds of thousands active participating users.

The ONLY reason it works as well as it does is this volume.

This is why many small sites on SE platform struggle and why SE closed down Area 51 process - even with SO influenced volume bump AND (see #2 below) Google rank, new sites simply can't get off the ground.

2. Search Engine Optimization (SEO) - most importantly, high Google page rank.

Again, SE/SO built this up gradually over long time, based on efforts of said active participating users. Starting from scratch, your new site - decentralized or not - would have zero chance to compete with SO/SE given that as per SE's own words, most their traffic comes from Google.

If so, what would it look like?

The only way it would work/look like, would be (obviously, this would never happen and SE the company would never consider it) would be for them to relinquish the BRAND (specifically, URLs that Google page rank is based on) as well as all existing users.

In other words, SE/SO content would stay exactly as they are on the same (user-facing) web sites/URLs, but be run by a different entity (or decentralized set of entities); with the same userbase who may have been migrated to new back-end.

Whether that would be running on existing (provided for free or for pay) SE back end software, or on some replacement software, is not 100% important - it would have effects around the margins, but the "viable/non-viable" line is not that.

Improve this answer
0
10

Could a decentralized platform work?

No, it won't work. There have been numerous proposals over the past 20 years over how a "decentralised"/"community owned" social network could function, but in reality none of these proposals ever materialised into a successful project. Even websites like Wikipedia which run on donations still have a large body of employees and don't necessarily function under completely democratic principles.

In reality the only threat to SE's business is another company taking their data dump and then proceeding to build a clone of Stack Exchange while dumping tens of millions of dollars into advertising. But this new company wouldn't necessarily be any more democratic than SE's current management. And it could be even worse if an Internet giant such as Google takes over, as you won't ever have a chance to interact with management beyond a generic customer support line.

Improve this answer
10
  • 4
    can you link or expand on at least one of the failed proposed decentralized solutions? since this is highly dependent on exact implementation, it's good to know what exactly we're referring to as examples of unlikely and plausible paths to a good result
    – user1306322
    Commented Nov 1, 2019 at 22:13
  • 1
    @user1306322 Diaspora comes to my mind as one example. It's been around since Q4 2010, so hasn't exactly failed, though (per Wikipedia) only has less than 700,000 users in total. Compare this to Facebook, which has been publicly available since Q3 2006 and now has almost 40,000 employees and passed 100 million registered users in 2008, and had 2.3 billion active users in Dec 2018. We could discuss details, but clearly, Facebook's centralized development and hosting model has achieved something Diaspora's distributed one hasn't.
    – user
    Commented Nov 2, 2019 at 12:13
  • @acvn yep, Diaspora is the most famous example. Decentralised communities don't work, neither online nor in real life.
    – JonathanReez
    Commented Nov 2, 2019 at 16:32
  • 3
    Mastodon has 2.2 million users currently and as far as I can tell is still growing. It's more of a Twitter alternative but, I don't think decentralization alone is the problem.
    – Em C
    Commented Nov 2, 2019 at 19:49
  • @EMC Mastodon looks interesting. Do they have examples of heavily right wing communities on the platform, as a proof of their openness to any speech?
    – JonathanReez
    Commented Nov 2, 2019 at 20:38
  • I'm not extremely familiar but I know Gab migrated to it - theverge.com/2019/7/12/20691957/… talks about that, sounds like the creators didn't (and couldn't, bc decentralized) do much other than declining to add it in their directory, and mostly it was up to individual server admins whether to block the Gab instance.
    – Em C
    Commented Nov 2, 2019 at 20:44
  • 2
    To some degree that mirrors how different stacks have their own local site norms, though there's still a lot of overlap and the CoC of course.. but decentralization isn't and shouldn't IMO be a synonym for unmoderated - moderation is still necessary because of spam and trolls and human nature. I do think the concept of a social networking "protocol" instead of a complete package controlled by one entity is intriguing though.
    – Em C
    Commented Nov 2, 2019 at 21:05
  • @EmC doesn't sound like a free platform then, unfortunately. It's okay if individual users block a channel they dislike, but site admins making decisions for their entire communities isn't much better than StackExchange making decisions at a corporate level.
    – JonathanReez
    Commented Nov 2, 2019 at 22:05
  • @EmC reading the article it appears that the "Fediverse" has its own code of conduct and not following it means you won't be listed on their official pages. IMO it isn't any different from SE's CoC.
    – JonathanReez
    Commented Nov 2, 2019 at 22:13
  • 2
    As I understand it, individual servers are totally free to operate their own local communities however they wish - going against the Fediverse CoC just means they won't advertise it for you (à la that one xkcd). So the model seems more analogous to, say, SE taking a site off HNQ, or site mods suspending people who post certain types of content, but those people are still free to participate on other stacks.. Anyways, we can agree to disagree, I just wanted to share an example that people might not be aware of :)
    – Em C
    Commented Nov 2, 2019 at 22:23
4

I don't see why not. They just have to follow the kind of model WordPress uses, where the server and software is managed centrally and any old fool can set up their own site using the platform (either with "powered by SO for Teams" banners or paid for that ad-free experience).

There would still be the issue of poor governance though, if SO decided that they wanted to control what kind of site was allowed, or what kind of content was allowed (no Q&A site for Illinois neo-Nazis for example) or mandatory. They would have to give up a lot of control over the sites to the new site owners, and I doubt they'd do that even if they wanted to lose the hassle of managing them.

Improve this answer
1
  • On the topic, it's useful to look at what Matt, the CEO of Automattic (Wordpress.com), had to say after they bought Tumblr. theverge.com/2019/8/14/20804894/… "We have peaks and troughs of openness on the web. I think we are exiting a trough."
    – Nemo
    Commented Jan 24, 2020 at 15:13
2

Is it time to make this a decentralized platform? Could a decentralized platform work? If so, what would it look like?

In your question you miss a subset of SE, as the only site that is really tied with other is SO and SO for Teams, and for me it's that one that would need to be more independent, or more protected from the users that use it.

As that bring another question: Why is a paid platform tied to a free platform? Possibly after we could argue that, the Hot Meta Post from SO and the HNQ got tweaked to hide problematic site/contents from paid customers that see SO.

I bring that point as in the world of each site, each site follow the guidance of MSE, but MSE brings a layout, and we have a certain liberty inside that layout.

An example in the handling of comments: on the site I am a moderator on, comments are really important for the diagnostic step and as such I rarely delete useful comments and I almost never move to chat, as the diagnostic is important, as much as the answer sometime. On that matter I differ from the leading of MSE, that comments are worthless and can be erased anytime.

It's why for me each communities are already independent somehow, with the tool subset MSE give us to moderate, but the fact that we got paid customers mixed with SO can bring bad business decisions IMO in the long term.

Improve this answer
5
  • 6
    If you think MSE is aggressive with comments you should see the site I moderate ;-)
    – Rubiksmoose
    Commented Nov 1, 2019 at 16:42
  • 2
    @Rubiksmoose I'm scared already ＼(º □ º l|l)/
    – Suvitruf - Andrei Apanasik
    Commented Nov 1, 2019 at 17:05
  • 2
    @SuvitrufsaysReinstateMonica As a member of that site where Rubik moderates, I love the better signal to noise ration of that site. I wish more SE sites embraced that comment discipline. The comment piles on MSE are mostly noise, though there is some pretty good signal buried in there.
    – KorvinStarmast
    Commented Nov 1, 2019 at 21:31
  • What do you mean by "...the only site that is really tied with other is SO and SO for Team"? Is it "...the only sites that are really tied with each other are SO and SO for Team"?
    – This_is_NOT_a_forum
    Commented Nov 1, 2019 at 22:21
  • @PeterMortensen Yes, I mean that. as if you are a member of a Teams, you don’t see that site in your profile site list. You have to go to SO, and then you see your team on the left with question showed for the teams and SO question in the display per default.
    – yagmoth555
    Commented Nov 1, 2019 at 23:53
1
+100

Oh, I was thinking about decentralized platform many times and here what I think.

  1. Decentralized platform is extremely difficult to create speaking about technology. Decentralized database, blockchain, online consensus, problems with domain name, open protocol, hackers who will have a lot of fun here... I don't say it's impossible (see Diaspora for example), but... well... it costs too much.

  2. We will need some "really democratic" management with elections, rules, constitution et cetera. And here main problem lies: "democratic management", if it's badly configured, can lead to disasters much worser than current one. If Trump and brexit does not convince you -- think about systemd.

  3. When it comes to money, I really don't understand anything here. How to pay developers who will develop this? With "StackOverCoins"? Who will make decisions, hire&fire devs, control process? Community? Who will decice about new year bonuses etc? Again, I'm not saying "this is impossible", but I would say "this is definetely difficult and can and will lead to conflicts

Even in Wikipedia there is "top-management" which is not democracy-based who are resolving conflicts and fixing issues when situation goes completely mad.

Why we want to do this? Why we want to invest in heavy technoligies (1) and put whole community at huge future risk (2)? Just because one specific company got overwhelmed by assholes?

If so -- we should just fork and choose/create another company.

Improve this answer
4
  • ngl, the systemd reference is completely lost on me :p
    – user1306322
    Commented Jan 27, 2020 at 15:15
  • Regarding payment, normally (for altruistic global initiative like this) something as major as creating a free alternative to a mismanaged locked down but popular software is usually done by selfless developers in their free time. Sometimes they accept donations via Patreon or Paypal. And the scale is different. I don't think anyone actually wants the entire everyday traffic of SO, so that brings down the costs and concerns for a lot of things. And the point of decentralization is that there is no one "top management" that can go rogue and bring everyone else down. Each instance has its own.
    – user1306322
    Commented Jan 27, 2020 at 15:19
  • In a broad sense, we already have decentralized nodes of SO -- all the clone sites which copypaste the content, some with proper attribution, some without. You could say they are owned by some unknown management, and their userbase consists only of those who mistakenly thought this is a legit site, and don't know it's a clone. If SO goes down, its current contents will be preserved by those sites for a long time probably. That's basically only one feature (independent backup) out of so many that people really want, not counting the backlog of requested features and new spins on the Q&A format
    – user1306322
    Commented Jan 27, 2020 at 15:24
  • Selfless developers already tried to make SO clone many times and all attempts failed. This project is larger than one which can be developed for free. Unfortunately.
    – Arenim
    Commented Jan 27, 2020 at 17:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .