Curated Governance with Stake Machines

Dimitri De Jonghe
Dec 4, 2017 · 9 min read

co-authored by Gautam Dhameja, Tom Debus and Günsu Pancar.
direct or indirect intellectual input by Trent McConaghy, Tim Daubenschütz, Don Gossen, Jernej Pregelj, Rhys Lindmark, Jake Vartanian, Luke Duncan, Mike Goldin and Simon de la Rouviere.

An image to attract the attention of the reader.

Humans have the tendency of being egocentric. Personal gains are more tempting than doing something for the common good. It’s our will to survive and thrive and it’s easy to forget that collaboration probably leads to more sustainable solutions. But not all is lost… I hope…

One approach could be to give someone a reward when they accomplish some objectives. Self-interest is fed by the reward and steers the human to do something that’s useful for “the system”. Let’s now think of “the system” as the common goods, public interest and global governance. Worth a try…

Looking at game theories I stumbled upon token-curated lists. And they look pretty neat to me. Here is my take on it: steer token holders to curate items that should or should not belong to a list. If your opinion reconciles with the majority, you win more token. So, one is encouraged to predict what others think about that specific item. Here’s the catch: “predict what others think”.

Now, governance of the common goods is a bit more complicated than a single list of curated items. People and governed objects have various states, funnels, interconnections, dependencies and so on. Here we explore token-curated lists as building blocks for more complex curated automata.

Curated Governance

Belgium. A small country with 6 governments, 2 kings & queens, 3 communities, 3 regions, 9 provinces, 3 official languages, coalitions, oppositions and what not? In the meanwhile it holds the record for 540 days without a government. As a Belgian, it kinda felt as if nothing really changed and daily life went on as always…

Governance by tribes takes over in self-organizing communities (district0x, Aragon, nativetoken). Like-minded people signal ideas, opinions and influence. It’s necessary to build anti-fragile networks that flourish whilst remaining sustainable. See how tribes and subcultures form around common interest and reputation in social networks, knowledge platforms, open-source, art, and so on. Tribes exchange ideas and curate information they care about. Curation processes sift out the relevant information such that we know what’s hot or not and detect scams and time-sucks.

Give expert communities a carrot and a stick and they become self-sustaining. Reputation (reddit, StackOverflow, HackerNews, comics, music, peer-reviewed publications, …) or financial gains (crypto-economics, prediction markets, freelancing, …) pushes community members to process information within their expertise domain (curation, modeling, review, ….). Curators and curated topics have an incentive to form an integral connection with the community.

Token-curated lists (Mike Goldin et al.) and curation markets (Simon de la Rouviere et al.) show potential for incentivized community-driven governance. Items on these lists are vetted by the curators and may yield benefits such as attention, visibility, extended permissions and so on. We’ll cover an extended version that models multiple curated lists and dependencies between them. Much like a finite state machine or automaton.

However alienating the above concepts might sound, they are actually native to our current societies. One can think of how people get elected or achieve status, how bills pass through parliament, gaining gradual benefits when climbing up a ladder (credit scoring, KYC tiers, researcher’s status, …).

Knowledge graphs such as in Sid Meier’s Civilization and Starships and Tokens illustrate the point of tokenized governance. Here, one needs to evolve their civilization by allocating budget on specific research items. For example: once a civilization learned the alphabet, it could move on in learning how to write. Writing combined with pottery allows one to gain the knowledge of map making and so on. In a democracy, the allocation of budget requires consensus and misuse leads to revolts and looting parties. A related interesting read is Trent McConaghy’s blogpost on Starships and Tokens.

Token-Curated Registries

The principles of token-curated registries (TCRs) are simple: a group of like-minded souls curate a list of items using a propose-challenge mechanism. One can propose an item to go on the list and supports the proposal using bonded stake with a token that is native to the system.

Sketch of Token-curated Registries [image credits: Tim Daubenschütz, Don Gossen]

Each proposal can be challenged or not. During a challenge, stake is matched and a voting round starts. The winning party feasts from the looted stake and the losing party is slashed. When a proposal loses it is removed from the list.

This mechanism incentivizes curators to participate in the list, as each proposal for change could lead to a token reward. A useful strategy is to comply with the majority of the community by estimating or predicting the outcome of a challenge. Interestingly, individuals are rewarded by predicting community goals…

Labels & Rules

When dealing with multiple “tribes” of experts or distinct verticals, it might make sense to spin up multiple TCRs or allow to filter by labels and signals. Curation groups might then focus on different governance objects, hence separating signal from the noise.

A Labeled TCR. A Proposal might appear on multiple times on the list under different labels.

One example for Labeled TCRs can be the categorization of music by genres. Projects such as MusicMap and Spotify’s algorithmic map allow for a visual interconnected exploration of music. Currently there are only one or a few curators (the website owners) that use their expert knowledge to map songs onto genres and find connections between genres. If one can incentivize a community, subculture or other stake holders to map our music history in return for reputation, attention or token, this could lead to a drastically new approach to music discovery by curation.

A curated playlist of top-10 songs in a specific genre. Each genre can be regarded as a label or a filter.

One step further would be to attach label-specific rules. This allows to model interactions and links between TCRs. Many interesting use cases follow: permission groups, data governors, policy with hierarchy and escalation, funnels for reputation and standardization, curated knowledge domains, social graphs and groups, ...

Here we explore the notion of Finite state machines (FSMs) in a curated context. FSMs are a type of automata that contain a finite number of states and transitions (state memory + combinatorial logic). Automata are used to model broad classes of physical and behavioral processes.

Stake machines model governance using the concepts of curation and FSMs. This can give a community an incentive to participate in global goals and avoid steering towards a tragedy of the commons. To achieve this, we’ll need a few building blocks:

Schematic of a Labeled TCR with Rules

Token & Stake

Tokens are units of stake.
Stake is used to express opinion.

Tribe & Opinion

Tribes are opinionated community groups where the members hold token.
An opinion is expressed by means of bonded power (ie staked token or curation market).

Proposals & challenges

Proposals capture every change to the governed “state” of the community.
Challenges are finite epochs where opinion is expressed upon proposals through voting or similar.

Proposers & challengers

Proposers want add items and bond it with stake.
Challengers oppose to a change and bond it with stake.
or dual:
Challengers want to address issues and proposers solve these issues and thereby taking the issued stake as a bounty

Labels & Rules

Each state in a stake machine has a label and group together proposals.
Each label can implement a specific set of rules.
Rules themselves can be proposals and may be challenged.

The rules model the conditions under which changes may occur for a specific label. Hence it’s possible to constrain the proposals, challenges, token holders and model relationships in the system. In order to resolve and validate the rules, all the necessary information needs to be available to the system or token holders.

An example of state rules with dependencies


Say we want to implement a governance funnel with some stages. Each stage can have a label like 1, 2, …, X, X+1. In order for a proposal to be PROPOSED in stage X+1, it needs to be ACCEPTED in the previous stage X. This can be implemented by attaching a rule to the label that checks if the proposal is ACCEPTED in stage X. Such rules can be enforced by the system using some smart contract language. Similar rules can be devised to have label-dependent MIN_DEPOSIT or restricting the curators to a specific label.

Governance objects can gradually evolve through a stake machine as long as they comply to the rules set by community. Restriction of the curators per label enables tribes, roles, permissions, KYC tiers and so on. Think of splinter groups that fork from a set of rules and make them more stringent to make membership more “exclusive”.

As with any system, complexity and security tend to be at odds with each other. Simple queries and logic is advisable at the semantic level of the rules.


Open-source governance: Github, RFCs, forks, peer review, …

Open source code, documents, papers, standards, etc undergoes a curation process that’s governed by a group of experts that maintain a repository. These communities implement a form of hierarchy to reward behavior that’s in line with goals of the project. Frequent contributors decide on which changes will make it into the project or not.

For example: code changes are represented by pull requests (PRs) and request for approval. When approved, then everybody who downloads and runs the code will use that change. It’s important that these changes don’t contain malicious code or viruses, hence curation is done as a means for security audit.

Stake machine for open-source code governance. Notice how Rules can apply both on governed objects as well as curators and proposers.

It’s easy to setup a stake machine that would incentivize the community to crowd source the project with code and audits. The roles in the community are Requesters that contribute PRs and Reviewers that have proven their skills and project mind-share in the past. In order to belong to this group, again a TCR per group can be implemented, making the list of curators curated again. The incentive can lead to more project engagement and larger communities because holders earn tokens that represent reputation, attention, governance, donation, …

Curated Utility Networks

In a following example, we can explore governance in tokenized systems such as public utility networks. Validation and policing of proofs in return for a token reward is framed as a staking mechanism, as well as protocol changes and corresponding voting weight. It’s closely related to Proof-of-Authority.

Curated governance in a purely fictional public utility network.

The above picture is a bit hyperbolic, but in the world of blogs it doesn’t really matter.

Of course there are many other governance objects like smart contract code, oracles, scaffold-based education, curated AI’s, data quality, semantics, rules, norms, and so on…

Wrap up

Exploring the potential of curated governance is quite an exciting pastime. This blogpost is incomplete of course. Still many important pieces need to be developed, proven and so on.

But, following efforts by and for the community are probably the way forward — or just make me happy:

Feel free to drop a line in the comments or email dimi<at>bigchaindb<dot>com.

co-authored by Gautam Dhameja, Tom Debus and Günsu Pancar.
direct or indirect intellectual input by Trent McConaghy, Tim Daubenschütz, Don Gossen, Jernej Pregelj, Rhys Lindmark, Jake Vartanian, Luke Duncan, Mike Goldin and Simon de la Rouviere.


2017/12/07 Minor fixes and more attributions

Thanks to Gautam Dhameja, Don Gossen, Rhys Lindmark, and Luke Duncan.

Dimitri De Jonghe

Written by