Moderating a DAO

Our fight against infrastructure monopolies would be fruitless if we were not actively planning how to decentralize Pocket’s governance. To eliminate the final point of centralization in Pocket’s decentralized infrastructure stack (us, Pocket Network Inc) we must eventually cede control to a community-governed Decentralized Autonomous Organization (DAO). As part of this vision for total fault tolerance, we are researching a governance interface that will enable the DAO to utilise any blockchain and any organizational technology (see our whitepaper for more details). In the meantime, we’re also thinking about how the DAO itself will operate. One Pocket DAO governance mechanism we designed, with the help of the Aragon One team, is on-chain moderation.

DAOs as Communities in Need of Stewardship; Moderators as Community Stewards

It is naive to expect a DAO, like any community, to operate effectively without stewardship. Low voter engagement is a widespread issue for which many solutions are being developed. However, there are a number of other governance processes which also deteriorate with low engagement, either because they are manual (requiring a volunteer to drive an action forward) or because they are complex (requiring more attention than the average voter can spare).

Most communities have norms which define membership rights, expected community behaviour, proposal requirements, and the quality standards that contributions must meet. Supervising and enforcing these community norms is a process that is both manual and complex. So who will do it? If we leave this question unanswered, we risk organizational paralysis.

In our research, we have observed many DAOs resolve this problem by delegating the responsibility off-chain. DAOstack’s Genesis DAO formed a working group known as the Accountability Task-Force (ATF) (which some now wish to automate), who would be compensated for the responsibility of monitoring proposal progress, keeping contributors accountable, and defending against attacks. The first version (AGP-1) of the Aragon Governance Proposal (AGP) process delegates the role to the Aragon Association and AGP Editors, who are selected by the Association to manage the AGP workflow, enforce proposal requirements, and forward compliant proposals to the voting stage. Aragon’s Editor role is in turn borrowed from Ethereum’s EIP process, Bitcoin’s BIP process, and Python’s PEP process.

We do not want to rely on off-chain stewardship, because this will inevitably undermine any on-chain governance mechanism that falls under the stewards’ purview. However, if we do not design formal structures to support voluntary stewards, with fair rules for compensation and accountability, would-be stewards will simply fork the community into an alternative organizational structure that supports their efforts.

The obvious solution is to implement on-chain stewardship. But how?

If DAOs are community-owned organizations, perhaps we can model our on-chain stewards after the stewards of traditional online communities: moderators. According to Grimmelmann (2015), a Cornell internet law professor, moderation can be defined as “the governance mechanisms that structure participation in a community to facilitate cooperation and prevent abuse.” Echoing this, researchers from Carnegie Mellon’s Human-Computer Interaction Institute, who interviewed volunteer moderators of communities on Twitch, Reddit, and Facebook, found that “users can be very effective at self-governing when given the tools to do so.”

Rather than reinventing the wheel, perhaps we can consider these lessons as we attempt to bring moderators to DAOs:

• Active Volunteers: Carnegie’s researchers observed that most moderators attain their role by being stand-out members of their community and volunteering for the role.
• Automatic Filtering: Community platforms provide algorithmic tools (e.g. Reddit’s AutoModerator) to filter objective content infringements, so that moderators can focus on decisions that require context-specific judgement.
• Kick Repeat Offenders: Accidental one-off infringements do not typically result in removal from a community, depending on the severity of the community member’s actions. However, moderators will remove repeat offenders.
• Quality Control: Moderators spend most of their time approving/rejecting content submissions and membership requests according to compliance with pre-defined community rules.
• Moderator Accountability: According to Carnegie’s researchers, “moderators… made decisions by executive fiat usually without community input, and could not be removed from their positions except in some cases by other moderators.”

Introducing Two New Apps for Moderating DAOs

During the last few months we have been working closely with Luke Duncan and Aragon One’s user advocacy team. After we shared with them our on-chain moderation use case, they brought in Aragon One’s developer team, who got busy building two new tools during a hackathon at one of their team offsites.

Filtering: Aragon Tollgate

Tollgate is a simple Aragon app forwarder that requires a user to deposit a fee before submitting a proposed action. This can be used to disincentivize spam or malicious submissions on certain actions. To ensure that good actors do not lose out, DAOs can pay back tolls to those who submit compliant proposals.

This replicates the filtering of the Web 2.0 AutoMod in two ways:

• Barrier to Entry: You can only initiate actions if you have managed to obtain the toll token. DAO designers can customise the strength of this barrier by choosing their own toll token and controlling its supply. For example, designers could restrict toll tokens so that they can only be earned by completing bounties for the DAO, so that only past contributors can initiate new actions.
• Penalty for Non-compliant Proposals: If your proposed action is rejected for not following the rules or norms of the DAO, you will not recoup your toll. The result is that members will self-moderate to minimise their risk of toll loss, i.e. check that they are following the rules before submitting actions. The strengths of this incentive can be customised depending on the source and supply of the toll token. For example, designers could implement a staking system, where some monetary token external to the DAO must be staked to mint toll tokens, these toll tokens can in turn be redeemed for the original stake, and any loss of toll tokens will result in the proportional loss of the stake if the member does not find other ways to recoup them.

This also means that moderators do not need to manually remove repeat offenders, they will naturally lose their proposal rights as they waste their toll tokens.

Moderating: Aragon Approvals

Approvals is an Aragon app that queues proposed actions (“intents”) and only forwards them (e.g. to the Aragon Voting app) if they are approved by a member with approval permissions. This can be used to batch proposals or enforce proposal requirements.

DAO designers can customise the authority of moderators by changing which actions take this route. For example, they may choose to only give moderators control over which proposed financial transfers make it to voting, or they may give moderators more control over membership rights by assigning the power to approve/reject proposed token mints/burns.

In our current design, we give moderators the power to approve/reject all actions except those which are initiated by other moderators. We justify this level of power because everyone else has the power to fire malicious moderators, by proposing a moderator token burn, and they only need one cooperating moderator to approve their proposal and forward it to voting.

So how do we apply these apps to the lessons we considered earlier?

• Active Volunteers: Enable any active DAO member to volunteer by proposing the minting of a moderator token (subject to approval/voting).
• Automatic Filtering: Introduce a toll which incentivises self-moderation, replicating the filtering function of the Web 2.0 AutoMod.
• Kick Repeat Offenders: Naturally remove repeat offenders as they run out of toll tokens to spend on new proposals.
• Quality Control: Enforce quality control on certain proposals by routing them through the Approvals app.
• Moderator Accountability: Unlike most Web 2.0 communities, give members the power to “fire” malicious moderators by, appropriately, burning their moderator tokens.

These apps are ready to use on the rinkeby testnet, but will need to be audited and deployed to mainnet before most Aragon DAOs can begin using them. Once they are, DAO designers will be able to integrate and improve upon the principles of Web 2.0 moderation, ultimately creating more organized organizations.

Thanks to Luke Duncan, Maria Gomez, and the Aragon One team for their support in refining our use case and building the apps. Thanks to Philippe Honigman and Pocket Network Inc for their comments.

Subscribe to the Pocket newsletter to stay tuned on future Pocket governance news.