Swarm conference
==

Brain dump taken during https://ethersphere.github.io/swarm-summit-2018/, unmaintained

## Questions

* what are the most recent specs / docs / etc?
> https://github.com/ethersphere/go-ethereum/ swarm-network-rewrite branch, this will be merged to a `master` somewhere (either their own fork or the usual upstream geth)
> https://swarm-guide.readthedocs.io/en/latest/ - latest docs
* are there alternative implementations?
> no - there's less benefit since there's no consensus involved - swarm should work with any rpc-supporting client
* is there a projected timeline for light clients? - hutch
> there's work to do and Status could become involved, else Q2 2019 when PoC4 is done
* Is Swarm-PSS going to be more suitable then Whisper as a messaging protocol? I think Whisper is better for anonymity but it would be nice to know more about it. - Andrea (@gravityblast)
> it's a strong candidate for sure - for many reasons, including efficiency, flexible routing.. there are open questions too though like immutability (or it being expensive to do stuff like a chat log) and permanence (no way of removing data), will need to dig deeper - mainframe has an implementation of what it could look like

## Log

### Viktor
* partnerships to to provide database, advanced features
* about pss: messaging system with sliding scale of anonymity, ground service for many applications
* lower barrier entry for world computer services
* stake-based service-level guarantees
* poc3 (q2 2018): inclusion proofs - encryption, mutable resource updates, pss etc - running on open/staging cluster - recommended for hacking
* poc4 (q1 2019): mostly feature-complete

### Mainframe, Carl:
* set up swarm chat channel, needs aws instance,
* [set up sos decentralized chat](http://sos18.mainframe.com) - requires something aws..

### Mainframe. Doug:
* how to contribute to an incumbent, start small contributions that
add value, eventually take over.. OSS business model different, we're
building infrastructure, swarm itself is infrastructure for other,
projects, help where you can

### simulation framework:
* Anton Evangelatov, devops from EF
* p2p/simuation package
* running simulations with different nodes, either in-process, child process, docker
* works with travis
* nodes are created, then sim framework generates events
* nodes can have different protocols
* used in chunk streams, pss, network discovery tests
* p2psim command line client for http api
* devp2p protocol simulation very easy
* visualisation component present in ethersphere repo

### testing chunk distribution:
* Janoš Guljaš, swarm team
* using simulation framework to test message
exchange
* streaming: history vs live chunk exchange
* more complex messaging scenarios than req-resp
* have static tests for: node sync, live/history
streaming
* also has dynamic simulation with (dis)appearing nodes, higher
level functionality
* highest level of tests: swarm cluster with docker/terraform
* there's a open test cluster: https://open.swarm-gateways.net/

### mainframe, distributed:
* what's needed to reach web3?
* addressing (break link between ip/location), payments,
communications, databases, computing services (offline, jobs,
heavy), incentivization (behavioral crypto-economics)
* building an SDK - a platform for accessing technologies
underneath, devoting resources to many projects underneath
allowing pluggable access to similar systems (ie swarm+ipfs),
essentially doing dapp framework
* support dapp development, develop a few own quality reference
dapps, focus on good documentation / simplicity for developers,
delighting developers is a key to success, **developer is
main customer**
* roadmap - sometime this year - devtools, functional service
layers
* 20ish ppl, hiring engineers

### ethergit:
* sofware innovation centre, based in madrid
* focus on video, heavy on engineers from content distribution
/ akamai
* storage on swarm, put governance-necessary information to
blockchain smart contract
* when things happen on repo, there are smart contract functions
called
* web interface available to browse ethergit repo, create repo
with "standard" governance models
* mirrors go-ethereum repo on swarm (cool!)
* planning to add issue tracking, sample contracts, bounties
* store issues/metadata in swarm with
* goal: get rid of github!
* haha: could make a contract that requires deposit to push,
no tests? penalty.. buggy code? penalty.
* planning to tackle repos/distribution as well (a la npm)

### dinner:
* developers needed, potentially to work on light clients,
erasure coding

### api:
* preferred way to use swarm is through http API
* several protocols being released: pss, bzz
* documentation revamped
* adding encryption?

### streem/artis:
* subscription payments
* balance flow rates in subscription graph
* basic digital income?
* discussion: sub vs micropayment? automate the latter makes it
easier, sub allows financial planning / less cognitive overload
* similar thinking / project: colony (Aron Fischer)

### reencryption
* https://www.nucypher.com/

### livepeer:
* transcoding to different codecs/bitrates intensive
* livepeer creates transcoding market
* tricky on privacy - transcoder needs data
* tricky on validation - did the transcoder do the right thing?
* truebit - step-by-step proofs of work that can be verified -
https://github.com/livepeer/verification-truebit

### whisper questions:
* do we support forward secrecy?
* do we have a custom keyboard?
* in general - questions were focused on secure messaging aspects

### psst:
* pss sockets - make pss familiar to socket devs
* tcp-over-pss?
* solve reliability issues by doing a TCP-like connection
over PSS
* inspiration from TCP over unreliable channel, messaging
windows etc

### Swarm on Red Hat OpenShift, Nikola Jokic
* how to run blockchain deployment in production
* manage upgrades, development, topologies
* template for deploying: [parity-openshift](https://github.com/DAPowerPlay/parity-openshift) - deploys parity nodes with spefific features
* comes with load-balancing, etc
* swarm under radar, hard to find code - missing project /
community management
* kubernetes - think cattle, not pets - hard to manage nodes
with identity

### Datafund
* social as well as technical
* ethics matter
* GPDR entering the picture, legal implications?
* economic incentives only lead to toxic environment
* personal data has value - manage it as assets in a fund
* open question: how to apply ethics to choice in technology
implementation

### Swarm encryption
* CTR mode with two sha rounds
* Allows partial disclosure by revealing output of first round
* Existential forgery vs XXX forgery
* References are hash + decryption key
* Encryption/Decryption happens in gateway, must trust it!
* Identity management in swarm could be managed by trezor/ledger

### Ontologies for Structured Data in Swarm
* LOD refine - tool for cleaning messy data