On Decentralisation Alex Feyerke / @espylaub Decentralize Camp 2014, Düsseldorf http://decentralizecamp.com/stuff Before we start: decentralizecamp.com/stuff, etherpads, so we don‘t decentralise our note taking, but take advantage of collaboration and a centralised index of information :D Hello everyone. My name is Alex Feyerke, I'm a frontend developer from Berlin, and my talk is called "On Decentralisation", because it makes me feel like an eighteenth-century psychologist or something. It was originally called
First I thought, woo, this is going to be easy. Jeremy raised a lot of excellent points on why centralisation is dangerous (imagine facebook closes down like geocities, while being much more interconnected). And we all kinda know what we‘re talking about, and I talk about this kind of thing all the time in the context of this Open Souce project I work on:
Hoodie. But the more time I spent on research and reading the clearer it became that the topic of decentralisation has a lot more facets than we commonly acknowledge as tech people, and that, as tech people, we tend to start thinking in the realm we're most competent in: technology. But the
Decentralisation Technical Infrastructural Economical Cultural Judicial But that hardly the only realm where decentralisation can affect the web [read] any number of ways. But before we dive in to those, let's go for the low- hanging fruit first, and take a look at the past. Maybe that has some clues for us.
Re-d14n One term that gets used a lot is the "re-decentralisation" of the web, which I think is a bit dangerous, because it implies this idea that everything was better in the late 90s. The gist of it is that back then, the huge silos like Facebook didn't exist yet, and the semi-closed offerings of access giants like Compuserve and AOL were crumbling. People had their own websites, self-hosted everything, and their URL was basically their means of identification.
Hello my name is Ses Tantek Celik. Conference badges had people's URLs on them, not twitter handles. Your domain was where your mail contact was, where your content was, where you lived on the web. You were in control.
ugh. This is nice, but it ignores the fact that having your own domain and running and deploying your own stuff was… difficult. Time-consuming. It was something for very patient people and specialists. People slogged through it because there were no alternatives. Yes, you could live within the AOL messageboard system, and many did, but that was quite restrictive.
When the first silos appeared, like Flickr, they genuinely made things easier, and publishing content on the web massively more accessible. And the self- hosted people? They used it too, of course. Flickr was built by people well- known in the community, it was a friendly silo, as Tantek Celik calls them. And it allowed connections with people who didn't have their own domains, the less tech-savvy could now also benefit from this new technology. Flickr was empowering, and it solved a problem: putting photos on the web was too hard, and connecting through and about them even harder.
web 1.0 wasn‘t for everyone And therein lies the fundamental problem with "everything used to be better": conceptually, it sounds better. In practice, the old web, web 1.0, was pretty elitist and exclusionary. Sure, with enough patience, "anyone" _could_ build a presence on the web, but as I said: ugh. Difficult. Unfamiliar. Hostile. Small audience. Steep learning curve. Not welcoming at all. The good old days were mainly good for a small number of people. The centralised services actually provided a valuable, empowering service.
Theoretically, self-hosting is a noble thing. But in practice, convenience trumps ownership. Just a few years later, all that distributed, self-hosted wonderfulness became stale, too big, too impersonal, too unfocused. Here's Wired in 2008 telling people to stop worring and love the silo, because blogs had seemingly reached the limits of usefulness.
Blogs were too text focused, photo- and especially video blogging was still difficult, professional blogs with teams of writers and editors appeared, it became harder to reach and connect with people.
• RSS OPEN WEB TECH FOR HUMAN CONNECTIONS RSS helped a bit, but what also helped drive bloggers to the silos was the open web's lack of an integrated backchannel. And there were feeble attemps at making more connections:
• RSS • Track- and Pingbacks OPEN WEB TECH FOR HUMAN CONNECTIONS - There were Track- and Pingbacks, but honestly, those systems have always been rubbish. Terrible to implement and of very limited practical use. I‘m pretty excited about webmentions, though.
• RSS • Track- and Pingbacks • Blogrolls OPEN WEB TECH FOR HUMAN CONNECTIONS - People had blogrolls and rings to enable some form of discovery, but those were always manual and also had very little attractiveness or utility.
Silos won because they met a genuine need In any case, nostalgia alone isn't going to help. Wanting to radically de-silo the web is a kick in the shins of everyone who gets genuine utility from them, and that's a _lot_ of people. Simply rewinding the web, giving everyone a domain with a blog, that's out.
Silos can show us what people want So if we acknowledge that, evil as we may think they are, for all their faults: silos have some sort of legitimacy because they are, fundamentally, enabling technologies for millions of people, then we can't simply strive for their destruction or quick replacement, at least not in the way we've been trying to. That's just arrogant and patronising of us.
Simply trying to replace them won‘t work Additionally, all decentralised Open Source attempts at building silo replacements have been met with failure so far. The reasons for this are numerous:
• Too tech-focused • More complicated • Less usable OPEN SOURCE SILO ALTERNATIVES - Too tech-focused: design, UX and marketing as afterthoughts. "Architecture Astronauts": tech came before human needs. - More complicated than silos: not actually feasible for everyday use for all but the most enthusiastic - Less usable than Facebook or Twitter so they Never reached critical mass
„Let them host nodes!“ was the new „Let them eat cake!“ Diaspora is the prime example. "Let them host nodes!" is the new "Let them eat cake." Big and centralised were obviously evil, so small and decentralised must therefore be good.
The centralisation of the web goes way beyond social networks and personal publishing But there's such a vast gray area in between, and the centralisation of the web goes way beyond Social networks and personal publishing. Maybe we should first ask:
Which problem are we solving? Which problems are we trying to solve, anyway? It's tempting to look at this from a purely techncial perspective, but centralisation and the issues that stem from it take many other forms on the web:
•Surveillance and security •Closed ecosystems and silos •Advertising, and private data as a monetizable resource •Techno-cultural hegemony •Monopolies and prevailing economic models All of these are either problems or more or less broken, and all of these could potentially benefit greatly from increased decentralisation. Let‘s talk about a couple.
Small pieces, loosely joined. That's one of the ideas fundamental to the web, right? And yet, this is completely opposed to our industry's idea of success:
• Fail • Be assimilated • Replace a large company STARTUPS ARE EXPECTED TO 1. fail 2. be assimilated into a large company 3. replace a large company In all cases, the result is a convergence of knowledge, power and connections. Hardly surprising that the web looks the way it does: dominated by a few huge companies. The whole underlying concept is opposed to small pieces. You're not supposed to bootstrap your way into a sustainable business, you either bet everything on disrupting something or other, or you go home.
Startup culture is toxic In more than one way It's actually doubly toxic, because startups don't exist in a vacuum. Berlin is trying to attract them, but apparently no-one has thought through what that will mean in the end. Most Berlin startups will fail, maybe one or two will become a big, self-sustaining company, and
That [leaves] acquisition by an US company. [This will] ensure that neither qualified people, nor tax income will be left in the city as soon as something of significance will emerge here. — Igor Schwarzmann Third Wave Weeknote 158 This is cargo cultish imitation without understanding.
Digital Mittelstand Igor goes on to suggest the European tech scene not categorically look to California and try to copy what happens there, but instead look closer to home and establish what you could call a "Digital Mittelstand", smaller, sustainable companies that actually provide value to everyone You might also use this to find ways t re-introduce long term thinking into the tech industry, not just financially, but also about your resources: people‘s data.
Digital Mittelstand Ja, bitte And there's enough people around to prove that this is viable, Maciej with Pinboard for example, or many others in Berlin and the rest of the world that are neither huge nor decadently funded nor famous, but that just work, both as a product and as an economic enitity.
Decentralisation THIS IS ALSO This is also decentralisation, and if you want to decentralise the web, this is an aspect you can work on, too, today. Economic structures and economic culture. That's one thing we're trying at Hoodie. We don't want to be bought. We don't want to create a massive monopoly. We want to grow naturally, make empowering products and live good lives.
Decentralisation THIS IS ALSO So working on decentralising the web can be something as unintuitive as teaching app developers how to properly price products, or how to design better onboarding flows, or simply giving them better tools to more easily build products and services. Because more successful smaller services means more diversity, and that decentralises users, data, money and power as well. And that's all in our interest.
Content Content is actually one of those decentralisation success stories, much to the dismay of the print media, record labels and, best of all, television. Shirky's Cognitive Surplus is actually being turned into culture, and it's finding an audience thanks to the web. On this level, the web is doing what it's supposed to, and it's often doing it, paradoxically, through silos.
Cognitive Surplus THE WEB AS AN OUTLET FOR OUR Of course, the web is only half the story, accessible devices for production are also important, but in general, more people can produce content at lower cost and distribute it more widely than ever before. But content is also a problem:
Media licsensing encourages centralisation At the same time, content also causes problems. Movies, music, games… you _can_ sell and buy these on small, independent platforms, but due to how licensing currently works, building media empires is highly encouraged, because acquiring content to sell is such an enormous hassle. And if you're selling content, you either want to have a selection that is as complete as possible, or a catalog that interconnects with other catalogs, so buying is still nice and simple.
The latter is very hard, for many reasons, which is why we get so much of the former. iTunes, Amazon, Netflix, Spotify, Steam, YouTube… having all content in a central registry is simply convenient and operationally preferable to the user. You get synergies with other content you sell, and there's higher discoverability of content, too. In short, there's an opportunity for user lock-in: the possibility that the customer will buy everything there, for convenience's sake. Which in turn is terribly attractive for the service.
Silos empower both producers and customers HARD TO ADMIT both customer and seller are actually often hugely empowered by these huge central catalogs. There may be better ways to empower them, especially in cases like spotify, but this is empowerment. And it‘s better than what we had before, at least in the music and games industry.
Size has advantages when it comes to content HARD TO ADMIT And this is where things get messy: this is mainly an economical and a behavioural issue. I suppose you _could_ theoretically go and build something that aggregates, say, 500 indie labels' online shops with a central shopping cart and discovery system, but you'd quickly find that you'd have to take a cut to keep running it, and suddenly you've built iTunes…
And indie music is an apt comparison. Yes, there are enthusiasts that will go to the record store around the corner, or buy the unreleased vinyl at the band's gig, or order stuff directly from the label. They exist, and they're analogous to people who use PGP, or host their own data and webDAV stuff from a mac mini in their closet via DynDns, but they're not the majority of people. And even they are swayed when something massively more convenient comes along. And currently, that massively more convenient thing is most certainly centralised somehow.
Centralisation on the web has causes that go beyond the purely technical Many content silos don't exist because they've forced themselves upon us, but because they work well and genuinely solve a problem. Their centralisation is firmly rooted in the non-web real world. There are economical, psychological and judicial reasons for centralisation that you very probably can't just hack out of this world. Let‘s look at a different aspect
Advertising and "Kostenloskultur" People overwhelmingly expect things to be free on the web. German politicians made up a word for this: "Kostenloskultur". This expectation directly causes centralisation effects: how do you fund something that should be free? With ads. When do ads become a feasible income strategy? When you get a lot of people looking at them. The bigger you get, the more users you accumulate, the more money you can make. Same with selling personal data: it's only really valuable in bulk.
And make no mistake, people have been trying to find alternative means of funding small producers of content on the web, flattr, paypal donations etc. But even if these things are totally seamless and and really well done, and I consider flattr to meet those criteria, you can help a bit, but you still have exactly the same problems as before: people expect things to be free, and the more users you can attract to your offering, the more money you make.
And then you're intrinsically disadvantaged towards something like YouTube, which offers content producers a cut of their ad income. Because YouTube is huge, it can offer better tools, more reliable services, larger ad clients, more users and better discoverability.
The next big thing is a billion tiny things BUT… So I like the notion of the next big thing being a billion tiny things, but this doesn't apply to all aspects of the web: economies of scale are a real thing. There are still limits to the quality of the things you can expect from a huge network of tiny producers. Making good things generally takes time, resources and organisation. And then there's this totally weird place where centralisation enables decentralisation:
Kickstarter and co. They're services that use the intrinsic benefits of centralisation to enable decentralised production of culture on a more professional level.
• Foster a culture of value • Build tools that enable small, not replace large HOW TO DECENTRALISE CONTENT? So how to achieve more decentralisation in a content context? - help establish a culture of value - build tools that enable diversification, discovery and funding of products and services. I count something like Stripe as one of these, because it completely removes the pain of receiving payments and empowers more people to charge for stuff. It might turn into paypal, sure, but in the meantime, it helps create diversity on the web.
Humans are connectivity addicts The web is about connections between humans. Connections create happiness. Likes create Dopamine. Being able to share, to be visible, to be validated, to be acknowledged in their existence. That feeling trumps nearly everyone's sense of dread concerning the privacy implications of such behaviour. You're not just up against huge, semi-malicious corporations: you're up against human nature, too. People value this experience.
Humans have trouble valuing the intangible Which is a nice way of saying: people generally don't care about things that don't immediately affect them. Sascha Lobo had a wonderful example at re:publica this month:
This is Germany's bird of the year, the Common Snipe. One of the reasons it's thriving is the Bayerischer Vogelschutzbund, the bavarian bird protection federation. You might think: hm, a regional organisation dedicated to protecting birds, how big and powerful could they possibly be?
120 full-time employees at 25 locations, with up to 75000 volunteers and a budget of 10 Million Euros a year, mainly donations. That's what our parents do when they care about something. The German organisations that fight for a free and open internet have trouble getting even a tiny fraction of that in donations, and they can barely pay less than a hand full of people for their work. Because people take the web for granted. Access to it is a human right now, remember?
V WEB Imagine a Bayerischer Vogelschutzbund for the web. 120 full-time employees and a budget of 10 million just in Bavaria. That could enable so much, from lobbyism to education to technical development and unified communication in the name of a more free, more open, more decentralised web…
Privacy, Surveillance and Security The invention of the airplane is the invention of the aircrash. There's inherent dread in all our beautiful inventions, and the internet is no exception.
open = vulnerable The fundamental problem, as any psychologist will tell you, is that openness and vulnerability go hand in hand. This is true of humans as individuals, it is also true of the systems they construct. It‘s the basic schism in the open web.
The whole thing is a shitty battle of attrition between what we all want for ourselves and our families and the ways we need community to survive as humans — https://medium.com/message/81e5f33a24e1
[It‘s] a Mexican stand off monetized by corporations and monitored by governments. — Quinn Norton Everything is broken But what else is there? Surveillance is inherent to the internet, just as traffic deaths are inherent to traffic. You can minimize the risk, but you can only do so by reductions in freedom and convenience, or an improvement in technology. https://medium.com/message/81e5f33a24e1
ABS Airbag ESP And these many improvements in technology just work, they're all passively implemented systems, you don't have to _do_ anything. ABS, ESP, Airbags, crumple zones, sensor systems: dozens of these technologies exist, some became law, and they all work because they require no effort or even knowledge of their existence. Except for putting on your seatbelt. And it seems absurd in hindsight, but people fought the seatbelt tooth and nail, because it reduced their freedom and convenience.
In fact, people still do. Says Felix Schwenzel in his re:publica talk a few weeks ago. But it's not entirely true, is it? You can minimize the risks of traffic through
• Law • Culture • Psychology • Infrastructure INFLUENCE TRAFFIC SAFETY THROUGH law, through culture, through psychology, infrastructural changes… many ways. Maciej said yesterday he'd even appreicate laws that force some sort of standard on him when it comes to user data.
The best security features are those that require neither effort nor even knowledge of their existence. But the thing is, as Benjamin Franklin said, that you can't have both freedom and security. You can have 100% safe traffic, but that would require taking humans out of the decision-making process completely. But what you can do is acknowledge that the best security features are those that require neither effort nor even knowledge of their existence. Because people are - lazy, habit driven and stubborn - terrible at calculating risk, and therefore the value of security measures
The best security features are those that require neither effort nor even knowledge of their existence. The easiest way to make the web safer and better is basically to not tell anyone you're doing it. Which brings us nicely to the next part:
Technology There was a distributed social network in 2004, called Appleseed. There was a completely impenetrable and private p2p sharing and communication tool called W.A.S.T.E around the same time. There was PGP, there was Diaspora. There's been distrubitive tech around for ages. It never caught on. Because simply providing the technology isn't enough.
Technology There was a distributed social network in 2004, called Appleseed. There was a completely impenetrable and private p2p sharing and communication tool called W.A.S.T.E around the same time. There was PGP, there was Diaspora. There's been distrubitive tech around for ages. It never caught on. Because simply providing the technology isn't enough.
We believe that when a technically- savvy elite of enthusiasts build tools and technologies for themselves, that technology will eventually trickle down and help less technically-savvy members of society.
And, just like trickle-down economics, it doesn’t work. — Aral Balkan Trickle-Down Technology Tantek Celic called the people who make these things "architecture astronauts", because they build massively complicated technology that is very far removed from the wants and needs of the actual humans who might have use for it.
„Alternatives“ Much of this open source tech comes along as a more secure alternative to existing tech, but they're not. They weren't as good as what they sought to make safer. They were just safer, but in all other aspects, less desirable. It's completely unsurprising none of them have succeeded at making any notable impact on the web so far.
„Alternatives“ If your decentralised or more secure technology is set up like this, it's at best a temporary solution for a small group of enthusiasts. If it's more work than what people already use, you'll have an extremely hard time. If it's not massively better that what they already use, you're doomed.
It‘s made of people You need people to do this, many people. The web is, after all, about people. The internet is the place where no-one knows you're a fridge, but the web is for people, for people to connect. To connect, they either need to be in different places that are interoperable (i.e. email), or in a single place that has critical mass (i.e. Twitter). We've already established that we‘re sceptical of the latter.
POSSE Post on Own Server, Syndicate Elsewhere It's not practical to move to dispora if none of your friends are using it. so the first step is interoperability (PESOS or Reclaim Social Media-style, or Indieweb's POSSE and brid.gy work along those lines: you have your own server, but you're still connected to all your friends.) I really like this, and I found Tanteks recent talk on this really motivating. But the problem is: the silos provide benefit that goes beyond merely connecting with your friends, even if this is the main attraction.
Meh. Sweet! Instagram simply makes my iphone's photos _better_, and it's fun, quick and easy to use. Facebook's size is a guarantee for utility and stability. POSSE can't help me there. I can‘t POSSE to instagram, for example. And you can‘t realistically do this with every new service. And have you seen the devices people are increasingly are browsing with? You can‘t really build a website on a phone.
You can‘t really build a website On Glass. In Oculus Rift. Over Siri. There is an actual global hardware trend that works against self-publishing, because these devices are only for consuming services, not building them. And Plus: You're demanding a trade-off, or even a sacrifice from users. Since most of them don't even appreciate that a problem exists in the first place, I think this is going to be a niche solution from the word go.
„Alternatives“ And it's naive to assume that there can be Open Source alternatives to every silo. How would that even work? Who's supposed to build and maintain that at a level even remotely comparable to the big silos? Flickr's been around for a decade, and it's a fairly simple use case. Where's the Open Source version? What would that even look like?
We need great self-hosted applications, which we can use to manage our emails, personal pictures, documents, private messages with friends, blog posts, etc. —Tom Dale Progressive Enhancement is Dead I used to think that opening Open Source to more people from more backgrounds could help, bring in UI and UX people, designers, concept people, have a more holistic view of Open Source development. And that is absolutely neccessary and will help Open Source immensely, but it won‘t beat the silos. Consider what you're up against: Google. Facebook. Apple. Amazon. Governments. The companies spend billions a year buying up young blood, like monopolistic vampires.
Today I'd say the only way to maintain composure in the face of this opposition is in fact to stay tiny, and to focus on a field where it's not a liability if your decentralisation project only gets picked up by nerds and experts, but actually the whole point: build infrastructure instead of services. Build protocols instead of products.
Stealth d14n Our efforts at decentralisation need to blend into the background, become the underlying principle of everything. People in general simply don't value their privacy enough to actually invest money or make any additional effort to learn new software or make convenience sacrifices. There was a medium article yesterday about endangered intelligence sources, and even people who know they‘re being surveiled will often rather take a risk than wait for a secure opportunity to transmit information. Even they.
if(pain > inconvenience){ doSomething(); } else { //meh(); } In my opinion, decentralisation of the web will not come through individual services, like Diaspora, or widespread sale and adoption of something like the freedom box, Protonet, ArkOS or or similar host-at-home/in-the-office solutions. They will undoubtedly help a bunch of people, but mostly businesses and nerdy people like us. The vast majority of people don't care, because they can't directly feel the negative effects. The Stasi would ruin your career if you said the wrong things to your neighbours. That was a daily, real threat. We‘ve got a lot of crazy stuff happening, but it‘s not that intimate for most people.
If decentralisation is to become a mainstream thing, it can't be a thing at all. It has to be a fundamental part of the technology and the platform, and not a feature of the service. And it will gain traction not because it is the morally right way to build things, but because it offers tangible operational benefits for those who implement it, like not having to manage your users' data, scalability, security.
And infrastructure innovations are surprisingly feasible to achieve for small teams or even individuals, just look at bitcoin and the blockchain as an inspiration.
Take something like maidsafe, a blockchain-based storage system. The general idea is that data isn't stored in a single place by a single entity, but broken up, encrypted and distributed across everyone who uses the platform. Instead of paying for a service with your personal data, you're paying for platform use by allowing other participants in the platform to use your machine for storage, just as you use theirs.
That's where we get back to Hoodie. Hoodie lets you build complete, data-driven web applications with user auth, data sharing etc. really quickly, from the frontend. We‘ve got CouchDB to build on, and with that, we want to build a toolkit that helps people build web apps, but with a new, different, and more decentralisable structure underneath. And this takes multiple forms:
noBackend We‘ve got this noBackend ideology, which basically states that building for the web doesn‘t have to be this complicated. Lots of people will do just fine with another layer of abstraction that hides the backend behind a human-friendly frontend API. And this simplifies development, it massively lowers the barrier to entry, and it lets more people make better stuff. This hopefully leads to more small products and services.
Offline First Local user data One Database per User We package in a sync engine and a database design (one database per user, to oversimplify it) that allows you to easily keep user data on the user machine, which has a number of benefits (again, recall Jeremy warning you about giving other people sole control of your data), but it‘s also a foundation for the future. CouchDB is brilliant at replicating data between nodes, which means you can later build p2p systems with it. And it‘s master/master replication, it doesn‘t really care about a distinction between client and server. The design is in place (one DB per user), the tech isn't yet, but we‘re trying to lay an infrastructural foundation. That‘s not optional, it‘s just there.
Dreamcode We aim to make the tools great to work with on every level, so we help people migrate to the easy thing, so they get the good thing. It the tech equivalent of a Trojan horse.
hoodie.account.signUp(`name`, `password`); hoodie.store.add(`todo`, {desc: `Say hi`}), A toolkit that is a pleasure to use, but also imbued with values and opinions and a foundation for what we think makes the web better. But you don‘t have to care about that.
Don't tie d14n to a service. It's never going to take off as long as content through critical user mass is a factor. And decentralisation as part of a single product or service doesn't really help the web as a whole. I wouldn't bother trying to sell your thing as safer or more secure or more private, because nobody cares. And by now, it's hard to imagine what would make them care. Don't build individual services or products in a decentralised way. You'll be expecting people to sacrifice interoperability and content for benefits they don't care about. That's going to fail.
• Interested audience • Less brutal competition • Your work is multiplied • Small team, big impact • Business opportunity GO FOR THE INFRASTRUCTURE LAYER Go for the layer beneath, where the makers, are, not the users. - your audience is reachable and cares - disruption in this space is totally possible, less competition for critical mass - If you build a single decentralised service or product, there's only that. Build decentralisation into the infrastructure level, and your efforts will be multiplied by everyone who builds on top of it - your project can be smaller and more manageable - you can sell your expertise on the platform when others start building on it. No ads, no data selling.
So yes. It‘s not just a technical issue. Appreciate the fact that humans are creatures of habit, and may have to be tricked or educated into acting in their own interest.
espylaub
takatsunobuhiro
kobayang
riru
pistatium
coe
soracom
devopstaiwan
yosuke_furukawa
oracle4engineer
karawoo
miyake
whywaita
jensimmons
lemiorhan
paulrobertlloyd
maltzj
danielanewman
brianwarren
michaelherold
keavy
morganepeng
soudai
afnizarnur
philhawksworth