DEF: 2 Title: Equality Protocol. Author: @DemocracyEarth. Comments-Summary: No comments yet. Status: Active Type: Paper Created: 2020-01-08 License: MIT Replaces: 0
A probabilistic score for human identity over peer to peer networks.
Quis custodiet ipsos custodes?Juvenal, Satires 2nd century AD.
Blockchain networks opened the possibility of new institutional models built with open source code, able to resist censorship and scale participation globally. However, Bitcoin’s original White Paper description of “one-CPU-one-vote” shaped the industry to think governance centered around machines, not people. Although a fundamental right to privacy bent early blockchain design towards anonymity, the lack of a robust notion of unique identity renders its governance practices into plutocracies: membership is often defined by stake ownership, enabling large holders to swing votes according to their own preferences.
Legacy identity mechanisms verify humans implementing practices that require disclosing personal and private information to an identifier. Any centralized repository that monopolizes identity in a marketplace —either through Know Your Customer (KYC) practices with banks, political coercion by totalitarian parties or data mining as practiced by digital corporations, eventually gains Orwellian notoriety risking the legitimacy of social governance processes. The emergence of blockchains as decentralized networks also pose a risk as personal information will live on an immutable ledger without any possibility of modification.
We propose a credential mechanism based on a principle described by Edward Snowden during the 2019 Web3 Summit in Berlin: “It is important to differentiate between verifying an identity, and verifying the right to use a technology”. The present approach does so by circumventing the need to request any data point from individuals; instead it infers the probability of an address belonging to a human based on its interactions with Distributed Autonomous Organizations (DAOs) — i.e. entities capturing decisions that require governance input and consist of contracts expressing choices that cannot be automated away, thus entailing human entropy.
The Equality Protocol inverts the focus taken by traditional verification mechanisms: it assumes by default that all addresses on chain are replicas and entirely not human; having that as a starting ground, it computes metrics that point towards increased uniqueness, drawing from the intersection of DAOs in which any given address is a member, as well as their relative presence in the social graph of blockchain-based transactions. An individual may join different DAOs with different addresses, and obtain an aggregation of the scores by demonstrating control over the corresponding set of private keys. By doing this mechanism incentivizes a behaviour that is opposite of the web, in which you are forced to use the same email address over and over again leading to the totalitarian concept of a one-dimensional identity. A successful implementation of an equality oracle will prove hard for any living person to demonstrate she holds the private keys to a collection of public addresses that will add up to a maximum score of 1.00, signaling she is (probably) entirely human.
Finally, the Equality Protocol approach creates a meta protocol against which other identity protocols can measure their legitimacy, and then reputation flows from organizations to individuals and back. It addresses the question of "Who verifies the verifier?" by creating an intersubjective space that feeds from a subjective function with valid participants voting on the legitimacy of indexed DAOs using quadratic governance methods, and an objective function that applies a Gini Coefficient to measure the egalitarian allocation of economic rights within a DAO on the blockchain. The Equality Protocol is designed to enable the implementation of a social layer built on top of distributed ledgers that can deploy borderless democracies, Universal Basic Income mechanisms and credit scores, without the need to sacrifice privacy and using social markers that incentivize participation on the blockchain economy to earn rights.
Throughout the year of 2019, two phenomena in the blockchain ecosystem have fundamentally augmented the scope of metrics available for a probabilistic and multifactorial identity score: (i) the maturation of governance contracts known as DAOs (Decentralized Autonomous Organizations); (ii) the emergence of different Proof of Humanity protocols based on Turing Tests and Web-of-Trust schemes —each having clear trade-offs on the feasibility to establish a scalable long term solution to the problem.
DAOs — Decentralized Autonomous Organizations: The DAO acronym refers to a class of smart contracts devised to automate the execution of organizational governance and decision-making. This institutional modality emerged as a possibility due to the creation of Ethereum, a blockchain network that permits Turing-complete computations, leading to the growth of smart contract development. By deploying DAO contracts into the Ethereum blockchain, organizations allow their participants to maintain real-time control of funds, and vote on its subsequent allocation with governance rules that are formalized, automated and enforced by the conditions encoded into its smart contract.
While traditional smart contracts are aimed mostly at delivering automated outcomes as they perform financial functions; DAOs differ by structuring a set of decisions that cannot be automated away and instead require some form of conscious attention for their functioning. Being intended for governance, they imprint human entropy on chain, offering a promising path as a substrate for anti-Sybil solutions. Examples of DAO contracts with parameters applicable as inputs for the Equality oracle include MolochDAO, DAOstack, Kleros Courts and Aragon DAOs, as well the recent efforts to standardize contracts that enable legal compliance by nonprofit and political organizations led by PeepsDAO, and for-profit ventures, led by The LAO. It is a composable solution that can enable the trust present in any kind of institution to be plugged into it, and therefore it is not bound by the Ethereum ecosystem, and it's — still — fragile ability to establish trust. As an example, it can take as an input a DAO that mints a national NFT by using a zero knowledge proof that takes records from traditional, nation-state institutions as oracles. Other non-EVM protocols, such as the Idena Network, can be also be included by allowing token swaps.
WOT (Web of Trust) using TCRs (Token Curated Registries): Web of Trust schemes consist of identity certificates that can be digitally signed by other users who, by that act, endorse the association of their own public key with the person or entity listed in the certificate. Adding crypto economic incentives to a Web of Trust is the use of a Token Curated Registry where curators can benefit from proper membership validation. While that prevents the centralization of validators, the possibility of dishonest participants taking over the graph presents a challenge. Examples of projects utilizing this model include HumanityDAO and Kleros' upcoming Proof of Humanity video platform.
Machine Learning Resistant Turing Tests: A machine's perception threshold can be measured using Turing Tests: tasks designed to tell robots from humans apart. Therefore, a Proof of Humanity consensus requires new forms of the CAPTCHA test, demanding specific tests of human processing capability that cannot be matched by machine learning. Such an approach has been suggested by Vitalik Buterin in theorizing a solution to the Sybil Attack problem with "a proof-of-work algorithm dominated by human labor and not computers", and has recently been applied by the Idena Network, as it outlines a novel way of achieving Proof of Personhood by making the resolution of machine learning resistant proofs synchronous.
Idena implements a live event held over its entire network, where participants are required to simultaneously solve FLIP tests. A FLIP test consists of picking between two sets of the same images disposed in different order, where one makes sense as a story, and the other doesn't. Solving a FLIP requires a semantic interpretation of the relationship between objects, which is common sense for humans but difficult for computers: the average AI accuracy on solving a set of FLIPs is 69%, while Human accuracy is at 95%. Given that tests cannot be captured by existing machine learning, Idena successfully provides a Proof of Personhood. It is not a strictly anti-Sybil solution, since there is a probabilistic margin of error: although highly unlikely, a person with exceptional ability could solve more than one set of FLIPS within their allotted time and earn more than one credential within the network.
However, Idena demonstrates that "liveness" — a synchronous event — can play a critical role in Sybil prevention, as the time constraint prevents a single entity from solving more than a few sets of FLIPs. Such a method enables a modality of validation that involves no data point except that of proof of conscious cognitive ability. The tradeoff is a significantly high coordination cost to achieve recurrent, simultaneous solving of FLIPS: all nodes must continuously participate in the synchronous events, otherwise their identities expire.There is no mechanism to ensure that those who pass the FLIP test and are thus granted a credential continue to maintain control of their keys over time. This creates a scenario where people could pay Mechanical Turks to collect several genuine identities which could later be handed over to a single person.
As the blockchain ecosystem is converging on smart contracts standardizing the structuring of organizations and identities, Democracy Earth Foundation is constructing an algorithm able to output a probabilistic score for Proof of Humanity to all addresses interacting on chain, facilitating democratic governance and preserving privacy. The Equality Protocol is designed in the same way that algorithms like PageRank (developed by Google in 1997) were once created to make sense of the web. It serves as a proxy to estimate the chances of any given address corresponding to a single human participant within the network.
Equality Protocol works as an intersubjective consensus. Drawing from the social ontology of John Searle, Intersubjective Consensus, as the name indicates, takes identity under a lens of intersubjectivity. Searle proposes that contemporary society is permeated by invisible, but pervasive, "status functions", which are intersubjective entities that emerge from the interdependence of a two-level ontology of facts. The lower level concerns objective, brute facts — independent of human subjectivity — while the upper level concerns institutional facts that rely on some form of ‘collective intentionality'. Examples are money, property, governments, marriages, stock markets and cocktail parties.
Current blockchain-based identity certifications may be categorized as strictly objective: they consist of cryptographically signed statements that go no further than timestamping a hash with an encryption algorithm. The failure to provide an input for subjectivity accounts for a central challenge: the inability to address the question of "Who verifies the verifier?". Looking at identity outside of the scope of centralized or governmental verification, requires the construction of new status functions embedded in collective intentionality that can substitute for the shared subjectivity legitimizing former sources of validation.
The present approach is designed to create an intersubjective space able to address this need, accounting for measurements of both 'collective intentionalities' and 'brute facts' by combining a subjective function that provides legitimacy to the score based on Quadratic Voting, and an objective function that measures the Gini Coefficient of any DAO existing on the Ethereum blockchain. It will create a Democratic Index and assign a score to every Ethereum address relative to the intersection of DAOs in which it belongs as a member or its position in the social graph of blockchain-based transactions.
Democratic Index
Fig. 1 Every address that belongs to a DAO will be weighted by the Equality Protocol with a percentage of the Democratic Index obtained from the intersection of DAOs that constitute the identity of such address.
Fig. 2 In order to counter weight false positives on the Gini Coefficient, addresses that achieve a high score will be granted the right to vote on the personhood of members in other DAOs analyzed by Equality Protocol.
Fig. 3 The Gini coefficient for democracy ranges from 0 to 1, with 0 representing perfect totalitarianism and 1 representing perfect democracy. It computes a score that measures the share distribution by each segment of addresses belonging to a DAO.
Fig. 4 A Democratic Index is calculated for each DAO, as a function of their position in the Quadratic Voting rank and their Gini Coefficient.
This implementation implies that an initial arrangement needs to be made for the first cohort of applicants to be approved — after which, it can continue on its own. We plan to host an in-person validation event at a popular conference such as Devcon, where DAO members with a solid public reputation can be the first to be approved and will kickstart the Quadratic Voting component.
Described by author Glen Weyl in the book Radical Markets from 2018 and in the original paper published by him in 2012, Quadratic Voting is a mechanism that consists of granting each voter an equivalent amount of credit tokens, and allowing them to distribute it on a given set of choices, expressing the intensity of their preferences. However, each additional vote cast on a single choice costs an amount of credits that increases at a quadratic rate: 1 vote costs 1 credit, 2 costs 4, 3 costs 9, and so forth. This means that voters are able to express the intensity of their preference as much as they want to in support of a cause, but zealotry is penalized at an exponential rate: more votes being put in the same option quickly exhaust the voter's budget. By doing so, it provides a clear output on whether the intense preferences of minorities outweigh the weak preferences of the majority.
In April 2019, Democracy Earth Foundation partnered with RadicalxChange Foundation to implement the first Quadratic Vote in the history of the US Government, for the Legislature of the State of Colorado (globally only a precedent in Taiwan can be found). Results show how Quadratic Voting is particularly useful in ranking a long tail of items. Instead of resulting in a few items at the top, with no clear signal over the rest of the proposals, Quadratic Voting succeeded in capturing nuanced preferences over the entire list of proposals.
Equality DAO
Within Equality Protocol, we use Quadratic Voting to rank a long tail of DAOs from -1.00 to 1.00, where validated identities will provide their input on the different DAOs according to their ability to be Sybil-resistant, i.e. ensure that no one person controls more than one identifier within their respective domains. Manipulation attempts by an individual such as creating a false DAO with multiple Sybil addresses, and therefore generating a false positive in the Gini Coefficient (described below), are contemplated by the mechanism of allowing negative signaling in the Quadratic Vote, which can shoot a DAO's index score down. This can additionally be balanced by the requirement of a minimum of assets — valuable as money — per capita so it can make spamming the system costly.
Philosophy: Equality DAO is a radical, egalitarian democracy that governs human identification on Ethereum. It evaluates the personhood of members in other DAOs on Ethereum. That evaluation is an ongoing and never-ending process. After all, new DAOs may emerge at any time, the reputation of existing DAOs may change, and the subjective evaluation of DAOs by Equality DAO members may change in arbitrary ways. We have designed Equality DAO so that the evaluation of the other DAOs is as flexible and accurate as possible. In the following we explain the mechanisms and rules of this democratic process.
elections should be a real time ongoing never ending process.Santiago Siri, June 2019 via Twitter.
Membership: Equality DAO is an egalitarian democracy, i.e. each member has the equal number of voting rights. Membership can only be acquired by those Ethereum addresses that have a certain probabilistic score for human identity based on the Democracy Earth Intersubjective Consensus. All Ethereum addresses with an Equality score greater or equal to 0.5 have the right to join Equality DAO. When a current member‘s score falls below 0.5, it will automatically be kicked out of the DAO. Also, Equality DAO members may resign their membership voluntarily at any time. Nevertheless, given a sufficient Equality score they may reclaim their right to join again.
Eligible candidates: The Equality DAO election candidates are other DAOs in the Ethereum ecosystem. The admission of new DAOs as eligible candidates is handled by automatically adding DAOs of certain DAO standards. When the Equality DAO is summoned, all DAOs of the DAO standards Moloch Version 1 and Moloch Version 2 are included as eligible candidates. All DAOs to come in accordance with these smart contract standards will automatically become eligible candidates for the deliberative agenda of Equality DAO.
To add a new DAO standard to the eligible candidates, current Equality DAO members must achieve a petition with signatures from at least 50% of the members. Vice versa, a petition with signatures from at least 50% of the Equality DAO members may remove a DAO standard from the deliberative agenda.
(Quadratic) voting: Each Equality DAO member receives one voice credit per eligible candidate. In other words, one candidate is one voice credit in the democratic process. Voters may arbitrarily allocate their available voice credits across eligible candidates - for example, casting all credits on one candidate, or distributing the credits across a few or all candidates. Here comes the radical twist: in doing so, for each candidate respectively, the casted voice credits of a particular voter translate to effective votes at quadratically increasing costs. Voters may cast voice credits as pro or contra votes. In terms of voice credits, the minimum threshold to cast to a particular candidate is one.
Voters may at any time reverse or modify the allocation of their voice credits in accordance with the rules above. Thus, casting votes is never final and can be reversed at any time.
Following is an example with 200 eligible candidates. Therefore, voters had 200 voice credits at their disposal. Let us assume that a voter X casted 12 votes pro DAO A (144 voice credits) and approximately 7.5 votes pro DAO B (56 voice credits). If voter X now learned that DAO A was subject to a Sybil attack, he may update his voice credit allocation. For example, casting 10 contra votes to DAO A (100 voice credits) and 10 pro votes to DAO B (100 voice credits). Furthermore, if a 201st eligible candidate were added, each voter received one additional voice credits. Voter X now had, including the 200 voice credits already allocated, 201 voice credits in total at his disposal.
After all, voters may always revise and update their voice credit allocation across the eligible candidates. Voting is always on; there is no grace period. Thus, the chronological availability of candidates and information is completely irrelevant for voting in the Equality DAO, leaving no security loopholes for strategic agenda manipulation. Equality DAO voting is a real time ongoing never ending process.
Relation of votes to subjective inputs: The subjective input of Democracy Earth’s Equality Protocol rates DAOs from -1.00 to 1.00 where -1 indicates the lowest and 1 the highest possible score of human identification. A particular DAO can reach the maximum subjective input of 1.00 if and only if all Equality DAO members casted all their voice credits in favor of it. A particular DAO can reach the minimum subjective input of -1.00 if and only if all Equality DAO members casted all their voice credits against it. The following formula describes the translation from votes to the subjective input of a particular DAO. Let n be the “number of Equality DAO members”, c be the “number of total available voice credits per member”, and v be the “sum of effective votes pro and contra for a particular candidate”.
Let us assume, for example, that at a particular moment in time, Equality DAO had three members, each of them had four available voice credits, and that DAO A had two effective votes in favor. (See the corresponding calculation above.) Thus, the subjective input for DAO A would be 0.33 and further, a piece of the subjective input for the Equality of each of its members would be 0.33 as well.
Minimal Anti-collusion Infrastructure
Democracy Earth Foundation's co-Founder Santiago Siri joined the MolochDAO using selloutdao.com - a bribing contract written by Mariano Conti from MakerDAO during the ETHBerlin hackathon. The smart contract entailed that for 0.5 ETH (USD $75.00), a buyer could send a proposal to MolochDAO through Mariano's address using his pre-existing shares as a member, and then vote on it with his shares (but without his active involvement). It was an experiment on transparent, “automated bribery” aimed at testing the boundaries of blockchain governance. Santiago obtained 389 votes in favor - and zero against - his proposal to join MolochDAO from other members and Democracy Earth Foundation is now an active member of the guild.
This is one among many examples of collusion that can happen under a blockchain-based voting system. By generating a record of transactions, blockchains can facilitate bribery with smart contracts able to reward users able to demonstrate they have voted in a certain way through a publicly verifiable transaction. For the implementation of Equality Protocol, we consider Vitalik Buterin's proposal for a Minimal Anti-collusion Infrastructure as a path for maintaining on-chain privacy and ensuring that no such trusted record can be generated under the Quadratic Voting function of the Democratic Index.
Under such a protocol, a user could prove to have voted a certain way by pointing to the on-chain transaction and providing a zero-knowledge-proof that the transaction is the encrypted version of the data containing their vote. However, they have no way of proving that they did not send an earlier transaction that switched their key to some new key, thereby making their proof unreliable, since the new key could have been used to vote in a different way. While there are still possible vectors of attack (one could sell their private key), the Minimal Anti-collusion Infrastructure outlines a promising approach to address on-chain privacy for voting mechanisms.
The Gini Coefficient measures the distribution of income across percentiles of a population. It ranges from 0 to 1, where 0 represents perfect equality and 1 represents perfect inequality. As the work from Balaji Srinivasan on quantifying decentralization demonstrates, "centralization" and "inequality" have striking similarities; one can think of a non-uniform distribution of wealth as highly unequal, and a non-uniform distribution of power as highly undemocratic, or centralized. A similar calculation can be made to measure how democratic is a DAO.
Drawing from the two basic primitives involved in every DAO — shares and addresses — a Gini Coefficient for democracy is employed in Equality Protocol to measure how egalitarian is the distribution of decision-making power within its structure. This calculation can require a minimum of assets — valuable as money — per capita so it can make spamming the system costly. A graphic representation of the Gini Coefficient for democracy can be made through the Lorenz curve, showing voting power distribution by plotting the addresses percentile on the horizontal axis and the cumulative shares on the vertical axis. With 0 representing perfect totalitarianism and 1 representing a perfect democracy, the Equality Protocol will compute a score that measures the share distribution by each segment of addresses belonging to a DAO.
Future implementations of the objective function can be also be made with different power indices that aim to express exponential correlations between stake size and their corresponding "decisiveness" (i.e. the ability to swing a vote) within an organization operating under a weighted voting system. In that sense, the work of Brukhman and the Banzhaf method point out that under certain distributions, smaller stakes have infinitesimally small or zero decisiveness, rendering them valueless. Conversely, sometimes adding marginal stake to a position disproportionately increases its decisiveness. Additional metrics can be taken into account such as the number of ongoing votes in every proposal, or the evolution of the DAO over time (traction). A high position within the Democratic Index ranking is therefore designed to incentivize the creation of more egalitarian organizations.
The very nature of public blockchains cries out for universal equality. It cries out for rule by the people - not by one person or a group of people, but by all people. By you! We think that Equality Protocol is the most equitable governance process to predicate this view of democracy. While Equality scores constitute people on Ethereum, Equality applications are the spaces where universal rule-making happens. Equality applications are the building blocks for the federation of the universe - giving each person equal say in how universal laws and policies are formed. Such applications may range from universal basic income and quadratic funding programs to elections and other decentralized governance that is not financial at all.
Equality Protocol accounts for two variables: account addresses and their respective Equality scores. It stores the Equality scores of listed accounts on the Ethereum Main Network. As soon as the objective or subjective input for an account changes, the entry for this account is updated. Equality applications are fully interoperable with other chunks of code on Ethereum, such as token systems or financial applications.
At an on-chain engineering level, developers of Equality applications set a minimum required Equality score for accounts to transact with them. Hence, only accounts that meet the Equality score requirements may claim the right to participate in their application. See the figure above and consider the following examples. Only accounts with 0.84 Equality and up may claim the payout of the basic income application. Or, only accounts with 0.90 Equality and up may vote in the democratic body application. The scope for action is limitless. Each Equality application is free to determine its custom required score. Thus, it is up to developers of Equality applications to find the best trade-off between Sybil-attack resistance and openness. Experience will tell what Equality scores indicate trustworthy personhood.
In a nutshell, Equality Protocol is a censorship-resistant and open source constituent for those who are building truly social applications on Ethereum. It lets you build a world with equal political power for everyone.
Money is understood in its fundamental role of serving as a measuring stick for economic activities. However, as economist George Gilder puts it, currencies based on commodities doom valuations with self-referential loops (in the same way that atoms cannot be measured with atoms, or minds cannot be gaged with minds). Simply put: measuring sticks cannot be a part of what they calibrate. Time instead, has its roots in a grid of measurement beyond commerce. It is the only irreversible physical constant, making it the ultimate frame of reference for all measured values.
A protocol able to formalize humans can mint tokens for participants based on elapsed time since verification. Therefore, the most salient aspect of a time-based cryptocurrency is its connection to a robust identity protocol. From an economic point of view, governance and identity have a symbiotic relationship: a vote only has value if it comes from a legitimate voter. Thus, the matter of legitimizing digital entities is fundamental to building a working political economy. While strictly financial transactions can be completely anonymous, political transactions only have value if able to ensure legitimacy. By using time as a reference, our aim is to create a humans-only crypto economy where the starting point is the same for everyone, regardless of the place of birth.
Approaches aiming to prevent a Sybil Attack must ensure that it is possible for an individual to obtain one validation within its domain, but unfeasible to obtain two or more. Equality inverts the angle with which former protocols address this challenge: instead of establishing a binary output (human / sybil), it defines entities within a continuum, approaching identity from a probabilistic lens. Equality circumvents the formalization of a unique process for identity verification by creating an intersubjective space that draws metrics from the embedded trust and sociality already present within Decentralized Autonomous Organizations, as well as the objective measurement of their democratic virtue. As it assigns a score to addresses based on the intersection of DAOs into which they belong or their relative position in the social graph of blockchain transactions, it can begin to formalize a path for an intersectional blockchain-based identity, as outlined by Immorlica, Jackson & Weyl on “Verifying Identity as a Social Intersection” (2019). Finally, Equality can serve as a framework to benchmark the informational threshold needed for a consensus able to formalize humans in blockchain networks.
These are some of the minds that have contributed to this paper by reviewing it.
Primavera De Filippi (CNRS & Harvard University), Glen Weyl (RadicalxChange, Microsoft & Princeton University).