Commission for Global CoordinationGlobal Secretariat · Established 2041
Artificial
Governance
A common framework.
A continuous mandate.
Open study series / Governance

A DAO is an arrangement among people, not an artificial mind

Shared rules and records can help people pool resources, recognise work and keep commitments across distance. Their most promising role is to make cooperation easier to understand, carry out and improve.

What this could make possible

Make a shared undertaking easier to join and improve

Imagine a community fund whose members can understand its rules, follow its spending and propose useful work. Routine commitments can run reliably while people retain a clear way to evaluate outcomes, resolve disputes and change direction.

A useful promise: commitments people can inspect

Vitalik Buterin's 2014 terminology essay describes DAOs as organisations with continuing human involvement, distinct from fully autonomous artificial intelligence. It is a historical proposal for definitions, rather than a permanent standard. The useful opening is cooperation around shared assets and procedures. Imagine contributors in several towns maintaining a common repair fund: they can see the budget, the conditions for payment and the authority required to change those conditions. Software may make agreed steps repeatable and records easier to inspect. The design question is where that reliability helps people work together and where judgment needs to remain available.

Inside the Commission

Moving the Commission's current rules onto a blockchain would preserve the same consultation deadline, service exclusions and circular mandate. A reliably executed bad rule remains a bad rule. Shared records could make some actions easier to inspect, but somebody would still select the measurements, validate reports from the physical world and decide who can propose a change. An immutable water ledger cannot establish that an unregistered settlement has received enough water. Those facts require people, instruments and an accountable process for disagreement.

Recognise contribution beyond the initial investment

Backfeed's 2016 economic proposal explores community assessment of contributions, reputation and token rewards, including possible equity, access and currency roles. Its model explicitly relies on human judgment for value that a computer cannot verify alone. This offers a productive question for a community fund: could someone gain recognition through maintaining shared infrastructure, rather than only by arriving with money? Our illustrative answer would make proposals, expected outcomes and evaluation criteria visible before work begins. Contributors could understand what they are undertaking, while residents could influence which work matters. Recognition would be reviewable when experience changes the assessment.

Keep four promises clear enough to honour

For this site's experiment, distinguish access to essentials, a common dividend, additional rewards and political voice. This is an editorial design to test, not a complete model attributed to Backfeed. A dividend distributes available net revenue; contribution rewards recognise agreed work; an equal vote expresses membership. A guarantee of essentials would require its own funding and obligations. Keeping these promises separate lets the community reward useful effort without making every person's standing depend on a score. It also makes the finances intelligible: the same surplus cannot fund both a full dividend and a full reward pool.

Make the account broad enough to notice care

SourceCred represents contributions through a graph of people and work, allowing recognition to follow dependencies. DisCO brings care into cooperative accounting alongside paid and commons-producing work. These are different approaches, but both invite closer attention to what a simple output count misses. In our example, reporting a failure, diagnosing it, repairing it and helping households during the outage may each matter. A useful account could combine recorded work with resident testimony and a review by people outside the immediate team. Some support can be funded collectively, avoiding a demand that every kind act become an individual claim for payment.

Give correction a place in the original agreement

Jeffrey Wilcke's July 2016 fork announcement records choices facing Ethereum participants around The DAO intervention. It is a contemporary account of a disputed event, useful for seeing that people still make consequential decisions around software rules. Our proposed fund would discuss correction before a crisis: who can pause a payment, what evidence is required, when a pause expires and who reviews compensation. The purpose is to preserve confidence that mistakes can be addressed. A limited remedy can protect contributors and recipients while leaving an inspectable record of why the ordinary process was interrupted.

Reading behind this discussionTo Fork or Not to Fork

Build a modest institution worth joining

Begin with a bounded community project and a budget people can afford to experiment with. State the work, membership, payment conditions and review process in ordinary language before choosing technical tools. Test a missed deadline, disputed contribution and month with less revenue. People should be able to discover what happens without decoding a contract. Compare the effort saved in administration with the effort added in evaluation and participation. A useful outcome might be a repair completed, a dependable shared account and a newcomer able to participate confidently. Those are meaningful advances even before any wider network emerges.

Which shared commitments could software help people keep, and which decisions should remain open to human evaluation and correction?
Your experiment / Recognising work

Make room for work that matters.

Change the split between the common dividend and contribution rewards, then compare activity-based recognition with an assessment that includes care. Reduce available revenue and inspect whose income changes while each person's political vote remains equal.

Open this experiment