Ethereum Community Conference 03/18
A few notes I’ve taken during the event. For those who couldn’t make it.
Yet another great conference last week in Paris. The Ethereum Community Conference was mostly adressed to devs and the current challenges they’re facing. A lot was said about scaling in all its forms.
Quite a few familiar faces here including Trent McConaghy, Stefan George, Vlad Zamfir, Karl Floersch and a special appearance by Vitalik Buterin no less.
Hat off to Trent for his amazing presentation on Token Engineering and to Karl Floersch for an energizing Saturday morning 9am presentation on Plasma & Sharding.
What follows are my non-technical notes of the event. For a more advanced coverage feel free to check the recordings.
Simon Polrot — Identifying & Managing legal risks
Where Simon gave us a few practical advice on risk management strategies.
Everything is risky. There is:
- Tech risk.
- Financial risk (volatility, liquidity etc.)
- Legal & tax risk. In the blockchain world, everything can be sued and fined.
A lot of the regulatory risk analysis has been conducted around US law.
The risk is actually global. Regulators are starting to freak out. Reinforced by the crypto boom and recent price drops.
All these regulator activity leads to more uncertainty and consequently more risk.
Tentative legal risk scale (from low to high risk):
1 (low)- dApp is not dealing with value (infrastructure, proof of something)
2 (medium)- dApp is dealing with simple exchange of value
3 (high) ICOs, token sales & DAOs (dealing with large amounts of values)
4 (high) Financial dApps (full exchanges, derivatives products and gambling)
3 and 4 you should hire a lawyer very soon!
The potentially applicable regulations that teams and investors should keep track of:
- tort law
- tax law
- consumer protection rules
- data protection rules
- corp law
- KYC/AML
- Accredited investors
- Sectoral reg (relevant to the application of the blockchain)
The impact on projects could be:
- Difficulty in putting the structure in place (bank account)
- Large tax liability
- Class actions
The main points that will significantly weaken a project:
- Collecting large amounts of money from unknown people
- Operating without legal structure
- Not paying any taxes anywhere
- Deceiving people in intents (99% of the time this is how it starts)
Ways to manage legal risk:
- Become familiar with the legal/tax/blockchain rules that might be applicable.
- Incorporate a legal structure
- Hire an advisor
- Do not mess with partners, clients and investors — they’re the most likely to sue you.
How to operationalize this:
- Get a local advisor (where the legal entity is setup)
- Meet with more than one lawyer, pay attention to the most annoying advices
- Do an honest risk assessment: what are the risks of not doing such and such? What are the risks a project is taking?
- pay fines
- pay taxes
- be in litigation for x years etc.
Risky assumptions
- Illusion that a project faces no legal risks because dApps are decentralized
- are you making any money from the apps?
- Belief that a project need only to take into account US regulation
- are you making business outside the US?
- Belief that there is no risk because a project is located in a tax heaven
- be careful about “artificial arrangement” and “substance”
Lefteris Karapetsas — Rotkehlchen
Where Lefteris looks into the issue of filing taxes on crypto gains.
Many transactions are happening.
On different entities and through different jurisdictions.
Tools available for filings:
- bitcoin-tax
- cointaxes
- cointracking
- and more..
The problem with these tools is that users are handing their own data.
Rotkehlchen is an open source solution that offers data encryption. Starting with Germany.
The goal of Rotkehlchen is to help users get more responsible and take ownership.
A premium model would offer:
- dedicated support
- priority in feature requests
- data synch between devices
- priority usage of ETH nodes for faster responses
Patrick McCorry — Smart Contracts for bribing miners
Where we explore the fascinating field of how to break consensus forming.
What is a bribery attack:
51% of network hashrate is bought or rented from existing mining hardware
Two ways of renting:
- in-band
- out-of-band
Is bribery a realistic scenario?
Miners have long term interest in securing a network.
But that could not be the case in the short term.
Take BCC vs BTC for exemple where miners can become opportunistic in the short term.
Is there evidence of bribery? There have been allegations but no evidence.
In any cases, if there is bribery, you need a wealthy adversary.
Vitalik once asked if 51% attacks risks were acceptable to a community. +50% of respondents said that a 51% attack was an acceptable risk every 10yrs.
Identified bribery-style attacks:
- whale transactions
- script puzzles: by solving a puzzle a miner gets a price. A process to divert hash rate.
- proof of stale block: When winning a block a miner can publish it or keep for himself. If he decides for the latter, he can sell it in a smart contract to a malicious agent.
- censorship con: the briber openly want to disrupt a chain. An answer to this form of bribery is to use an “uncle block reward policy”. All uncle receive a partial reward depending on when the block is accepted. The longer before acceptance, the lower the reward. For it to work a briber needs at least 25% of the hashrate.
- History revision con: briber wants to reverse a transaction. A briber would need 3 accounts to execute: (1) one to create and fund the contract, (2) coins to be double spent (3) an account to receive double-spend. If all steps are satisfied, the bribers wins. This attack is all or nothing.
- goldfinger con: a wealthy adversary wants to reduce the utility of a crypto. Specially valuable in an environment with many competing cryptocurrencies.
Patrick finished his talk asking how these attacks would unfold in a PoS setting.
Trent McConaghy — Ocean Protocol, Towards a practice of token engineering
Why jellyfishes all over your slides Trent? Because jellyfishes are cool. If AI was an animal, then it would be a jellyfish.
AI research in the 90s was about fixed size data sets and do what you can to come up with some fancy research algorithms in order to improve accuracy.
After the early 2000 researchers decided not to focus on algorithms but rather on data set sizes. Increasing them by 1000x.
As more data was added, accuracy improved. From 65% to 95–97% in some instances.
Which was quite embarrassing for people with advanced PhD in algorithms.
That’s when Google decided to go buy access to all the data it could. In essence, it became a data company more then an AI one.
The default incentive today is to hoard data in closed silos.
“Show me the incentives and I will show you the outcome” Charlie Munger
Blockchain are incentive machine. They get people to do stuff. Ocean’s goal is to change the existing incentives to silo data.
Tokenized ecosystems are like evolutionary algorithms (survival of the fittest algorithms).
A block reward is like a fitness function to minimize an error. Agents are the miners, actors etc. And there is even a system clock. Evolutionary systems can’t control besides incentivizing and killing.
If these things are so similar, let’s approach token design like optimization design.
What do optimization design look like?
- formulate the problem: objective and constraints
- find an existing solver
- only if needed.. design a solver
! if the solver doesn’t converge you can change the constraint.
(1) Formulate the problem:
Who are the stakeholders, what do they want and what are the possible attack vectors? - because each of these attack vector are going to turn into a constraints.
(2) Try existing patterns/crypto economic primitives. There are various patterns for curation, proof of compute etc.
A closer look at Curation:
TCR (binary models), Stake Machines (status models), Continuous-valued membership (bonding curve), Hierarchical (using label), Non-fungible token.
A closer look at Proof of Compute:
With a root split between data (proof of replication, proof of space time, proof of data availability, proof of service receipt) and computation (zero knowledge proofs, interactive proofs, secure enclaves).
Let’s look into bitcoin:
Objective function: maximize security of network.
Where security is expressed in hash rate which is compute power which is electricity.
How well did it work? Bitcoin is on track to use more power then all of USA in 2019.
Let’s look into Ocean:
Objective: maximize the supply of relevant data.
Where token reward happen when data is been made available and curated.
Patterns: using TCR for its actors. In some instances, the Ocean team had to come up with its own patterns such as Proof of curation.
Raine — The culture of crypto investing
Where Raine asked us investors to remember what drove us first to the space. Maybe is was through financial interest with some initial knowledge.
The entry point in crypto investing is always an exchange. What do people that on-boarding for the first time see socially? Not much. Maybe a chat box. Also called a “troll box”.
What are characteristics of offline communities that creates welcoming environments?
Offline communities might have “open house” events. Places where people can feel comfortable.
If we want wider adoption, Raine advocates for a progressive on-boarding.
How could we incentives more inclusive community behaviors she asks? If people get push back during their initial experience, they don’t come back.
The thing about cryptos is that it brings individuals together with similar interests and different backgrounds.
Inclusiveness is about who come back and who doesn’t come back.
Once the on-boarding phase is completing, another key offline protocol of welcoming communities is the presence of moderators.
Moderators are often long time community members. These moderators are leaders. Which in turn asks the question of our relation with leadership. A difficult discussion in a community built around decentralization.
There is a strong difference between management and leadership. Code might be law but it shouldn’t be conflated with leadership, it’s management.
The thing is you can’t fit the whole social system on a blockchain and we need to spend more time thinking about it.
Bridging the Dapp — Scaling now with Parity Bridge
Where we were offered an in depth intro to Parity bridges.
What problems could be solved?
- ethereum still doesn’t scale — though it’s been live since 2015 and it’s not likely to scale at least for another 12 months.
- A lot of good projects have been in the pipe for 2 years and are in pre-deployement phase, waiting for scaling to happen.
- Giveth, Colony, Gnosis are just waiting because cost of transaction makes not viable.
There are 3 options for these projects:
- wait some more
- deploy on testnet — which doesn’t allow for real value exchange
- explore work arounds on the state channels front
Parity bridge has two parts: Home and Bridge
A bridge consists of:
- software that connects nodes
- EVM smart contracts
- Authorities
How does that play in terms of security?
- Authorities have to be trustworthy & can be equal to foreign chain security.
- Relays require sign off by a majority of authorities
How Polkadot help to mitigate scaling issue?
- By offloading transactions to another network with more throughput. Currently Polkadot could offer 10x-100x better throughput then mainnet.
- By deploying smart contracts on each network. One part of a dApp can live on the mainnet while most functionalities can live on less expensive chains.
The current state of Parity:
Users deposit Ether into the Homebridge contract on home and get the same amount of ERC20 token on foreign. If users want to withdraw, a similar process happen.
Usefulness of Polkadot:
Useful where more than 2 transactions occur between deposit and withdrawal.
Security considerations: Once you offload to another network, the security is that of the other network.
Vitalik Buterin — Scaling for Plasma
Where Vitalik introduced a new scaling solution for Plasma.
The project has been lead by Vitalik and Lightening creator Joseph Boon.
Plasma is an attempt at scaling ethereum by creating a layer of smart contracts that interact with mainnet.
While the current prototype requires the user to download and validate each smart contract in the Plasma system, the newer version would only focus on a handful.
Users would be able to generate plasma coins by sending a deposit to a contract and simply track the tokens they have created in the system.
Verification only happens at the relevant index to the user (tracking any plasma coins that a user might own or care about).
The solution has yet to be tested.
The impact on exchanges could be huge as it restrict the need to providing order book functionality and insure losses through Plasma contracts.
Matan Field, Adam Levi — The DAO Stack
A firm is orientable but not sub-scalable
Once you have the setup you can orient it to any purpose. But it scales sub-linearly.
The free markets are the opposite. They are super scalable but not orientable.
Enters the DAO Stack: a new form of human association that is orientable and super scalable.
The DAO stack combines firm and market features.
Use cases:
- collaboration among developpers
- asset management
- curation (the way Google is a curation of websites).
- marketplace
The DAO stack is powered by Ethereum.
The DAOscape
There is no way to build a complex system from end to end. However, you can design its elements and then allow to happen.
The DAO “letters” are these elements. Then need to design interactions between these elements. And finally allow for supers scalability through governance.
On governance systems, they:
- can be divided in the do’s and don’t.
Do’s: ex. if 50% approves, then x happens.
Don’ts: ex. limits like global constraints.
- have actors (token, reputation, avatars). The face of the agency
- have controllers
- have universality
On holographic consensus: allowing small groups of people making decision on behalf of the larger majority. Only when these decisions are predicted to be in line are they projected to the majority.
Aron Fischer — Colony
Where Aron focused on permissiveness in the Colony Protocol — A platform to organize collaboration in projects.
Today collective governance mechanisms are missing within our toolbox.
If you want to use the blockchain to achieve the goals of collaboration you need to solve:
- speed
- cost
The general principle is don’t go to the blockchain unless you have to.
In Colony, you only go on-chain if there is a dispute.
On voting: people love to talk about how to implement it on-chain.
For many people voting and collective governance are two sides of the same coin.
The problem with voting:
- it’s slow
- it’s expensive (cognitively)
The core idea is to vote on-chain only if there is a dispute.
The smallest part of a project is a “task”. There is an administrator, an evaluator and a bounty to the worker.
- worker: if the work is accepted, then the worker is paid.. after three days. Every security involves delay.
On balance in designing Colony:
- things need to be easy
- but it should be hard to cheat
Dispute only happen after a rejection to a task and a objection to the rejection.
Disputes can be escalated to a higher domain but at a higher cost.
Reputation kicks in to solve disputes.
Reputation is earned by doing good and lost by doing the opposite.
There is significant downside if a dispute is lost by a landslide (90%).
On the current obstacles that Colony is facing:
Hard to assign what a task is worth up front. There needs to be mechanism to be able to adjust bounties.
Karl Floersch — Ethereum Scaling: Plasma
Minimum viable plasma:
About scalability while maintaining decentralization.
A plasma chain that is as secure as the root chain.
That’s the guiding rule.
Sharding: currently scaling is constrained by computation resources on the blockchain. We could very well replace all computers with a central super computer.
Sharding creates a super computer by combining all computers in the network.
Phase 0 sharding doesn’t require a hard fork. But only offers data availability not state transitions.
Phase 1 sharding upgrades. Offering account abstraction, ewasm etc.
Users pay block proposers that will in turn pay block validators.
Validators are randomly sampled to publish collations. Randomly sampled to avoid having them identify one shard.
Adrian Brink — Scaling with Cosmos, Tendermint and Plasma
Where Adrian started by distinguishing between two types of consensus:
Nakamoto Consensus
In order to be safe we need long network latencies. Given long enough if there is weak consensus, then you’ll start seeing divergence.
BFT Consensus
Hasn’t been applied to blockchain much. But some good things, among them shorter confirmation time.
The Tendermint Consensus is a BFT consensus.
Its properties:
- instant finality: as soon as sent, it’s final.
- efficient light-client proofs: in Nakamoto you need to keep up with header to check. In BFT you need to update.
- safety in asynchronous networks. Block times can be reduced to ~1sec.
The limiting factor in Tendermint becomes the state machine. Parity has great room for improvement though most projects shouldn’t be built on the EVM (good for contracts maybe not so much for DEXs).
The limiting factor in the distributed ecosystem becomes the limited scalability of some blockchains vs others.
Side thought: State channels!
Vlad Zamfir — Governance
Nodes might be:
- consensus forming (miners)
- service providers (exchanges block explorers)
- infra providers (relay nodes)
- personal authenticators
Software repos can be governed:
- for profit
- by an open source community
- by a standard body
Blockchain will have different ways of governing their nodes and repos.
Governance 101
Governance is a decision making process that impacts the governed resources.
Actors that are upstream the governance process are participants and actors downstream are stakeholders.
Governance can be understood as an coordination game.
Coordination is hard because of imperfect information. Ways to solve that:
- focal points
- expectations
- common knowledge
Governance norms are safely reinforceable expectations of the behaviors of participants. Governance norms structure coordination mechanism.
Using a governance process relies on legitimacy. Legitimacy doesn’t mean that is good. It doesn’t entail consent or fairness. It’s shared adoption.
Apathy is when participants renounce participation.
Ways to replace a decision process
- revolution: abandonment of a governance process
- forking: copying the governed ressources
Governance is the process of establishing and maintaining the legitimacy of decision making processes.
