Distributed Cooperative Organizations (DisCOs) and SourceCred

This post describes how Distributed Cooperative Organizations (DisCOs), a governance model SourceCred is exploring, rewards care work. I hope to explore other aspects of DisCOs in the future if there’s demand.

POINT OF PROCESS: I’m not advocating here that we start a DisCO. Not yet anyway;) It appears Sociocracy has gained momentum, and the Sociocracy Working Group (which I’m also a part of) is working on a first pass at a Sociocracy implementation. I don’t see this as conflicting with DisCOs, per se. The models seem very compatible, and we could start with Sociocracy, then incorporate elements of DisCOs over time if we choose. We could even eventually migrate to a full DisCO it seems without major structural changes.


At the Governance Jam, I presented on the DisCO framework. There were some unanswered questions, particularly around care work (something we’ve struggled to quantify). I volunteered to write a post, thinking it wouldn’t take long. Whoa boy…Turns out there is a lot more to the framework than I’d previously realized! So I’m just focusing on care work first. Fortunately, the first DisCO implementation, the Guerrilla Media Collective (GMC), has been dogfooding the framework since 2018 (around when SourceCred started actually) and has created a wiki with extensive documentation of their implementation.


In this post, I use the GMC implementation to illustrate the core DisCO concepts as they relate to care work, comparing and contrasting to how a SourceCred implementation could look. Note: this is highly speculative!

First impressions

Nearly everything in the framework feels solid and value-aligned. It’s what you’d expect from a bunch of decentralized governance experts. It’s also fairly process-heavy. This makes sense for GMC, but may be a bit unwieldy for SourceCred right now. There are also a lot of mechanisms around valuing work that are made redundant by the SourceCred algorithm. Fortunately, I think DisCOs compliment our existing systems and Sociocracy quite nicely. The framework is fairly modular, and SourceCred could use the entire model, or just integrate additional modules as needed. In particular, I think DisCOs have promising solutions in several key areas: valuing care work, decentralized decision making, structuring equity and cashflows, and mutual support (onboarding, accountability, etc.). There is also the exciting possibility that SourceCred could plug into existing DisCOs, automating a lot of their process around valuing work, without (hopefully) changing the model too much :exploding_head: But in this post I focus primarily on care work, as I think that is our most pressing need :heart: .

What is Care Work?

In the GMC, which is a translation agency, work is divided into “productive work” and “care work”. Productive work generally means work billed to external clients (i.e. paid translation work). Care work is everything that leads directly or indirectly to paid work, including community building, emotional labor, logistics, conflict resolution and other forms of often invisible labor. DisCOs put “care work at the core”, seeing it as the foundation of all other work (see feminist economics).

In DisCOs, care work is divided into two types:

  • Care for the health of the collective: The organization is seen as an entity itself, with a “spirit” that must be cared for. In GMC’s case, this includes business development (e.g. following up on leads for translation work), attending to clients, financials (accounting, paying bills, etc.), following through on committments…It also includes internal maintenance of the DisCO, including changes to the org structure (e.g. voting in new members, tweaking governance processes). Everyone has access to this type of work (or at least training), ensuring that it is distributed throughout the organization. This promotes decentralization and avoids creation of an overpaid “coordinator class” or underpaid “administrator class”.
  • Care for the individuals within the collective: The organization seeks to build trust and intimacy among all members. This promotes a “collaborative culture” that enables consent-based, non-hierarchical coordination. GMC does this primarily by mentoring and mutual support structures. Mentoring is bi-directional (both parties learn from each other) and is available to every “committed” member (committed members are roughly equivalent to our “core” contributors). Every member has a specific person supporting them, and every member supports another. This gives everyone a safe space to express themselves and be heard within the group. Conflict resolution (more on that here) is also handled through the mutual support system, ensuring care work is distributed. Who supports who is listed on their public wiki.

From my personal experience in SourceCred, 1-on-1s seem to be the safest contrainer to get really honest and be seen, so was happy to see it structurally embedded in the model. We’re already doing something similar it seems, e.g. with @joiecousins getting Grain for work supporting @LB (and Cred too now??). I’m currently talking with @Jolie_Ze about regular weekly checkins that resemble bi-directional mentoring/support. DisCOs also suggest emotional checkins at “working circle” meetings (more on this below). Which seem to mirror what we already do in many meetings. They use community rhythms to align on goals, which looks a lot like Agile development. Curious to hear any thoughts from the Community Cultivation branch, as many of these ideas seem similar to what we’re exploring. I also like that “admin” work is spread around. We’re already seeing contributors in “roles” owning such tasks express a desire to have help. This also enables decentralization, in particular with infrastructure.

More detail on care work can be found in the GMC wiki page on care work.

Valuing care work

Due to its subjective nature, care work is valued in hours. This raises two immediate objections:

  • How do we know the contributor is working the number of hours they say?
  • How do we know if the contributor is proficient enough to bill for this type of work?

These concerns can only be overcome, DisCOs say, by continually building trust. In the GMC, only committed members (those that have gone through the “dating phase” and been voted in; this is similar to our “core contributors”) can bill for their care work hours. This work is typically stewarded in working circles, which are subgroups that map to domains of work; each working circle has a steward (or co-stewards). This seems to map roughly to our “leads”, or leaders and delegates in Sociocracy. There are also quarterly “credit retrospectives”, where the group re-evaluates the per-hour rate of care work (more on this below), who is qualified to bill for care work, the amount of care work being done (or not done) by individual members, and whether certain time based-tasks should become “credit-value” tasks (i.e. billed per task vs hourly). In SourceCred, one can imagine time-based tasks transitioning to Cred-minting tasks as we create models for them (e.g. we create a meeting plugin that mints X Cred to meeting paricipants). Credit retrospectives are also when also when GMC contributors perform self assessments and peer assessments, but that could be a whole other post :sweat:

Hourly rate

GMC calculates the hourly rate for care work so that it has “fair equivalence” with other types of paid work (translation, editing, etc.). More on this calculation can be found here, but essentially, they estimate an average hourly rate for translators (who get paid by the word), and set the care work rate equal to that. Translators getting paid by the word average 25 /hr, so care work pays 25/hr.

@Jolie_Ze has created a Google Sheet where SC contributors are encouraged to log their hours. Not created for DisCOs per se, but could be a useful data source should we estimate an hourly rate in any system.

Open value accounting (payout formula)

This likewise could be a whole other post, but I’ll do my best to give an overview of the model and how it relates to care work.

Every month, GMC distributes all “available liquidity” (money in the bank after expenses are paid) according to “shares”, which operate basically like shares in a company. Instead of salaries, members receive “dividends” based on the number of shares they own. This enables members to work as little or as much as they want. This is basically how most contributors are paid in SourceCred, except contributors are paid proportional to Cred instead of shares–unless they have a sponsorship (which is more like a salary). In the DisCO documentation, the term shares is often used interchangeably with ‘credits’. I’ll just use shares here, to avoid confusion with Cred.

There are two types of shares in GMC, “livelihood” shares and “love work” shares. Livelihood shares are earned via paid translation work. Love shares are earned from unpaid translation work that furthers the org’s mission and contributes to the “knowledge commons”. The monthly payout is split 75/25, with 75% going to livelihood shares and 25% going to love work shares. This means that if, for example, the collective begins doing more love work than livelihood work, the “dividend” (payout) per love share would decrease, and the dividend to livelihood shares would increase, increasing the financial incentive to do livelihood work. This seems similar to budgets in companies, allowing GMC to steer resources at a high level of abstraction. Cred scores work differently, in that a contribution’s Cred score “collapses” all distinctions down to a single score. However, we’re already starting to differentiate Cred in various ways. For instance, using the Cred budget feature, we place limits on how much Cred is minted on different platforms, which roughly translates to work types (e.g. GitHub is dev work). Likewise with custom weights. For instance emoji reactions in the #props channel on Discord mint more Cred than emoji reactions in #random. As we build out our governance, and become more structured, I could imagine a distinction between Cred that clearly furthers the project’s goals, and side project that may or may not “move the needle”–though since Cred is retrospective, “side projects” may get revalued as more important later on.

Care work "tax"

In GMC’s “startup phase”, care work was paid out separately from livelihood and love shares, using seed funding. Now that GMC has entered the “stable phase”, care work is rewarded using a “dynamic redistribution model”. Similar to a tax, members who perform less care work than others have a portion of their dividends diverted to those that perform more care work. This feels similar to our Balanced policy payout formula, which imagines a “perfect world” where all Cred pays the same amount of Grain, and contributors that have been historically overpaid have some of their weekly payout distributed to those that have been historically underpaid. In DisCOs, the formula imagines a perfect world where everyone does their fair share of care work. Specifically, everyone is expected to earn the same percentage of their total income from care work. If you are below this percentage, the DisCO calculates the care work hours you “should have” worked to meet the percentage, and the dollar value of that (care work hours x hourly rate). It then redistributes that amount to those that did more care work, in proportion to how over the percentage they are.

The following example provided in the DisCO Elements guide shows how a monthly payout would be distributed in a simplified DisCO with only three members. The below graphic shows how $10,000 (the monthly payout) would be distributed before the care work “tax” is applied. Note that if all three members perform an equal number of care hours, no adjustment would be needed. This would be the final payout.

But, of course, some members have logged more care work hours than others.

The below graphic shows how the tax would be applied. Note that, in this example, the percentage of care hours everyone “should have” worked is set to the highest percentage (Joaninha, who recieved 7% of their income from care work). In SourceCred, we have contributors that focus primarily on care work. So it would probably be unreasonable to benchmark to them. It might make more sense to benchmark to average care work hours. Or simply target a percentage we think would be optimal.

This can be a bit confusing, so I made a spreadsheet to wrap my head around it. It lets you play with the dividend ratios, hourly rate for care work, etc., and see how it affects payouts.


The more I delve into DisCOs, the more excited I am. The theory is solid. They’re drawing on lots of shared wisdom from existing traditions. They’re also realistic about the messy realities of utopian visions such as SourceCred’s, and have the scars to prove it.

The thing that excites me most is that they offer a fresh perspective on building trust. Which I’ve come to believe from my experience in DAOs and researching them, is the primary bottleneck to meaningful decentralization. Most DAOs are designed for low-trust environments, which allows for permissionless and pseudonymous contribution. This is a huge advancement. It offers many new affordances. However, designing for low-trust environments often creates barriers to building trust between contributors. This leads to low-bandwidth communication. Which can, at worst, lead to many of the things DAOs aim to disrupt: panopticons, hidden unaccountable hierarchies, abuse, and more centralization than the companies they aim to displace. DisCOs, by putting “care work at the core”, structurally enable high-bandwidth communication. Which could lead to a resilient “web of trust”. This web of trust could be a solid enough foundation to build truly decentralized systems. Or that is the hope. Still licking wounds from my past traumas and always wary. But in my head all the right ideas are synthesizing…:exploding_head:

Phew! OK, if you’ve made it all the way to the end, curious to hear any thoughts or critiques!

Below are some more resources on DisCOs if you want to delve deeper :point_down:



Quick note: it’s actually @joiecousins who is getting the supporting-me Grain. :slight_smile:

1 Like

Oops! Fixed.