Decentralized Web Summit
we're live streaming this at www.decentralizedweb.net so watch along
if you haven't been to the Internet Archive before, we think of this as our Temple of knowledge
we're one of the largest digital libraries in the world - if you look at the back of the room you can see servers
the library of congress, if digitised, would fit in about an inch and a half of one of those servers
the Ford foundation got together Mitchell Baker, Brewster Kahle and Tim Berners-Lee and asked for a moonshot
rather than talk about the new library of Alexandria, Brewster talked about Locking the Web open for good
many of you are the builders of the next decentralized web and baking openness into your code #indieweb
Joi Ito likes to talk about the conspiracy of open - Mozilla, Internet Archive, Wikimedia, EFF keeping the web open
Mitchell Baker - Mozilla Chair
today we have 2 concepts: Decentralized and the Web - I want to talk about the Web
you might think that the web is about the browser; or that the web is about HTTP
the web is not about a particular technology but about 4 key principles
while we build a decentralised web or a Web of Things, what do we care about?
1. The web is Immediate. Safe instant access to content via a universal address without needing install
2. The web is Open. Anyone can publish content without permission or barrier and provide access as they see fit.
3. The web is Universal. Content runs on any device or platform. We do this through standards.
The w3c has a Social Web working group, and Mozilla is active in this.
4. The Web has agency. A User Agent can choose how to interpret content provided by a service on your behalf.
The browser can change typeface, protect you from malware, prevent tracking. The browser works on your behalf
as we think about the decentralized web, these key traits show up: Immediate, Open, Universal, Agency.
for the last few years Vint Cerf has been talking about the digital dark age - content blinks on and off
We built the Wayback Machine to give the web a memory, but that's a patch
Vint said "we all die" "when I die I want there to be a record of what I did on this earth"
there's nothing more satisfying than being the chief internet evangelist in this room
May all your packets land in the right bit bucket [amen]
Lessons learned form the Internet:
Collaboration and Cooperation were key to the internet - get everything talking to each other
open design and evolution was key, and anyone can join by following protocols
we didn't impose any business models on anyone, so you can choose how you do it
modualrity is key, layered evolution has been enabled by information hiding
E Pluribus Unum - from many networks we made one thing, the Internet
thinking about archiving there are a series of things we need
Compression schemes like tarball and so on only let you store and recover particular objects
storage and software with versions is now growing and possible
but the web is a complex structure that connects lots of things together
the internet archive takes static snapshots of web pages that are time-indexed
the hyperlinks need to be re-formed to resolve to in-Archive pages so it can be self-contained
this archive requires continuous crawling, and decide when it is a new instance
a system that is accumulating web pages - can the set of all sets contain itself - is the archive in the archive?
the web can barely contain itself as it is - can LOCKSS work - is there space to replicate it?
Hyperlinks deteriorate over time - do we need a permanent link system?
HTML rendering needs to be backward compatible - even the earliest HTML still renders well, but there are hazards
I uploaded my slides from Powerpoint to Google Docs, and found Google Docs can't render TIFF though PPT did
I have some 1997 Powerpoint slides - I pulled up one in current Word, and it was corrupted.
Power corrupts, Powerpoint corrupts absolutely.
there are issues with permissions on things that the web points too
I would like for web pages to be automaticallly replicated and archived as we edit and publish them
I'm not sure I want all my edits shared and archived, but certainly all my published versions
Is there a role for the Google Docs like real-time replication and synchronisation?
could we have a common reference scheme across copies
is there a role for a Pub/Sub mechanism for co-operating archive entities?
lot of metadata is needed to replicate the Time Machine of the Wayback machine
it is OK to have control of replication and distribution while still archiving what we can
the other question is can we render the objects pointed to - we need software to do that
50 years from now we may have software that no longer runs, which means we need to emulate old hardware
running old Windows versions in emulation in the cloud - we will need to get legal rights to do this
imagine that there are multiple alternative resolution targets for a link - doesn't matter which is used
when we move around copies of things, how do we make sure we can still point to them
newspapers and magazines have editions; on the web we may need snapshots at intervals, or define editions
desirable: automatic archiving on publication - a service you sign up for? how funded? [ask @pinboard]
registration of rendering engines and permission systems
Do we have automatic malware filters? How do we know it is safe
do we have varying fidelity levels - everything, surface visuals without links, what else drops out
how do we even describe how we have lowered fidelity and why
once archived, is a page an indelible instance usable in a court of law?
does this make an official record?
could we make this self-archive be encrypted and unlocked after 25 years?
what metadata can I put in to trigger opening up access after time
Wrapping in containers may be a way to preserve the software as well as hardware
apps are taking over eyeballs and advertising - can we include them too?
google is finding a way to run native apps in the chrome operating system, which is an answer for apps
we need to index things that run in the apps, and archive them
I have an image in my head of a diorama of the Jurassic period in a museum - people will model them
You didn't mention digital vellum. Is that to far down the stack?
my favourite term for this is digital vellum - that was focsued on an object and its rendering
what we're challanged to do in the web is more difficult to preserve the links
I'm Samantha and work in the VR space - how do we have this conversation about 3D VR on the web
the app space for me has grown almost out of control - it takes too long to find them and run the one I want
I don't want an internet of things to have an app for every light switch
some people are talking about using web apps instead of standalone apps for this and that looks promising
there are various models to fund things - you can put ads in apps and the web, though subscriptions models work too
Samantha: I like showing people that you can use the web for everything
given the rise of intolerance and anti-semitism, what about the right to be forgotten?
The right to be forgotten is weird as we have to remember everything that must be deleted so it doesn't come back
it's not just the stuff that we generate, it's stuff that other people generate that impacts privacy
people do ask to take a picture with you when they are taking selfies - we can grow that etiquette
we have a library of emulation here - the emularity is ours that lets you play Oregon Trail again
someone said it's kind of like the beatles here - Paul McCartney was here and John Lennon just walked in
Tim Berners-Lee once said "the pieces were all there, I just needed to do was step back and connect them"
thanks for inviting me and thank you everyone for coming. re-decentralizing the web is a topic near to my heart
anybody who just woke up here in a yellow light-filled church with sacred relics containing all human knowledge
Vint and Bob Kahn had done their work 20 years before -1969 vs 89 - totally different music
the internet was there and I could use it - my boss did write "vague but interesting" on my proposal
my intent was to make something Universal - I have seen a lot of documentation systems that made you use theirs
some were matrix oriented, some were tree oriented, there was this heterogeneity of systems
we had to put code on these documentation systems to convert them into universal web pages
HTML had a lot in common with the SGML markup that people were using in my department
HTTP was a concoction of SMTP-like headers with the HTML content
the URL, then called the UDI, was made to look as much like a unix path name as possible
the double slash came from the Apollo name system, where it could refer to another machine
incremental change and only adding the pieces that we needed was key
it was never clear that the web would take off, but it did
the internet was designed without the nation being a concept- it's not obvious where users were
making a website was downloading the code, running the httpd daemon and putting pages on the web
so just by linking from your blog to other people we made a web of intelligent discussion
in the years that followed, many wonderful things have been done on top of the web
creativity built many wonderful things, even giant social networks where people spend all their time
when you talk to people they are quite frustrated - they have their friends on Facebook, photos on Flickr
then you want to share between the systems, you have to import either photos or lists of friends
you have to build an app, learn the Flickr API and the Facebook API and everything is stuck in silos [km: try brid.gy]
the web was designed to be decentralised so everyone could participate by having their own name and domain
but instead everything is locked up in silos
the consumer gets a lot of things for free, but they sold their soul of privacy to the marketing machine
you suddenly get targetted with all kinds of things - because the system realises that you have a child
the only way to make money on the net seems to be with advertising
what's wrong with this picture? It's a myth. We don't have to be happy with this.
when I go for a run my wristband tracks my run and them gets uploaded to the internet
if I uplaod all my running it can maybe work out when I need new shoes, but that isn't that valuable
the proposal is to bring back the idea of a decentralised web - power to the people
breaking the model where all your data is in different silos - we have project called solid
we're going to use web technology but separate the acts from the data that is used
when we store data where we want, we can have apps that connect to more than just one silo
we had to use the domain name system because it was there, but we could add a .archive TLD to back things up
there are systems that say "let's not use human readable names" - but they can be tricky
you should think of the URL as a name, not a location - you can attach public/private keys to it as a name
what we are all aiming for is to make the web better in lots of ways and more reliable
with HTML sub-resource integrity when you link you can give its hash as well to check it is what you linked to
a lot of the issues about naming, about should we fix DNS?
when I come up with a name like 'whencanwemeet.com' they want to sell it to me for $100,000 - not good
we write standards on how these different projects can use different pieces of each other
The Social Web Working group at the W3C is very much a part of this work
the web authentication working group is also a related area
I expect to see the merging of the world of sync and bits of the cloud, and the web itself
a lot of websites now have links to git repositories for their history - maybe we should surface that history
when we can see commonalities and difference between versions, you can have distributed sync and editing
I wanted to say how I am frustrated with silos, and excited that we are going to re-decentralize the web
first Happy Birthday [applause] - I liked the values. Can you give us 3 promising projects?
people ask me for top 3 websites and I always refuse - I think they are all trying different things
it's not a race to the moon, but people working on fuel, on rockets and on cabins that fit together
I tried to get my friends to not use walled gardens, but the design was too good. How do we get better design?
it's a good question. When you use any decentralised app you are getting an id for them.
Making the sign-up easy is difficult especially when there is more than one choice.
PGP is a great system, but it is very hard to use, and hardly anyone is working on improving it.
people working in the walled gardens do garden well, but the jungle outside is more appealing in the long run
if we do store all our data on something we own we can switch apps much more easily
thanks for you vague but interesting idea. The web was defined for documents, but so much of apps is data over the web
are there any standards for unifying that data layer?
The Solid project is that - github.com/solid - learn turtle and drink the Linked Data Kool-aid
the data stores have generic apis so you can add facts to a file. we need some standardisation, but need apps
we have to standardise how we store contacts and friends and invitations and acceptances
I'm wondering how you feel about global access to the internet - Africa is seen as mobile only not mobile first
how can we give people this unsiloed access we had in the 1990s
the number of people using the web when we started the Web Foundation was 10% of the world it's now 40%
over the next year or two we'll cross 50% and the web users will be in the majority
the digital divide has become bigger as we assume people are online by default
I'm Max: the web has got a lot faster and better over the last few years, but often via side effects
webRTC was made for video but meant we could make distributed file systems
one possibility is that the early adopter community is developers - we've used IRC for ages, Slack is a silo
we're using twitter here, which is centralised for a reason. Slack isn't - we could use IRC
we're talking about moving the web to content addressing, but there is a lot going on with Bitcoin too
web 1.0 was documents, web 2.0 was dynamic what is web 3.0 ? is it bitcoin enabling transactions? what is the frontier?
yes the web is a web of documents; when you run a web app, you are exploring data. Exposing data to users
exposing the web of data is key- people have apps for powerful ways to look at data
data should be archivable in standards so we can still have it in 20 years
the library community has been excited by linked data - but when we decentralise do we still have URLs?
HTTP urls were not designed as locators but as identifiers - we put a document then a # then after that meaning
so after the # you can use these identifiers to mean different parts of the document or what it refers to
what I would like to do is stick to using HTTP for the namespace, but allow different URL extensions
Sir Tim Berners-Lee said that decentralizing the web was an exciting possibility
Brewster Kahle – "Locking the Web Open – a Call for a New, Decentralized Web"
This has been a fabulous coming together towards rethinking what we might build out there
Tim Berners-Lee said HTTP is not cast in stone we can see what it would become
there is opportunity and possibility here to change things
Jeff Ubois said "can we lock the web open?"
over the last 25 years we have had millions pour our lives into the web
as Larry Lessig says Code is Law and our code defines the web
the way that we code the web will say a lot about how we live our lives online
we want to represent the values and structures we want to reinforce and enable
we should encourage universal accuracy to all knowledge
we want the first amendment baked into it
the web is huge - we collect a billion web pages each week - it has attracted people to come together
but it's not available everywhere - in China and Russia you can't get to the internet archive
web pages blink on and off line - the average life of a webpage is 100 days
it's a very ephemeral medium, and it is also not private
GCHQ, the UK's NSA equivalent watched all of the readers of wikileaks and handed them to the NSA
in the library world the idea of being watched as you read has a long and dreadful history
so being afraid to follow a link is a terrible thing
but the web is fun - it's a jungle out there but it's fun jungle
the web isn't reliable, it isn't private, but it is fun - 1 out of 3 -we can go for the trifecta
the idea of building a decentralized system I want to distinguish the web from the internet
with the internet, any piece can get nuked and it will still work - it is resilient to certain kinds of failure
a decentralized system is more difficult to make than a centralised one
if you watch all the people transiting an ISP to a webserver you can block people
the internet isn't like that. I asked Vint how hard it was, and it was 5 guys locked in a room for a year
the Amazon cloud has large scale use and datacenters all over the world - they'll move to different servers
this kind of migration based on use is desirable - can we make the decentralized web do that?
we want to make it reader-private - Wikipedia is a wonderful thing for people like me who should know things
also writer private is important, and that is easier than reader-private
lets build a time axis into this new web - let not have to make a kludge like the wayback machine
lets archive not just what the website looks like but the websites themselves and their data
can we make it live in multiple places and be able to roll it back in time?
and can we make it so that people can make money on the decentralized web without a 3rd party?
we have some technologies that are fricking awesome
at the time the internet was invented, encryption was illegal
we won those crypto wars in the 1990s, so now we can use crypto
for peer to peer, Bram Cohen and Bittorrent have made peer to peer work; with WebRTC you can do it in browser
with signing and encryption you can know that you are not getting a corrupted version
With blockchain and bitcoin you have a money system and also an API to money
but the mind blow for me was changing the idea that only the stupid survive - only the simple systems live
these are not simple systems - they are complicated but succesful
we can build a system out of some of these component to do something intersting
about 25% of websites are build on wordpress - can we make a decentralized wordpress so you don't need a server?
could we serve a wordpress equivalent from everywhere and nowhere?
we want it to work on existing browsers so you can use it on your phones without downloading anything
it's got to have good naming and snappy performance
it's got to be fun to post and comment, and it needs an identity system to delegate authority
extra points if someone can put a bitcoin in a slot to pay for your songs
and we need archiving along the way
I am going to do a live demo of linking to http:127 something and a big hash and it will retrieve on my machine
it will retrive from a distributed system and show my blog and a search field for it
if I search "vint" it is running code in my browser you'd normally do server side
We need easy names - there's namecoin in the room; maybe ethereum could do names so we don't type hashes
we're going to need institutions that will serve at high speed, like the archive and ISPs and CDNs
ISPs want to make things faster - so it's like everyone has akamai built in
updates of decentralization is not trivial - mutable torrents and IPSF do make this work
a decentralized identity system - bitcoin addresses for signing posts? soild crypto and existing tools
if we can piggyback on someone else's system we can make that work
so we can have WordPress, but decentralized. A lot of the pieces actually exist and need to be put together
we want a reliable, private and fun piece of infrastructure we can use for a very long time
we can lock the web open, bake the first amendment into the code and make it irrevocable
we can extend the work of Vint, Tim Berners-Lee and Mozilla Foundation and build something now on top of it
Panel – Peer to Peer Networks
IndieWeb says that you should all be able to own your own data, to set up your own domain
we diagram the internet with arrows going in one side, arrows out the other side and a cloud in the middle
we drew it as a cloud because we didn't have to worry about what happened inside it, we could think about the ends
As Tim Berners-Lee said he built on top of what Vint and co did before & just worry about the layer he was working on
with IndieWeb we take the existing web for granted, and build on top of that for people to connect their own sites
what this group has been doing has been looking at the pieces inside the cloud that aren't working for everything
and replace them with other protocols that solve those problems better, for problems you may not know you have yet
I want them to explain the problem they are solving for you lets start with @zooko
for 20 years now I've been working on what is now Tahoe-LAFS which solves the problem of censorship and permanence of data
Vint described the problems that Tahoe is all about -it's the strong web because links are not so fragile
I'm also working now on Zcoin which is like bitcoin but with privacy for the transactions
Tim mentioned the absence of payment being built in, and that has turned into a weakness we need to fix
I work on the IPFS project which stands for InterPlanetary File System.
with IPFS developers can publish data without it being in one place - this is an upgrade to the fabric of the web
while developing IPFS we realise that developers have common peer-to-peer problems so we built a library for this
the problem is the web isn't always reachable, and IPFS lets you get access to things that are closer to you instead?
it solves the problem of finding content on the network, you can run content or apps anywhere using ipscend
I've been working on a project called ethereum a decentralised answer to HTTP POST -
ethereum lets you run an application that would be on a server & allow requests to make transactions without a server
a POST request is what you create when filling in a form on the web, you press a button and the server responds
with ethereum you can press the button and you get a response without having to have a server set up?
Indeed, it all happens magically. It's very good.
I work on webtorrent - this is a torrent client that runs in the browser, that is native to the web
the torrent protocol is the most successful widely deployed peer to peer protocol in the world
but bittorrent and bitcoin and tor all are applications that you have to install, not use through the web
so the goal of webtorrent was to make the bittorrent protocol work natively on a webpage
so I as a website owner could make the videos on my site be hosted in a peer to peer way
the visitors to my site watching the videos could help distribute them to other people watching webtorrent.io
so the problem you're solving is "i want to post my videos, but not to the centralized place"
it's more general: if I'm a scientist with a big dataset I can stream from 1 computer to another without intermediaries
I can see some overlap here, but I don't want to get the projects fight that solve the same kind of thing
what we look for as users is does this abstraction work for me? do I need to do some special thing to make it happen?
the tension is between centralized and decentralized in cycles
you first build one version in one place and it is centralized, then other people copy it and make it decentralized
if you get it right and make it a protocol, then someone makes it a business, does a good job and everyone goes there
it ends up recentralized then - youtube is good example of that for video becasue it was easier than your own site
another example in the programming world is GitHub - git is a decentralized protocol
but GitHub made a central site that was a good place to share your code with other people and centralized
we've had a big wave of centralization and things ending up in silos for one company
I think GitHub isn't quite right, as competitors could up their game if they did it badly
the providers give a relatively small service and the protocol still is decentralized enough that we could move
think of the damage that will happen in the meantime - there will be lots of broken URLs
right, I use GitHub for my code and I would never move anywhere else because there would be a lot of broken links to it
the web breaks piecemeal, because individual links go down site by site - it's resilient or antifragile as it heals
but people build a better service that puts all the links on one server so they stay up
but they are creating fragility debt - they get taken over or make a management mistake and it fails in one go
we lose all of Geocities overnight, for example.
one of the challenges of the centralization/decentralization process is that a site death can take out a big chunk
it takes a long time to heal from that big chunk vanishing. How do your decentralizations defend us from that?
the lesson from the GitHub topic is that the links matter we have to make it start at the layer of the URL
Tahoe-LAFS urls are the centre of the architecture
git vs GitHub is key -github gives us a nice UI to the underlying decentralized git protocol
whereas Facebook does not have decentralized data underneath so it would be harder to move to a new place from that
i'm sure Facebook does have decentralized underlying data storage but the rest of us don't have access to it
I'd call that distributed rather than decentralized
we're really talking about the user experiecne and the developer experience
we want to give our users the ability to access the app independent of the network they are connected to
the way the web works today does not provide enough resilience to provide access wherever we are
we're very used to high connectivity, but when we move to a more disconnected environment with high latency
and if the servers are far away the experience crashes
and that's why I think the web is losing on mobile
well, it's not losing on mobile to a distributed system
each app is building a little browser that only works on one website and handles its own caching
the apps aren't decentralized but they have superior user experience in terms of networking
this is starting to change web standardization like service workers in the browser do allow offline apps
the ux is what matters in all these things
the distributed social networks lost out to the centralized 1 because making choices is harder, and 1 iterated well
i think that decentralization vs centralization is orthogonal to bad ux vs good ux
i know many centralized systems with awful ux
I worked on OpenSocial at Google -we tried to build a process where you could write an app and run it on any network
but each network wanted to make themselves special, and Facebook was big enough to say 'use our API instead'
we'll give you access to more users if you use ours
the problem there may well be trying to overlay openness on an oligopoly rather than rebuilding fully decentralized
that's fair - we were building on the structures we had - we built OAuth as a way to trust one silo with another silo
but not a pure edge to edge communication you still needed servers
one difference is that Facebook's centralized social network site is massively profitable because it is excludable
and Facebook therefore has enough money to be able to afford better ux
and that is more important then the distributed things and making choices
but if you have enough money you can fix those problems
by exclude you mean only Facebook can show ads on our site, or GitHub can charge to keep private
GitHub is more usable because it can afford to spend that revenue on making it so
so to turn that round, how can you make your systems excludable so people can make money on them to improve the ux?
I don't know - I don't want exclusivity - I don't want most of humanity excluded from something -
some kind of vague but exciting business model that are decentralized that don't have the option of excluding
ethereum is the leader of that - how humans can organise wihtout exclusion
successful things like Facebook had a good UI before they were profitable
that's because they had investment to build a good UI so they could make profits
I remember circa 1997 Google had just one search box which was a better UI than Alta Vista
If I go to a vc and say "I want to build this thing where I have no control" they aren't going to give me money
some of them are - there are VCs like USV who are funding OpenBazaar
we have to be aware of all the unfair advantages the centralized systems had at the beginning
we had to accept all these firewalls so it is now possible to connect to my neighbours house even if we
the networks were designed with centralization in mind, so it's easier to build that
I'll push back a bit - the web was designed to let you have servers everywhere, but client server design took over
now we can't host services from our homes any more
if you wanted to run something in the browser, until WebRTC you had to connect back to a server first
I have to talk to a server to get to his laptop next to mine
the centralization that is there by design is that you need a web server
but what changed from Tim's day is that you can't run a webserver on your own machine and have anyone see it
if all these things work well together, we still have to first go to a server to get the connection
do we need a way to request a static resource from one of these
in Brewster's demo he had to go to a server, but now we can run an ipfs node purely in the browser
can you explain who pays for all this decentralized stuff? I have a big web hosting bill - can I dump this on you?
people seem to think that you can have content up there permanently that no-one pays for
one answer is to align incentives in the system: with bittorrent you have to share to get access - we call it tit for tat
at the moment users sell their identity and the advertising space on their screens to provide them with the content
maybe they'll have micropayments to pay the sites they're using, or maybe the bandwidth will come from their machines
there could be micropayments to the users for their hosting and payments to the sites for the content
one of the things that bitcoin taught us is that it possible to put value on the network, not a central service
you can incentivize file storage on the network by paying to the nodes that have files available -see filecoin
you get 4 answers as it was a very good question
there is a model where you still pay as currently for hosting, but you are not vulnerable fro them to spy on you
we can use security to separate the links from the hosting
any kind of consensus process has scaling limit beyond which it becomes centralised, but ethereum is trying something else
bitcoin has thin clients wihtout having to bear the costs of a full node - can ethereum have thin clients?
the go team have prototype of a thin client; we'er working on a hybrid client that syncs fast but is still a node
because ethereum is an app platform, the problems are magnified compared to bitcoin
Thank you for pointing out the elephant in the room - centralization allows tollbooths that produce businesses
Panel – Naming & User Identities in Decentralized Networks
the rise of socal network services has grown use of the web, but it has driven centralization
if you talk about censorship in China, that may help them think, and Identity Theft is growing due to bad naming
Christopher Allen is one of the OG's in the room -
@Muneeb works on blockstack - his advisor assumed he wouldn't finish his PhD, but he is defending it soon
certificate authorities are involved in saying who websites are, and these can be betrayed or compromised
this happened and certificates were issued by iranian intelligence for Skype, google and microsoft
these certificates were in use for over a month; namecoin could make this more secure
namecoin repurposes bitcoin to be a naming system rather than a currency
if attacks forge name records in namecoin it's equivalent to stealing bitcoin
there may be cool service we want to use, but our data is in these silos
we built a multiuser client for Solid that TimBL and his team have built
I was listening to a presentation by David Clark, and he did a thought exercise:
in the time between launching your browser and going to Facebook, how many parties are you blindly trusting?
that list turns out to be more than 10 - something is wrong when you trust people you don't know exist
with blockstack, to decentralize the web you need to get rid of these trust points
blockstack is a production system, it's been running for 2 years
I'm the co-author of TLS, and that was supposed to fix centralization, then, but centralization creeps back in
we've had a lot of 1st world problems here, but I'm worried about the underpriviiliged access to the net too
I have an MBA student teaching women in Afghanistan how to be entrepreneurial, and there is real risk there
you've already shown us that identity is more important then stopping your little brother impersonating you on FB
with TLS there was a lot of competition - VISA and MasterCard had separate efforts; a patent-holder had another 1
Microsoft had a spec that was just different enough for them to control it
one of the reasons TLS was adopted was that it let you choose which certificate authorities to trust
this has changed as we have fewer browsers, and they make it harder to manage CA's - they creep back in
we have cycles from centralization to decentralization - how do we change that and lock it open?
decoupling of data and application is key, as that is a liberation of services
when you look at a platform like FB you have a profile and ID tied into one block, which prevents other services
what does solid provide here then?
the data is not tied to the app - you can change the app any time
the pendulum swings back and forth between centralization and decentralization
there were a lot of peer to peer companies in the early 2000s and only a few survive Tor, torrents
there is a graveyard of peer to peer systems
what has changed is that bitcoin gives a neutral playing field with a Billion $ bounty if you hack it
there have been lots of attempts to replace the cert authority in TLS
most of them just shuffle around the trust authorities, they don't remove 3rd party trust
bitcoin gives a way of removing 3rd party trust entirely
how did you move from the general inclination to throw stuff on the blockchain to what we have today?
this is the CTO of amazon, this is his account on blockstack - he owns werner.id
to change a single character, you would need to take down bitcoin, which is a couple of billion dollars
theoretically you might think things are possible - I believe in building production systems
we don't want to tie ourselves to any particular blockchain; we design it so it works if bitcoin fails
you need to pay attention to performance and user interface from the beginning
can you name a decentralized system with a truly great UX?
reputation is very hard - there are lots of ways that people game reputation systems
Cory Doctorow – "How Stupid Laws and Benevolent Dictators can Ruin the Decentralized Web, too"
this is like being back at the O'Reilly P2P conference in 1999 [tç: 2001]
I'm here to give dieting advice. When you go on a diet: throw away all your oreos
if you're serious about not eating Oreos, your best bet is not to have a bag to eat
use your willpower now to get rid of the Oreos so you don't eat them when your will is weak
it's called a Ulysses pact, after his journey with the sirens - make sure he can't fail due to later weak will
it's very tempting to centralize things - there are lots of advantages to it
the way the web got centralized today was because people like you made compromises
we only sense relative difference, not absolute ones, so each compromise is incremental
before you know it you're suing to copyright your code or putting in back doors fro the NSA
I am not better than the people who made those compromises: you are not better than them
we hyperbolically discount future costs comapred to present benefits
if you don't want to eat oreos in the middle of the night, make it harder so you need your car to buy them
the way to avoid making compromises in future is to take them off the table in the present
Brewster said this when we could use google docs' JS libraries to build p2p
the GPL worked at locking things open by not being able to compromise
eventually it became absurd that you would pay licences to spin up VMs, so GPL'd linux won
sometimes companies want to commoditize their rivals product
systems that work well but fail badly are doomed to die in flames - the GPL fails well
Microsoft was right to be freaked out by the GPL - their programmerrs had the instinct for openness
the GPL worked so well, no-one ever sued them
DRM Digital Rights Management tries to take over your computer to enforce you not seeing things
DRM is legally enforced anti-tampering - it's being used by Ford to force you to buy their parts
Missing: Panel – As we build a Decentralized Web, what values do we want written in the code?
Missing: Panel – Security in a World of Black Hats
Panel – Moonshot Challenge – What could you do to Decentralize Scientific Journal Articles?
the problem with the art world is provenance, so we built ascribe to track provenance on the blcokchain
with BigChain DB we have transaction db under the hood - which you an install and track
so we wanted to bring bigchaindb to the web which is IPdb - a public database for the planet
shared global compute infrastructure is emerging, but this is the database for it
we talked about hashing content, but media has multiple formats, what if we could has a concept not the content
we see a lot of media sharing that loses attribution - we have decentralised database and content ID
we use image recognition technology like google research image search to resolve to the same identifier for images
we walk the chain of metadata to get a complete record of the work
we can surface metadata conflicts and correct attribution
content addressing provides easy redundancy, while retaining attribution so the fans can pay the creator
Dat is backed by the Sloan foundation and the Knight foundation
Interledger is a financial protocol handler
here's a wallet - we have bitcoin wallets, cash wallets and more. These examples are fake money, in 2 ledgers
I'm going to send $10 so they get 8 euros in their system - we route between ledgers
we took webtorrent and added interledger. I download a file, and pay the seeder as I go
if you can send money to people without thinking about you, you can send payment money to lots of different apps
InterLedger is inspired by IP - it sends packets of money across different ledgers
we're looking for people who are interested in compensation and getting artists paid
we wanted to read Vint's article in Science "who would pay for public access to scientific data?" but it's paywalled
to read Vint's article would cost $30 for a one day Science pass
I talk to scientists about how they publish their data - a lot of them use Dropbox to keep it simple
if we're doing a decntralised tool for scientists we need to do a lot of UX research
if you look at dat.land there is a hash which is a public key you can drag docuements info
this paper is a PDF and LaTeX files, and a CSV, but git can't view big csv
I can drag more data into this, but I can create a checkpoint to share or reference in a paper
a tool called ScienceFair that lets me search for a cancer paper, it gets the paper from the decentralised web
usually the problem is usually not storage, but bandwidth
by using decntralized web you get a lower bandwidth bill
articles have copyright issues -do the data?
we leave copyright to others; we reference data wherever it is by what it is
have a look at dat-data.com
if you're successful how will it impact the world?
open knowledge for everyone in the world
share global data that's always in sync for the whole planet
making the decentralized web possible through shared data
Closing Remarks: What happens next?
This is Great [arm wave]
we have got the idea of changing the web protocols to be decentralized
we have implementations that run in browsers
with zeronet, IPFS, dat and webtorrent
can we make a rock band actually make some money by selling directly to the web?
how do we reinvent scientific publishing in a way that isn't gummed up by the industry
what do we need to do now? VC Funding? Foundations? is that's what's needed?
do we need a sequence of conferences with milestones? what if there are awards
I have been amazed by the people who came together - tomorrow there will be more sessions for implementors