Summary 

Governance is crucial to the future of the Mina Protocol for key decisions to be made effectively. For example, what are the rules and processes for making changes to the protocol? And how should the development of the protocol and ecosystem be funded? To realize Mina Protocol’s vision of an internet of true things powered by users of Mina’s proof of everything, these decisions also need to be aligned with the wishes of the Mina community. Consequently, all key stakeholder groups in the ecosystem need to have a voice, agency and support to participate in decision making.

This raises the question: how should Mina’s Protocol’s governance be organized to harness the intelligence of individual community members to make the best decisions? To answer this question, this is the first in a series of blog posts that explore collective intelligence where groups of people are organized at scale to solve public problems in ways that often outperform individual people alone.

First, we introduce collective intelligence and insights from biology, including the decision making processes of brains and swarms.

Inspired by these insights, social science researchers have developed high level frameworks for organizing groups of people to make decisions as a ‘hive mind’. We summarize these frameworks and frame Mina Protocol’s governance around them since some of their methods are already being implemented.

Ongoing experimentation and learning is central to collective intelligence. Public problems have become increasingly complex, including the emergence of wicked problems that have multiple causes and nonlinear dynamics so there are no obvious or clear solutions. Feedback loops and interdependencies mean trying to solve one aspect of the problem may create other problems. Proposals for solving problems cannot be verified in advance but only from practice by implementing and testing them along the way.

Since governance is a complex, ‘wicked problem’, it is impossible to know at the outset how best to configure Mina Protocol’s governance processes to optimize their efficacy and alignment. We conclude by explaining our experimental and incremental proposal so that Mina’s community can iteratively add, test and improve governance processes based on real world learning.

The sum can be greater than its parts

Imagine harnessing the cognitive power of a large number of individuals to tackle society’s most pressing challenges. This is the essence of collective intelligence where groups of people are organized at scale to solve complex public problems in ways that often outperform individual people alone. 

Collective Intelligence has roots in a long intellectual history. Ancient Greek philosophers argued that the expertise to govern a city was found not in exceptional individuals but in the community as a whole. Today, there are many societal examples of collective intelligence, such as:

• Wikipedia where people freely create and edit content to create a global knowledge commons.
• X’s crowdsourcing of fact checking of tweets and social media content.
• E-petitions where citizens propose topics for politicians to debate and even ideas for new laws.
• I CHANGE where citizens monitor their cities and turn their city into a living lab. 
• NASA’s efforts for people to collect, measure and classify data and make scientific discoveries. 

Learning from biology

There is order in biology across levels of organization from cells to organs, organisms and even societies. Each level solves its own type of problem whether metabolic, physiological, anatomical, behavioral or societal. One finding is that collective intelligence features across these scales of organization since evolution has re-used similar processes to solve these different types of problems. Consequently, collective intelligence spans a range of disciplines, including biology, computer science, ecology, organizational psychology, political science and sociology.

Towards one end of this range is the study of brains that can provide insights about how to organize a collection of unintelligent neuronal subunits into an intelligent collective. Specific communication and functional links are needed to connect decision making subunits in ways that can be updated based on real world experience and learning. Subunits act on local information based on feedback and rewards, and their behavior is sensitive to the behavior of other subunits closeby. 

Introducing swarm intelligence

Further along this range is the study of swarms, including ant colonies, fish schools and bird flocks (see Figure 1). There has been extensive research about the decision making process of honey bees to select a new hive. Scouts leave the swarm to search for suitable sites and then return to report about them through their waggle dance- a series of movements that represent the direction and distance to a new site. Scouts report for longer and more excitedly about better quality sites so that other scouts will also seek them out and return with their own reports. Importantly, scouts report about a site only if they have first hand experience and have visited it themselves.

Figure 1 Example of swarming in the natural world

This process can be modeled in terms of decision making subunits (individual bees) that represent alternative options (different hive sites) and compete and inhibit each other until their excitation exceeds a threshold, causing their option to be selected as the swarm’s decision (see Figure 2a). There are similarities with how brains make decisions where neuronal subunits, representing alternative options, compete and inhibit each other until their excitation exceeds a threshold, causing their option to be selected as the brain’s decision (see Figure 2b).

Figure 2 Similarity between models of brain and swarm decision making
Figure 2a Model of mutually inhibitory bee decision making	Figure 2b Model of mutually inhibitory neuronal decision making

Building social consensus

At the other end of the range is the building of consensus amongst groups of people. Social choice theory has long studied how individual opinions, self-interests and values compete and propagate through groups, and are aggregated so that the group accepts one option as its collective decision even if everyone is not equally satisfied by the outcome.

However, some assumptions of social choice theory are not realistic, such as individuals forming opinions independently of each other. Swarm intelligence is now providing insights about the dynamics of how opinions emerge, spread and compete in networked groups.

Organizing collective intelligence for groups of people

Inspired by these insights, social science researchers have developed frameworks for organizing groups of people as a ‘hive mind’ (see Figure 3).

Figure 3 Principles for organizing collective intelligence

Who should participate?

The decision making subunits

Members of the group should be competent for the task at hand. They need access to distributed sources of information to ensure they do not use the same flawed information. 

They should be cognitively diverse. The ‘wisdom of crowds’ relies on everyone making mistakes in slightly different ways so that people with a tendency to overestimate can be corrected by others who are more likely to underestimate. 

Rewards for participation 

They also need to receive value from their participation whether due to intrinsic motivation or external rewards. This will help to ensure they feel they can contribute their views equally and independently. 

How should they participate?

Communication between members

Members must be able to communicate effectively at a reasonable cost. They can operate on different timescales whether synchronously when they interact at the same time or asynchronously when they participate at different times. Groups may even be more productive if they work asynchronously with bursts of rapid communication and synchronized activity.

Functional links between members

Solving specific problems requires organizing members around one, some or all of a range of ‘core functions’ (see Figure 4) and choosing the most suitable methods for each (see Table 1).

Figure 4 Core functions of collective intelligence
Table 1 Examples of methods for each core function of collective intelligence
Core functions	Examples of methods	Examples of Mina using these methods
1) Observe and gather data to identify a specific problem	Discussions forums Crowd surveys and polling  E-petitions	Poll on MinaResearch
2) Analyze and model the data to understand the problem and make predictions	Forecasting methods NLP & sentiment analysis	Discord bot surveys
3) Generate options that could solve the problem	Hackathons Incubators & innovation programs	Submit zkIgnite funding proposals Research & draft MIPs
4) Deliberate on these options then choose the best ones	Focus groups Citizens’ Assemblies Voting	Electors in zkIgnite SMEs in the new MIP process
5) Implement the chosen options and take action	Sandboxes Controlled trials Monitoring tech	Enact MIPs Fund zkIgnite teams
6) Learn from the actions taken	Retrospectives Wikis Knowledge base	Reviews of MIP and zkIgnite processes

Linking people and machines can significantly increase the potential of collective intelligence. But smarter tools to collect and share information do not entail more intelligent results. Machines are good at some core functions but poorer at others. With each step up from gathering raw data, it becomes harder to automate processes that become less routine and involve more qualitative and contextual considerations. Effective machine-human teamwork is needed. 

Deliberation of competing options

Deliberation, in particular, needs effective facilitation, specific tools and even dedicated spaces. Through learning, listening, dialogue and debate, the full range of different arguments and opinions need to be considered, including trade-offs. Members need to feel free to provide judgements independently and respectfully without fear.

Thresholds for group decisions

Fair and effective methods, most often involving majority voting, are needed to aggregate support for different opinions to arrive at a collective decision. Different decisions involve different thresholds: the quorum of participation- must everyone participate in the vote or only a certain number of people?- and the size of the majority, e.g a relative majority of simply more than half or a supermajority perhaps more than two thirds? 

Real world experience and learning

Proposals for solving problems cannot be verified in advance but only from practice by implementing and testing them along the way. Ongoing experimentation and learning is perhaps the central function of collective intelligence (see figure 1). 

This is one of the reasons why there is renewed interest in using the methods of collective intelligence to respond to major societal challenges to the legitimacy of democratic decision making. 

Public problems have become increasingly complex, involving a variety of interests of different stakeholders and needing diverse sources of information and expertise. Problems are difficult to define and may change during the time between when decisions are made and take effect. 

‘Wicked problems’ may be the most complex. They have multiple causes and nonlinear dynamics so there are no obvious or clear solutions. Feedback loops and interdependencies mean trying to solve one aspect of the problem may create other problems. For example, tackling global warming is a wicked problem. There are many aspects to the climate system yet its complexity means that interventions in one area could have unintended impacts on other areas. Interventions have economic and social consequences, such as energy use, transport and urbanization, that could worsen other problems, such as development and poverty alleviation. Consequently, it is hard to rely on past experiences so new approaches based on ongoing experimentation and learning are needed to arrive at the best solutions.

How many should be in the group?

Large groups can be more intelligent but not always. Deliberative methods can be difficult to scale. Ensuring high quality deliberation online may not easily capture the richness of real life conversation. 

Effective collective intelligence also depends on its ‘we-ness’- what makes a group feel that it is more than just a collection of individual people. A strong culture is needed that makes its members feel bonded based on common interests and values, and a shared model of the world and what is important. 

Lessons from real world examples 

Researchers have analyzed a range of real world examples (see Table 2). Some examples failed due to poor design if they were too difficult for the public to use or too easy so that contributions were too simplistic, lacking supporting evidence or implementation plans and drowning institutions in unusable feedback. Other examples failed since they lacked political support; after all, successful projects could pose a challenge to traditional forms of power.

How Mina is implementing collective intelligence 

Mina’s governance can be framed around these core functions and is already implementing some of their methods. Lessons are also being applied from real world examples of institutionalizing collective intelligence.

There are two primary types of decisions that the Mina Protocol’s governance should consider: decisions about the protocol and decisions about ecosystem funding. The Mina Improvement Proposal (MIP) process is the primary mechanism for protocol decision making (see Table 3). A community based approach to ecosystem funding could follow a similar process based on existing programs, such as zkIgnite. 

Observe and gather data to identify a specific problem

The recently published governance proposals encourage greater community conversation on MIPs. This includes the initial research stage where community members can discuss ideas on MinaResearch and seek wider feedback, using polls to gauge community support to assess whether to formally submit the idea.

Analyze and model the data to understand the problem 

The governance proposals stress the potential of mass deliberation tools to capture different perspectives across the community and support robust discussion between dozens, hundreds, or even thousands of people. Mina Foundation’s Protocol Governance Team developed a Discord bot to survey the community, gather sentiment and summarize feedback about the proposals. Open sourced and accompanied with explanatory documentation and videos, this tool is available for the community to modify and adapt to its needs (see Lessons 4 and 5 in Table 2). Another blogpost will explore mass deliberation tools in more detail.

Generate options that could solve the problem

Protocol decision making

The MIP process is open to all community members to propose new changes and new features to the protocol, and the process is explained on Github (see Lesson 1 in Table 2). Authors use a formal template to submit MIPs to ensure a precise problem is clearly articulated and well defined (see Lessons 2 and 3 in Table 2). 

Ecosystem funding decisions

zkignite has been Mina’s leading innovation programme. It, too, is open to the community and provides a formal process for submitting high quality funding proposals. 

Deliberate on these options then choose the best ones

Focus groups

The governance proposals were shared with a group of community members for initial feedback, who provided detailed comments on draft documents. Wider deliberative activities are planned, such as regular community Town Halls, governance Q&As, discussion channels and forums. 

Citizens’ assemblies 

The governance proposals discuss the potential of citizens’ assemblies to randomly select a group of people that is broadly representative of a community or society. This group deliberates on a public question by considering the full range of different arguments and opinions and then recommends solutions to decision makers. 

Another blogpost will explain how zkIgnite already implements this approach when electors review proposals and collectively decide which to fund. That blogpost will also explain how a similar approach could be implemented in the MIP process.

Implement the chosen option and take action

Implement MIPs 

Every MIP should include a plan that explicitly states what needs to be done to implement the proposed changes, who is responsible for this implementation (in agreement with the MIP author) and what are the target dates for each milestone. The plan must be transparent so that the community can verify the changes are implemented as intended. 

Release funding for ecosystem building

Once the electors have made their decision, the final stage of zkIgnite involves providing funding, as well as opportunities to connect with industry experts and VCs, to the winning teams so that they can start building.

Learn from the actions

Governance is a wicked problem, so it is impossible to know definitively at the outset the best way to 

configure Mina Protocol’s governance processes to optimize their efficacy and alignment with the community. Consequently, the governance proposals emphasize an experimental and incremental approach so that the community can iteratively add, test and improve governance processes based on real world learning (see Figure 5). This can begin simply and gradually by increasing the complexity of additions and tests so that the scope of the Mina Protocol’s governance can grow. Doing so can build ecosystem capacity, confidence, and motivation to foster wide participation. 

Reviews of process changes could include a community wide conversation facilitated by mass deliberation tools, as well as smaller reviews by committees, to make recommendations on any further changes (see Lesson 7 in Table 2). For example, after the first on-chain vote, community feedback meant changes were made or the next on-chain votes so that accounts, who had delegated to other accounts, could vote directly without having to first un-delegate. Thanks to iterative feedback sessions with participants, new improvements were also applied to the zkIgnite program; for example, a scoring system for electors was implemented, a technical review team was introduced, voting criteria were changed and voting mechanisms were updated.

FIgure 5 An iterative approach to developing Mina’s governance

Let’s build Mina’s hive mind

Collective Intelligence provides a deep source of insights for how to harness the cognitive power of Mina’s community. Inspired by insights from biology, social science researchers have developed high level frameworks for organizing groups of people to make decisions as a ‘hive mind’. We hope to inspire ideas from the community to help build Mina’s hive mind! 

An experimental approach means that we look forward to working with the community to iteratively add, test, and improve governance processes based on real world learning. Please connect with us on Discord at the following channels:

#protocol-governance-general-discussion

#protocol-governance-announcements

#protocol-governance-surveys