Decentralized Web Summit 2016-06-07

we had lots of requests for press, but @qhardy of the NYT is hear and our incredible journalist Dn Gillmor

Kevin marks is live tweeting - follow him

if there is something you don't want tweeted, warn beforehand so I haven't already tweeted it

tomorrow we will live stream the talks and archive them at high res

Friends I bid you bright good morning

we are about to move into a phase where your laptop is less useful

put your device to the side and rise

Breakout Group report back

1: no reliance on politcal stability disagree on who can restrict others and on accountability

1: there is a tension between equal access and visbility

1: we want to build a standard way to address content and self-soverign identity that is cryptographic

1: we want to build a peer to peer encrypted channel without a central authority

1: hard to avoid regulatory capture

2: agreed on talking bout governace, interop, extensibility and cryto

2: there is alway some form of governance so we need to talk about it

2: interop - define minimum standards or building blocks not a monolith

2: community standards of censorship resistance and dealing with harrassment and abuse

2: lost fo discgreements on enryption and signing

3: agree that decentralization is a means not an ends

3: division of control baked into the infrastructure to resits capture and enclosre

3: emphasise the idea of the end user able to interface wiht the decenntraised system not a center

3: disagree on whether kinds of tech are bad or good or factors

3: complicating factors around access to kinds of data

3: how to prevent recentralisation and an endless loop so we'r not back in 15 more years

4: agreed on no central authorities -aspiration that files are content adressable

4: disagree on need to replace ICANN especially multi-state

4: idea: court institutional support for decentralised web

5: agreed we should be working systemic decntralisation on top of technical requirements

5: decentral governance is necessary to rpesevre decentral tech

5: content accessibility is import but has lots of sub-parts - rootless identifiers

5: not require http but map to multiple protocols

5: security requirements baked in on day 1; privacy baked in

6: agreed on openess of access and public data stored so everyone can access it

6: some disgreemnet about free of charge to the public access

6: disagreement on the idea of filtering and whether people are putting their data on the networks

6: and disagree on how you define sub-groups - inclusively or exclusively inherently

6: the hard problem is how you define the boundaries of groups

Just before this started a young guy showed up and wanted to show Van Jacobson his networking stack

If we can decentralize the internet, do we need to worry about decentralising the web?

we want to do both. There are incredible things happening in the dentralized web that will change the world

the web chanegd the world, but it was build on IP which assumed telephony point to point connectsions

you may find that what you want to achieve with a decentralised version has problems with the IP basis

in Named Data Network a packet has name duRI not an IP address. How does that work?

the web chanegd our model of how the world works - for most people the internet is irrelevant except for the web

wiht the web you give a name and get informtion back - it is a transctional model and we shoudl not build on that

rather than building on top of the web, we should push that transactional model down the layers

when you go to archive.org you run a stack of protocols, but as a user I think of it as a data hierachy

if your view of archive.org is as a set of machines, you need to wrap them in certificates to check

in the NDN model you announce the name, and use that to look up. You can't trust where you got it from

you have to get the trust from the information itself, not form where you obtained it

the basic structure gives a packet the url of where you got it from and a signature that asserts it is an answer

any security model that we are capable of building today needs certificates and addresses

when the data is not reliant on the network path you got it from but can be from anywhere, it works in new ways

what you are asking for is more important that the path back to you, so it can be asked anonymously too

the current abstraction means the router can't know if the endpoints have got what they needed. we can fix this

because the information is write once it is all cacheable

why is it better to do this at the lower level and not on top of IP?

we have a lot of structural problems with IP -

with about 2Bn users of the 6Bn world population it is hard to add the next people

the IP routing model encourages monopolies to form - it is hard to deploy in rural areas

one of the problems that Vint and the IAG discussed at length was how to get from phone to new protocols

you need a strategy to get from where you are to where you want to be - IP works as an overlay on anything

so you can run NDN over IP, but you could also run IP over NDN

how do you find the performance difference between NDN vs BDN/IP?

an IP identifier is very restricted; NDN is just a hierarchical structure, not special

the extra cost with NDN is the security as producer has to sign every packet, which takes more work

the sender is trying to prove to the receiver that they should receive that information

a lot of our specs are open source and open standard for adoption. Is NDN open source?

the NDN software base is all GPLv3 and is on github see named-data.net

what shell do you use?

I'm a dinosaur so it flips between bash and tcsh

can you give a 2 minute recap?

I worked with David Chung after dropping out of college

I then worked on PGP for a few years then moved to SF

I then came to Sf and met Bram Cohen and went to Mojo Nation

Bram said 'fuck this, it's not working' and went off and made bittorrent by throwing away 5 of 6 features

biran eno calls it being the first person to not do something people hadn't not done before

the I worked on zRTV for voice on the internet

and then I met Brian and worked on Tahoe, where we threw out different things than Bittorrent

now I'm working on zcash which is a distributed currency that has privacy built in

with bittorrent the more popular a file gets the more it is shared, with businesses that should work with money

as hackers we think of DHT etc as tools, but the most powerful tool is the feedback loop of business allocation

firms are vulnerable when they are a little successful when you are ressponible for your friends' pay

the second way is when the companies grow and are under threat from specialised shareholders taking control

big businesses are harmful; they become coercive.Can you stop them getting into the monopoly stage?

if you are trying to build good things you have to use this most powerful tool and know about its effects

this reminds me of the p2p conference in 1999 - I don't want it to be this long again

I used to use the line that the DCMA is so old people still though VR was a good idea. I can't say that anymore

what reignited lot fo people's excitement about decntralisaton was the success of bitcoin

you talk about how crypto can separate content from hosting and control

I reject "decentralize all the things" My goal is to uphold human rights

with math you can separate who you rely on for communication for who you may be spying on your comms

what have we learned from the first era?

I don't know

in the early day we made some Uber-ish bets on copyright law giving way

we had a de facto change where lots got normalised, but no de jure change so you can be sued for all you have

there was a technological determinism that judges would come round. are we still naive?

what we found that companies that get big enough can change things see Uber- why that works I'm nto sure

the taxi companies are local whereas uber is global, thta makes a difference

are market allocation and peer production in competition, and which is a better mechanism?

I don't like open loop answers where it's not clear how things are going to work

bitcoins tit fot tat wasn't money, but ti wasn't love either

do likes count?

my answer to CHelsea's question si that the monetary loops have high stability

(david reed) if the monetary loops are way too powerful, what happens if they're not linear like bittorrent

(david reed) how do to you prevent facebook from owning everything though network graphs?

when designing technologies that are designed to have social consequences, we have to think about this

the worst failure model we have at the moment is things not taking off; we haven't dealt with the opposite much yet

what structures do we need to achieve that thrust adn takeoff?

I really like sharing of knowledge and science - it's hard, but there is success there

I do like open source sharing which is almost like science, and it is hard too

someone reviewed employment data - one key thing was the lack of noncompetes in SV beats out Boston

also don't sign patents for your work unless they go into a non-agression pool.

Tamas Kocsis coded up zeronet on his own in 3 weeks

we're going to demonstrate Tomas's project zeronet

zeronet is a decntralised web project that allows you to integrate web pages without a cental service

when you run the zeronet client on your computer and open it in the browser - you can see decentralised websites

you can see how many peers are serving the site, but not all of them at once

so to take down this site you'd need to take down all these 45 sites on the web

thats right - the data is resilient

to host your own blog you can clone a node and then share the link wiht your friends

if you edit it it updates for your friends

there is a tutorial showing you how to use this comments are also decentralized

there is also a forum where comments can be shared

you can also follow blog feeds or comment feeds

if you're not running the client can you see the site?

there are people running zeronet gateways that share out to the web

zeronet hashes are the same pki as bitcoin so I can send donations to each site

there is a 10MB limit so it is for text, not for images

you could use IPSF for larger files

I'm impressed with zeronet - you wrote all apps from scratch. Can others do so?

what is the latency for publishing?

it goes out straight to all peers