The promise and paradox of decentralization
thediff.co
thediff.co
> The first downside to "Anyone can build anything" is that "anyone" means anyone, and the people to whom decentralized systems are the most attractive are the ones who are banned from other systems, often for good reasons.
This is an often overlooked point from the user perspective. Decentralization sounds great when a website censors content you didn't want it to censor, but most users don't realize just how much unwanted spam and abuse content gets quietly removed from public platforms on a daily basis. The sheer volume of spam or even just angry/abusive users on public platforms these days is hard to describe if you haven't been on the spam/abuse prevention side of a popular website.
Worse, the most abusive users tend to be the most persistent. They can become very good at gaming things like IP blocks, reputation systems, or voting systems. The best users can quickly get frustrated with even small levels of abuse and leave a platform.
Decentralization seems to work well for cases where individuals are privately interacting with other users they already know. It seems much more difficult to solve the problem of public forum websites where anyone can contribute content. That's a dream come true for spammers and abusers. I'm interested to see how this space evolves as these platforms try different solutions.
The bigger the size of a network, the more attractive it is to abusers, while at the same time being more difficult to manage.
Typically, they are constrained by some limitation like being for an offline group or requiring a vouching process to join.
Importantly, a blockchain is not very fit for an application like this because it is append-only. That makes perfect sense for a transaction ledger, where you want immutable history and transactions are reversed by entering a new transaction with the signs reversed, but you really want the ability to correct a reputation history by actually redacting false reports rather than just appending a correction.
George Orwell's 1984 features a main character whose job is just this: rewriting history by 'redacting false reports'.
Of course well-meaning people will set up these systems without considering the potential for abuse. In many cases, persistence is desired - it does matter who said what when, historical written letters are important documents, and records of actions also form institutional histories. Normalizing the 'redacting of false reports' can easily turn into rewriting history for propaganda reasons.
I suppose you could have a two-lane social media system, one channel where comments and posts were not anonymous and were recorderd for posterity, and one channel for anonymous ephemeral chatter. However, trying to run both channels on one platform might not work, legally or technologically.
punishing someone for breaking the rules - not vague "we reserve the right to fuck you in the ass for any reason or no reason at all" kind of bullshit rules employed by facebook/twitter/reddit/whatever - is not considered censorship even at the kinds of places where people are particularly sensitive to it - for example, people don't accuse moderators of censorship when someone posts porn on 4chan's safe-for-work boards and it then gets deleted.
besides, facebook/twitter/reddit/whatever, despite being centralized and authoritarian services, aren't equipped to combat abuse any better than decentralized entities, the only difference being that instead of blatant low-effort buy-penis-pills spam they are targeted for subtle manipulation at industrial scale.
if Facebook actually had a PR position of "we remove whatever the hell we want" it would be fine, rather they prefer to have nice sounding rules and then interpret them however the hell the want.
Personally I have a similar position towards net neutrality... if my isp wants to degrade connections to torrents and netflix/youtube that is fine, but the have to put it in writing in the contract.
If moderation is just the current group in power gets to decide what is moderation and what is censorship then I fail to see the difference. You just redefined the centralization because I guarantee that group will be smaller than the rest of the internet.
Yes, but this isn't some kind of repudiation of the concept. One person's justice is another person's injustice. One person's virtue is another person's iniquity. Moderation is a necessary component of any civil social structure.
Any competitor will immediately be inundated by spam, scammers, child porn, trolls, and Nazis to name a few. Since bad drives away good this will make their platform less desirable. The existence of a highly toxic refugee population deters anyone from creating competing networks.
I don't think this was planned or intended, but it works out this way.
Certs are pretty dang centralized.
Also, Your second sentence needs a bit more fleshing out because it doesn't follow.
Chain of trust is a decentralized reputation protocol and we use it frequently and effectively.
A contributing factor is that the culture that grew around them makes it uncommon to be able to manipulate your own certificates. If you set up a Web page with your own root cert, you can be sure it will never grow.
So "decentralized" in the sense that there are more than one, but it's not some hierarchy-less free for all where anyone can attest to their own identity and trustworthiness and the network just automatically accepts that.
I don't really understand the complaint because you literally can attest to your own trustworthiness through your own cert.
Having others trust you is a taller order. Obviously that isn't automatic.
I haven't been on one of those teams, so I hope you'll forgive the naivety, but:
These seem like solvable problems with decentralized systems. In both cases, someone has to go through the work of manually identifying the bad content, right? In a centralized system, that's someone working for the system - in a decentralized system, that's a random user.
From a technical perspective, then, the centralized "please delete this content" message is pushed out the entire system, while the decentralized message/action can be put into a blocklist/banlist that other users can subscribe to. I believe that this is how the Fediverse works, for instance, and it's definitely how adblocker blacklists work - so this kind of system is already in effect, and seems to be working decently.
If the volume of spam is truly extreme, then what's to prevent you from having distributed blocklists that are fed by automated processes (as opposed to manual additions), and users just subscribe to the ones that they trust?
From a social perspective, users seem to be willing to do this work themselves, given how driven the users of sites like Reddit are, with no more reward for posting high-effort content than a bunch of imaginary internet points, and how effective adblockers are.
To summarize - what prevents a decentralized system from taking the same approaches that a centralized system would employ, packaging them into blocklists, and then allowing users to choose which of those they employ? Same tech, different level of control.
I think the problem is that by and large, users are not willing to do this work themselves. When faced with a social platform that has a lot of jackasses on it, rather than individually curate their experience to remove the jackasses, most of them just leave the platform and find another one where this work is done for them already.
And this is why social networks have abuse teams. If it were totally up to them, they'd rather save themselves the expense, but users have shown that they will leave a platform that doesn't moderate, and so all social platforms are eventually forced to.
Understand that most of the general population operates on the level of "Somebody said something on the Internet that offends me; how could this have happened?" And that the maximum amount of effort they're willing to put in to rectify this situation is clicking a button. The realistic amount is that they wish it never happened in the first place. That's the level of user-friendliness needed to run a mass-market consumer service.
That being said I don't think it's impossible to have such a system in a decentralized way, there are incentive structures you could build to handle this.
Moderation is a fundamental requirement of any social system, and it's one of the two things that Web3 can't yet address meaningfully -- the other being addressability.
It's solvable in the same sense that email spam is solvable. Much like adblockers, spam is "solved" by re-centralizing.
> If the volume of spam is truly extreme, then what's to prevent you from having distributed blocklists that are fed by automated processes (as opposed to manual additions), and users just subscribe to the ones that they trust?
Most users are unequipped to evaluate that and disinterested in putting in a bunch of work to defend themselves against the flaws of the system at hand. Like adblockers or email, most members of the general population want it to be easy, automatic, and require minimal effort beyond clicking the button that gets them going.
Users generally want things to work for them. Investing deeply in protecting themselves because the system's designers didn't consider abuse is rarely towards the top of the priority list. People like things that just work, and the further a thing is from that the more adoption will struggle.
Because the incentives are misaligned; bad actors have, if anything, more incentive than the good actors.
Maybe for you, but not for me.
If you have data/industry experience in support of this, I am genuinely interested to hear about it, otherwise it seems like a case of my anecdote against your anecdote. It is not clear to me as a layperson that spam misclassification is indeed that big of a problem.
The problem is, even at 99.9% efficacy, with around 50% of all e-mail being spam (that number used to be much higher, and there's some question as to whether the drop to 50% is a trend or a reflection of failures to detect/identify), that's still a lot of spam getting through.
[EDIT] Seems like these type of systems exist in some projects reading from futher comments.
Bringing together the entire world in one public space is Twitter and Facebook -- i.e. possibly not a good thing.
It's recentralization.
Users of a decentralized protocol may use one or more decentralized networks but there's no need for a given specific decentralized network to be accessible by everyone.
This does require effort on the part of the user, but that might be a good thing. Handing things to users on a silver platter without effort on their part leads to ad-economy driven manipulation, etc.
But these don't actually need to exist. Decentralization doesn't have to mean a flat hierarchy, it could (against all trends) be a hierarchy with the user at the top, as social media once looked like. I could select the content I want to see by selecting the people that I trust and the content that they approve of, provisionally trusting the people that the people I trust also trust, and maybe extending that to an arbitrary number of hops. In that way I can choose my own mods.
This requires clients that work for the user rather than for the content creator, complete anathema to major browser vendors, who divide their time between anti-features to force their users to do things that they wouldn't want to do if given a choice, and candy for website owners/developers.
Labeling spam should punish the connections that caused it to be surfaced. That might damage serendipity slightly, but the modern web has run a bulldozer over serendipity with its engagement metrics algorithmically curating newsfeeds.
edit: one of the reasons I want (and probably many want) this kind of control is because I don't want spam, but I do want abuse. I don't want angry people filtered out as a rule, I want to be able to make the decision to blacklist specific angry people, or even people making stupid arguments in good faith. I want control over my own filter bubble. If that contracts my www to a gathering of friends and family, so be it.
If web 1.0 is simple content consumption, and web 2.0 is users providing content to each other filtered by powerful rent seekers, web 3.0 should be user governance. The "mod" system is an authoritarian lack of a governance process, and should be replaced with tools that aid collective decisionmaking.
I've been writing a Reddit clone that uses elected moderators, statistical sampling, and referendums to estimate the result of referendums. I'm curious what tools you think would work well.
https://efficientdemocracy.com
New name needed
Democracy only works when you have more energy available than people and that is about to not be the case after 200 years of extracting sunshine from the earths crust. The only thing we're efficient at is to waste energy!
Also "sign in with google"?
Your own http server/database!?! Do you mean you implemented them from scratch? I'm just using nginx/expressjs & postgres. Edit: rupy! interesting! Does ssl kill it's efficiency?
> Democracy only works ...
Energy limitations seem beyond my lifetime. I just want to see how to run a community based on the will of it's members. Efficiently using their time.
> sign in with google
It's not ideal. I originally implemented my own authentication, however implementing it "securely" is a huge task. My main problem is unknown unknowns. Even big companies have non-perfect authentication, like leaking emails during recovery/login. If the site gets any traction I'll expand the authorisation options.
What do you think I could do to improve the site?
I made my own secure user handling using this: https://datatracker.ietf.org/doc/html/rfc2289
We seem to be moving toward a society where grifters simply feast on the uneducated and un-discerning. The thing that makes it hard to police this is that the marks will themselves fight for the grifter, as we see with all the people fighting tooth and nail against their own access to health care or against treatments to stop them from getting horribly ill or dying in a pandemic.
Using Twitter as an example - they are going to be purposefully and continuously ejecting users. As long as the users are people who nobody wanted to listen to anyway then that is fine.
But the situation changes because sooner or later (potentially already happened) Twitter will eject actual community leaders and a community will need to reform somewhere else.
When that happens, open protocols and open platforms will help the community reform. It is quite hard to stop a community reforming on the internet for example because there are so many parallel communication channels that are basically free. So using Twitter isn't much of a risk because if they go bad then the community can reasonably move somewhere else.
People seem to expect "decentralised" leads to some sort of semi-utopia where everyone has equal status or act like some sort of saintly community driven by consensus. I don't see how it can work that way. It really just makes it easier to transfer power when the currently powerful start acting abusively.
You're saying this as if it's inevitable. It's not. The incentives are aligned to keep all reputable "community leaders" on the platform. If you're concerned about how a market driven organization judges reputability, well, (a) that's a different conversation, and (b) check to see who you're aligned with.
"Actual community leaders" is a matter of perspective though. "Actual community leaders" don't need a megaphone like Twitter, because they're already leaders that have an audience. Social media is a way to tap in to a broader distribution platform, and it seems appropriate that said platform would restrict that access to those that serve the interests of that broader audience.
For better or worse churches frequently served as community hubs in the 20th century. The pastor would essentially serve as a moderator who was supported by the church members pooling their money.
Of course pastors do more than this, but it’s up to the pastor to make the church a “safe space” by whatever standard the community expects.
I skipped copying it all because it’s right there; not limiting my reply to that snippet.
Yeah information network signal attenuates over longer distances (space and time).
The internet created an ansible for assholes to hassle us.
Knowing this, fuck the open web. I want an email provider that just straight up blocks messages from senders I haven’t approved.
I’d like Signal to go the extra step and make communication over a 1:1 link, maybe using Wireguard, an option, as a fallback and to put other options in front of less savvy users.
Spam is just advertising by agents that won’t kowtow to government. I don’t need schemers in big biz or Africa.
Decentralizing of communication will never happen running some platform centralized around rugged individualist hackers motives. But we don’t exactly need FB, Twitter, etc, either.
Our phones are just TVs for business opportunities. I’ll just send my adventures straight to people who care and skip the noise thanks.
abuse is a danger if identity is volatile.
If it isn't, things look very different. So put a price on identity, imagine the domain-name for that. And voliá you have reputation building.
English-language media is currently doing this to a lot of smaller cultures and languages, replacing them with a uniform monoculture. And that’s because the same thing happens when communication tech is centralized in a small number of megacorps.
The original appeal of the Internet to me was the potential for a vast amount of interesting subcultures. So while decentralization does enable a lot of messy and undesirable stuff, I think it might be a requirement if you don’t want everything, everywhere to be exactly the same.
Except it's not obvious. There's very few examples of linguistic dialects and cultures being forcibly assimilated by larger imperial conquest before modern times (the French Revolution is a good dividing point). Even France, one of the best examples of a large, centralized, unitary state throughout most of European history, didn't really make major strides in stamping out the langues d'oïl and langues d'oc that weren't French until the French Revolution. Previous notable large empires--e.g., various Persian empires, the Roman Empire, Mongols, Islamic caliphates, the Ottoman Empire--were often noted for the high diversity of cultures; that monoculture is a necessary component of a state is largely driven by 19th century ideas of nation-states, which (again) don't really exist before the French Revolution.
If you want to think in terms of decentralization you have to stop thinking in terms of broadcast, influence, publication, and broadcast. In a decentralized system you only influence those whom you are directly connected to and only if they wish to consume it.
If the goal of your online presence is some form of attention seeking behavior then be happy with Twitter and Facebook. If on the other hand you wish to share and expose absolutely everything without embarrassment or violations of privacy decentralization is probably something amazing.
True decentralization is a natural but ephemeral quality of networks with relatively few nodes. That is, scales where each individual node has enough internal resources to accurately represent the entire network. Human networks are parameterized _very roughly_ by Dunbar's Number (100-250), while modern computer networks are much more capable.
As a small network scales up, decentralization becomes unnatural; nodes and links require too many resources to accurately model and communicate the network state. It becomes net energy efficient to introduce layers of dedicated networking nodes, routing protocols, and entire subnets. It is not a fluke that "Web3.0/DeFi" technologies so far are either a) tremendously inefficient or b) end up centralized.
If you want to preserve true decentralization at this scale, you need to enforce it. Prohibit centralization in any form. No more central networking/control nodes. Perhaps there are clever schemes where you can e.g. shard the entire network uniformly across its nodes, but the bottom line--the paradox--is that some central authority must decide on and enforce the sharding/decentralization policies and actively prevent centralization from emerging.
Frankly, decentralization at almost any meaningful scale in our modern global society is a myth. It's a myth tightly coupled to the American origin story and its philosophical roots in the classical liberalism of Locke et al. and Western/Reformation Christianity.
From a hard scientific perspective, it makes no sense why to sacrifice the welfare of a system for the sake of maximizing the welfare of an arbitrary component that depends on the welfare of the system. Especially when we are confronted with systemic I/O imbalances on a global scale. To arrive at the myth of decentralization, you need the ingredient of (hyper)individualist ideology.
And for an ideology that actually build decentralised but just systems, I am strangely enamored by anarchists.
They(except Ancaps) have been working at this for about 1-2 centuries and it's a hard problem (and not all anarchists are trying to be smart about it). The only way that I can see to make it work (in theory) is to find a way to do confederations of confederations that keep everything more or less manageable with human interpersonal relationships and "sane" local rules on each layer and delegate global things upwards, with shortcuts and balancing mechanisms that make sure the state apparatus stays decentralised, nimble and controlled bottom up, not too down. I think Switzerland and Norway as well as some first Nations are closest to this, but I hope we'll all get there one day.
I view anarchism as another expression of classical liberalism, with the same structural faults re: decentralization, etc.
It's the difference between positive and negative freedoms (I think in germany and especially the nordics poor people are much more able to chase their happiness than in the US,in the sense a decent life without homelessnes, fear of bankruptcy due to disease and horrible working conditions is possible, not if you want to become a millionaire).
And anarchism was literally born out od rejection of classical liberalism and "classical liberals" don't really like them https://mises.org/library/are-libertarians-anarchists. (not a good source on anarchism, but just to give evidence that classical liberalism/libertarians and anarchists are not the same. There are socialist libertarians who are closer to anarchists but I think most libertarians would not align with them)
> And anarchism was literally born out od rejection of classical liberalism and "classical liberals" don't really like them
This doesn't quite align with what I've learned. Nonetheless, sectarian infighting between liberal ideologies is indeed a common theme.
I think you're right in that peer-to-peer doesn't scale. I also don't think many (reasonable) individualist ideologies espouse that it does either. Individualism isn't renouncing hierarchical power structures, but rather asserting that the power rests with the hierarchy's leaves (individuals) not with the root nodes (collectives). It's the same way the west leverages a republic rather than direct democracy for it's legal system.
Or put another way, the individual leaves should be in control of the nodes higher up in the hierarchy, not visa versa. Imo, your interpretation of the debate is missing the real argument being had here. Decentralization in this a political context is saying "power should be distributed to leaf nodes as much as possible". Naturally, those leaves will still organize themselves hierarchically in the name of efficiency (ie: I'll elect this official to make decisions on my behalf because i don't have the time to contemplate every bill/law being proposed myself). But that hierarchical delegation is still a distributed/decentralized power structure as long as the people can freely re-organize into a different hierarchy or elect a new representative at will.
I don't see how this works in practice. Power manifests in subtle ways. For example, you can have a pure direct democracy where individual voters nominally carry all the political power...
...but who decides what is on the ballot? Who determines the ontology of current and future policy decisions? Are closely-worded policies X ("ban abortion") and Y ("restrict abortion") the same policy with shared vote counts or different policies with separate vote counts?
If you democratize that power, you are subscribing for literally endless arguments over semantics.
See also district gerrymandering.
- anyone who can get 100k people to agree with them decides what to put on the ballot
- the parliament, which can shunt responsibility to the people after a best effort
- courts, politicians, in the end additional referenda settle disputes
District gerrymandering is also a uniquely Anglosphere-related problem that doesn't cause nearly as many problems in Germany, Switzerland etc
With no offense intended, us nerds on Hackernews tend to lose track of the simple solution ala "just ask people", "let people have a discussion", "common sense will sort it out over decades" while the rest of the population has no problems with things which aren't easily formalized
> "common sense will sort it out over decades"
More like centuries. USA has been a nation for 250 years yet we are teasing a second crisis of separatism. Immigration definitely plays a big role here. Political consensus takes generations to settle.
The nation thing is very much not the case for Germany and Switzerland at least. France more, and for homogeneous population maybe the nordics but the myth of the ethnically homogeneous long nation state is something that I mainly hear from conservative/right wingers who want to claim diversity doesn't work (not implying you are, just in general). The comparison with a colonial overlord carving out a centralised nation is also not too relevant, except that I agree that was bad and trying to impose European civic systems without also building the infrastructure and economy to support them doesn't work?
But Germany the nation was born in the late 19th century (Bismarck, Prussia taking over etc) and modern Germany has a complicated relationship to the old nation and nationalism in general. In Switzerland, they are literally 4 ethnic groups with regional fracfionalism tied together by will and democratic traditions, any ethnic stuff doesn't really apply there.
- Freedom of information So that leaves can error correct when corruption is detected
- Freedom to re-associate So that leaves can re-organize when the existing power structure becomes destructive to the leave's objectives. This may be a contextual or cultural shift rather than a direct form of corruption. (Eg: climate change may change many individual's priorities going forward). Imo pursuing this heuristic should preclude most forms of identity politics; I'd rather the leaves associate on philosophical priorities rather than on innate physical characteristics
- No special rules for leaves vs nodes higher in the hierarchy Or perhaps only more restrictive rules for nodes higher in the hierarchy
Your examples seem to have went back to a peer-to-peer model of decentralization; which I was agreed is inherently inefficient and untenable at scale. You need some hierarchical distribution of power, it's just that it needs to stay beholden to the leaves in the hierarchy. The person who decides what's on the ballot is the individual(s) elected/appointed to have that job. That person(s) is likely beholden to some pre-agreed upon rules for how to phrase questions, and any individual in the society can cry afoul if they abuse their position or if we need to update the rules with new considerations. All other leaves can choose to listen if they want, and choose to respond if they want, at whatever level of the hierarchy they believe is best suited to respond to the corruption. The hierarchy is not rigid, it's dynamic, evolves, and must be allowed to error correct as each individual sees fit. The only way that's possible is if it's driven bottom up rather than top down.
The objective should be to distribute and localize power as much as possible, because the more power is centralized, the more prone to corruption, less efficient, and less responsive to nuance it becomes. The exact laws and regulations that achieve that objective? Society is still working that out, but I'd argue separate judicial, legislative, and executive branches was a good move in the right direction. I'd also argue trial by a jury of your peers was also a solid move in the grand scheme of things.
I'd argue that same objective holds true for technological networks as well.
One of most successful decentralized system that is useful for what it was meant to do is DNS, which has this tree-like delegation in its core
lawrence@krubner.com
434 825 7694
Each individual naturally filters out information that they consider weird, the less they know the person that they are reading, the farther it is from their viewpoint, the more that idea will be passed over.
Each community is made up of the average of the opinions of its members, and we can collectively track this through the Overton window. The more people that don't know each-other in a community, the less their weirdness budgets are, therefore the slower the community is able to change its mind. This is where Dunbar's Number comes in, eventually you reach a point where the Overton window metastasizes and two factions in that community break apart.
So what we really need to be asking is...
1. Dunbar's number suggests that as a community gets larger, it should naturally split, why don't digital communities do this?
One obvious conclusion is that the internet sites that we are building them on top of do not allow the split. The website is owned by the web owner, and they want all communication on it, and they don't have any sharding built in, so we can't split.
But even when we build sites like Reddit that are inherently built of many communities and make it extremely easy to create new communities, we still see concentration of communities. So it is dismissive just to say it is the owners fault. It must somehow be that, because digital is so effective, the economies of scale within the digital space grows and even with the worsening of communication the community gets stronger with each member. That is, opinions within the websites Overton window have such good quality, that it doesn't matter that all discussion outside the window is ignored. And there are mostly more people entering than there are leaving, so it works out until it doesn't.
And thus as you say, the only option we have to reach beyond our local maximums is to split the communities soon after we reach something close to Dunbar's Number. Does that mean each community would just be 150 people? Does that mean each person could only be a community of 150 people local to them? No. You can have a community of the top community members, and that 150 community would have it's own Dunbar's Number. In fact, this is mostly what the House of Representatives and the Senate are, or was before 1920.
An interesting observation: the population of the USA since its founding until today has increased approximately by a factor of Dunbar's Number. In theory, this would call for the introduction of another layer of national bureacracy (say, a Regional level above the individual States). Yet our political structure has not meaningfully changed, at least de jure (de facto I believe it has changed substantially with the rise of capitalism in the early 20th century).
That is still true today. Anyone can still open a connection to a socket or make a server-socket available on that particular layer and can expect routing to work. Yet.
While it is tempting to equate "the Internet" with the world-wide-web - a term used less and less, they are two very different things.
The biggest danger to the internet is about control of the underlying networks. That countries can cut themselves off from the internet or monitor all packets being routed in large regions is problematic. Efforts into "decentralization" should start there.
That is a concern, yes, but I don't see it as the biggest one we have today.
The internet is already split up into service silos. Large corporations that service millions of users and control the flow and access to information has the same net effect as countries doing the same, and is much more prevalent today.
Do I care that China or Russia have their own internal networks and services that I can't access? Not really. The language is a bigger barrier in that sense if I cared about that.
I do care about public discourse happening on large private platforms where users have little to no control over the data they share or consume. This is what ultimately decentralization is meant to solve: users being in control of their data.
There's a lack of physicality, i.e. consider the difference between decentralized intellectual property (sharing patents etc.) and decentralized electronics manufacturing, food production, or transportation systems.
Those latter issues taken together result in things like the current disruption in global supply chains, for which both everyone and noone is responsible. 'Anyone can build anything' sounds good, but if your only chip source is China and they have an energy/pollution crisis and scale back manufacturing, then what? Wait a few years while the USA gets comparable facilities up and running, if that's even likely?
So perhaps you get economies-of-scale advantages with centralized manufacturing, but security-of-supply advantages with decentralized manufacturing?
Lately, I think of entropy as a friend more than a foe. all of the examples of privatization of technology he brought up, I think will eventually fall down due to the entropy. All these privatizations are attempt at optimization (hence the value extracted). But it is not free at all - optimization means specialization, and specialization means decreased flexibility and decreased chances of persevere the tendency of the world to digest systems and spit out chaos.
The “paradox” is things may be decentralized, but then finding and using those things in aggregate is hard.
Together with decentralization comes the idea that centralization is bad, which is wrong; centers are good, you just have to protect the sistem from abusive centers, and/or let people choose which center or center of centers they prefer.
In an alternare universe facebook has a moderation system users "subscribe to" that warns your client of what content you are likely to want to avoid but allowed you to "fork" it so that if you were displeased with their moderation you could run your own and have others subscribe to your moderation instead/too
The biggest problem with it is that, if a formerly-good server goes bad (or goes away entirely), everyone who used it is screwed, because their identity is tied up in the server. This means your most important criteria for choosing a server is stability. In practice, most of the user base picks old, established servers, hoping that the past predicts the future, and cementing a small oligopoly who can then use their power to direct the network’s future.
Real portability, like Matrix is working on and Scuttlebutt already has, helps with this problem. If your user ID isn’t tied up in a domain name, then you can try out hosting your own server, and switch to and from it without much risk, so more people will try it.
A possible half-assed solution would be to have your identity be linked to a public key you can move between servers.
Obviously non-bitcoin-like network will have nodes more important than others on which users will ends up relying. The solution is not to delete any kind of "centricity" but rather to make nodes easy to replace.
A similar point was made by Scott Alexander:
> HL Mencken once said that “the trouble with fighting for human freedom is that one spends most of one’s time defending scoundrels. For it is against scoundrels that oppressive laws are first aimed, and oppression must be stopped at the beginning if it is to be stopped at all.”
> There’s an unfortunate corollary to this, which is that if you try to create a libertarian paradise, you will attract three deeply virtuous people with a strong committment to the principle of universal freedom, plus millions of scoundrels. Declare that you’re going to stop holding witch hunts, and your coalition is certain to include more than its share of witches.
https://slatestarcodex.com/2015/07/22/freedom-on-the-central...