Higher-order fractal

in #governance6 months ago (edited)

With more and more fractal-like DAOs springing up, I think it is time to start thinking about how these fractals could interact and organize among each other. Here I will propose one obvious model, which uses a concept of what I think could be called “a higher-order fractal”.


In mathematics, a higher-order function is a function that either takes functions as arguments or returns a function. In other words - it is a function that deals with functions instead of just variables. In the same vein, a higher-order fractal would be a fractal composed of fractals or other kinds of DAOs instead of individuals.

This creates a fractal pattern, so obviously, it does seem like a natural solution for organizations called “fractals”, but it does raise some critical questions about how meetings and consensus process of this higher-order fractal would work.

In summary, I suggest that higher-order fractal can simply use the same Respect game and top-contributor-based consensus process as some existing fractals use, except the people participating in meetings would be representatives of DAOs (1 representative per DAO) and Respect would be awarded to a whole DAO. Consensus meetings of higher-order fractal should happen less often and there might be some additional small changes needed. I try to cover those in the "Design" section below. But first, let’s talk about why we might want this.


Respect game at the core of a fractal process, is a very straightforward way to create a common ground for all the DAOs, without much organizational complexity (which would be hard considering we are talking about DAOs organizing not individuals). This common ground can have the following benefits.

Means to measure performance of different consensus processes / governance solutions

We have already noticed that each new fractal often mutates a little in its processes. Some fractals make bigger changes, some smaller, but creators of a new fractal almost always put their twist on it. Plus you have kindred DAOs like Upscale (former Eden) and other DAO ecosystems which share a lot of common values with fractals (like Hypha DAOs).

All this variety is very healthy for the ecosystem. We don’t know which processes are the best, but through this variety of experiments, we create data that can be used to compare all of these different governance workflows.

This is where this higher-order fractal can be invaluable. One periodic meeting where each DAO presents their progress and then reaches consensus on rankings among themselves can create a very crucial data point for comparison of DAOs and their processes. Not to mention that rankings and Respect distribution that these meetings will output, will by themselves be a very interesting and telling metric.

Contributor presentations will get a lot more interesting and valuable

It’s one thing to listen to a presentation of what an individual did throughout one week, it’s whole another thing to listen to what a group of people did in a month or a couple of months. While the former typically just talks about an in-progress work being done, the latter can become a demo of an actual product of that work. This will provide useful content for what these communities are achieving that is meaningful to an outsider and not just old participants.

It is also easier to evaluate and compare products of work rather than promises of effort being spent. By extending time between meetings and allowing groups of people to be represented we will move towards presentations of products rather than effort.

Opportunity for smaller DAOs to get seen, get reputation, and attract more people and funding

Higher-order fractal creates an opportunity for positive collaborations of people to get noticed and get reputation. Reputation can lead to other great things for a DAO like new members and potential funding.

Competition and collaboration between DAOs

This I think can be a motivating factor for contributors. Higher-order fractal meetings will add more meaning to work performed by individuals in the sub-DAOs because it will add up to something bigger that will be recognized and evaluated at the end of the period.

It will also create opportunities for DAOs to learn more about each other and collaborate.


The core of higher-order fractal would be Respect game like in the typical fractals we know of. But instead of individuals competing for Respect we want to have fractals. For this we need fractals to be able to:

  1. Present contributions;
  2. Build consensus with other fractals on the rankings of contributor fractals;

The simplest solution is to have each fractal elect a representative. This I think should work perfectly for solving the first (presenting contributions). It would be individuals presenting contributions but they would be presenting contributions of a whole fractal instead of just theirs.

For the second, I think representative is not enough because, while you can prepare a presentation that presents the work of the whole fractal, you can’t really prepare for a consensus process on rankings where you would represent your whole fractal (its opinion on the rankings). Therefore, I think that after all participant DAOs present their contributions, they should be given at least a week to build consensus on their rankings. This will provide time for fractals to discuss the performance of each contributor fractal internally, within their own fractal. The final consensus would still have to be signalled by representatives of each DAO, so that there’s a clear point of contact from each fractal which allows consensus building among fractals.

So from a representative's perspective, consensus building could be a 3 step process:

  1. Poll his fractal for their opinion of how all the contributor fractals should be ranked;
  2. Negotiate and build consensus with representatives of other fractals (who are presumably doing the 1st step as well);
  3. Signal consensus pre-agreed among representatives;

If representatives do not properly represent their fractal (do not do the first step for example), it is up to that fractal to solve that issue internally and elect a better representative for the next higher-order fractal meeting.

There are some other changes and considerations to the fractal process we might want to take into account.

More time in between meetings

Considering that the overhead required to reach consensus (even through representatives) is bigger in the higher-order fractal, meetings should happen less often than a typical 1 week. Plus each participant DAO usually has its own weekly meetings as well. At least 1 month in between higher-order fractal meetings would be more appropriate but an even bigger gap might be sensible as well. This will give more time for fractals to make something happen, which will make their contributions more interesting.

More time for the presentation of each DAO

Since each representative will be presenting contributions of the whole group, over a period longer than 1 week, I think they should be given more time to present. Currently, I think the typical time for each to present ranges between 3-5 minutes. For higher-order fractal, 10 minutes might be more appropriate. The time given to break-out groups should be extended as needed.

Any kind of DAO should be allowed to participate even if it does not follow a fractally-like process

As long as they present a required interface (i.e.: elect representatives) I don’t see a reason for higher-order fractal to not include them. A variety of different competing consensus processes is good for the ecosystem (see Motivation section).

Furthermore, I would consider allowing 1 person DAOs to participate. This can be useful when a person is starting a project and has plans to accept community contributions to his project but has none yet. Then it makes sense for him to create a fractal that would eventually attract contributors. Meanwhile, he would still have a place to present his contributions regarding the project he wants to work on in higher-order fractal meetings. However, it is important that each DAO should be evaluated by the same standards regardless of its size. So 1 person DAO should normally not be able to compete with larger DAOs (assuming those larger DAOs are doing useful work).

Participants of a higher-order DAO should try to evaluate its sub-DAOs primarily based on their output and not their internal structure. This respects the autonomy of each DAO, which I think is necessary (considering we are talking about decentralized autonomous organizations). This is kind of like having some base respect for the privacy of another person’s mind (their beliefs and opinions) as long as the person acts in a socially acceptable manner.

I would make a proposition that bad governance processes would always eventually reflect in the outputs of a DAO. For example, if some sub-DAOs have a governance process that does not respect their members, eventually those members will leave and join other DAOs, which will eventually reflect in the output of that DAO and their rankings in higher-order fractal.

Higher-order fractal needs a consensus process

Since we are giving more time for presentations (so meetings will take longer), and consensus building among fractals will be slightly more complex, it might make sense to be more careful regarding who participates. Higher-order fractal needs a consensus process for adding (or at least removing) participating DAOs if nothing else. Another reason a consensus process might be needed is the scheduling of meetings.

I don’t think Eden+Fractal is suitable here because it adds another consensus problem (electing a delegate fractal) and as already mentioned these consensus problems cost even more in higher-order fractal. Therefore, a top-contributor-based process[1] would probably be more suitable. This means, however, that we will have to agree on a way by which each of the top fractals signal their approval of a proposal. Again, the question of what mechanism is used for participant DAO to decide to issue this signal is up to each participant DAO to determine.

Should people be allowed to participate (contribute to) multiple participant DAOs?

Generally when you have teams competing you don’t allow members of teams to overlap. So it would be natural to expect that being a member of multiple participant DAOs should not be allowed by the higher-order fractal. Plus competition might be more interesting this way. However, actually enforcing this would add a lot of complexity to the system and I would argue that we don’t have much to lose by allowing this.

The complexity that enforcement of this rule would add would be a bad kind of complexity where higher-order fractal would have to influence the inner structure (membership list) of a sub-DAO (which means infringement on the autonomy of a DAO). Not to mention that any complexity in higher-order DAO is very costly since changing any of the rules will require consensus from all participant DAOs.

One of the things that could happen if we allow participation in multiple sub-DAOs is that some contribution of a single individual ends up being claimed by multiple sub-DAOs. But then it simply means that these contributions “cancel out” when comparing these two DAOs.

If some contributor is contributing different things to different DAOs (helping them both), then it means that he is splitting his time and that reflects a certain amount of respect for both DAOs. It means that both DAOs are doing something right because they are attracting contributors.

Another argument against allowing participating in multiple sub-DAOs is that it would allow people to play dirty by sabotaging competitor DAOs from within. But this can be ascribed to be a problem of individual sub-DAOs as opposed to a problem of higher-order fractal. There are multiple things a DAO could do to protect from sabotage at a local level:

  • Transparency regarding who the members are - if members of a DAO are not anonymous then they would have to risk their reputation by trying to sabotage;
  • Review process for contributions (which is already being done to some extent in fractally-like DAOs) could catch any malicious contributions;
  • A good DAO should allow space for members to question the honesty of other members;
  • A DAO could (and probably should) have a mechanism for kicking out members who are malicious (detected to be sabotaging);
  • Simply not allowing people to participate if there's evidence of them participating in other DAOs;

It’s up to each DAO to decide what methods they apply. If a participant DAO fails to protect itself then it is useful information when evaluating its process and comparing it with other DAOs. Plus a higher-order fractal will not be able to protect from all cases of malicious behaviour. So DAOs will need to think about protection mechanisms for themselves anyway and I think it will be more useful for higher-order fractal to evaluate those mechanisms instead of enforcing them.

So my preliminary answer is that higher-order fractal should not have a rule against allowing people to participate in multiple DAOs even if those DAOs are competing in a higher-order fractal.


DAOs / fractals can gain a lot of benefits from organizing into a kind of DAO of DAOs. Here I propose one model for that which I call “higher-order fractal”.

The guiding principle behind higher-order fractal model is that the DAO that encompasses other DAOs should make a conceptual leap where its participants are smaller DAOs instead of individuals. This will give it a legitimate claim to be called DAO of DAOs. Other than that this DAO should mirror the design of a normal DAO composed of individuals as much as possible. This will make the system simpler and composable.

I propose that this DAO of DAOs should mirror the structure of a typical fractal because it seems like the most natural way to create that kind of DAO. Think of it like a periodic conference for the ecosystem of DAOs, except besides just presenting their progress they would also be playing a game to rank each other's progress. Besides providing many other benefits, this simple game enables bootstrapping a whole higher-level governance process without requiring founder DAOs to agree on any sort of complex ruleset upfront.

  1. E.g.: top 6 contributor DAOs over the last N meetings have votes in approving proposals, where you need 4/6 votes in order to pass a proposal.