Minimizing biases during proposal execution

Source: Midjourney

Summary:

• This week at My First DAO our community is voting on which members they want to elect for the DAO Council.
• Engaging the community, seeking diverse perspectives, and continuously improving voting mechanisms are essential steps in creating a more equitable and unbiased decision-making environment within a DAO.
• We present some of the most common cognitive biases associated with voting
• We will walk you through how we’ve minimized bias and what others factors arose during the voting period and how we addressed them.
• We also highlight a case where cognitive biases impacted the outcomes of a DAO vote

How we went about it (MFD perspective):

In our quest to establish a strong and inclusive DAO Council for My First DAO, we launched an application process that surpassed our expectations. It’s evident that the interest in leadership roles within our DAO exceeds the industry average; approximately 10% of our Discord members applied compared to an average of

Application and Qualification Process: To kick off the selection process, our team devised an internal rubric to evaluate the applicants fairly. The aim was to recruit 3–5 council members, while ensuring a balanced mix of experience levels. We set a requirement for at least two members to have prior DAO experience, while at least two others should be newcomers possessing valuable skills beneficial to our community.

Our overall rubric was as simple as:

1. Verified Profiles: We requested applicants to provide their Twitter or LinkedIn handles, allowing us to confirm that each applicant was a real person, not an impersonator or a duplicate account. It allowed us to establish a credible and accountable council, instilling confidence in our community that the chosen members truly represent their interests and aspirations.
2. Thoughtful Responses: In designing our application process, we opted for simplicity, focusing on asking the most pertinent questions rather than overwhelming applicants with numerous queries. We believed that the right questions would help gauge applicants’ genuine interest in our DAO.
   
   Here’s a question we asked in the application:
   Please describe why you should be considered for this role and include relevant information such as your background, skills, passion for DAOs, and ideas for contributing to My First DAO. By giving applicants the opportunity to express their ideas and vision, we were better equipped to identify the most promising candidates for our DAO Council.
3. Community Engagement: Active participation within the community was another essential criterion, reflecting their dedication to our DAO’s growth and development.
4. Diverse Expertise: To build a well-rounded council, we sought individuals with diverse skill sets, ranging from beginners to web3 enthusiasts, capable of enriching our community.

Inclusive Considerations: At the core of our DAO’s values is inclusivity. To achieve this, we ensured there was a diverse pool of candidates from various backgrounds.

The Final Selection Process: After careful evaluation, we narrowed down the pool to 10 candidates. To ensure a fair community voting process, we will leverage Upstream’s voting feature. Each finalist was asked to prepare a concise 3 to 5 sentence pitch, which community members will use to make their final selections when they vote. We will not be revealing anyone’s name or social media handle. This is a simple approach that can significantly minimize potential biases and allows our community to play an active role in shaping the future of our DAO.

Addressing the main causes of biases:

Problem #1: Influence of reputable or familiar names. As humans, it’s natural for us to gravitate towards voting based on familiarity or popularity. However, this tendency can introduce a degree of bias, as recognizing a name or person doesn’t necessarily equate to the best candidate for the DAO.

Solution: The concise 3 to 5 sentence pitch we asked our finalists to prepare is what community members will use to make their selections when they vote. We won’t include any names or social media handles. A blind vote addresses the social media handle Recognition also could lead to a bias of similarity if the person in question is directly relatable to the person voting — hence a blind vote addresses that

Problem #2: Token weighing. If someone has many DAOPhin NFTs a small group of voters can dominate the vote.

Solution: Instead of one vote per DAOphin (Token) every member gets one vote regardless of their # of DAOphins. This method ensures that those with larger contributions to the DAO don’t have a greater say in the voting process. Token weighting should be balanced to avoid centralization of power.

Problem #3: We overlooked the finalist submission process, and unfortunately, encountered an unforeseen issue with our private finalist Discord channel. Regrettably, one of the candidates got blocked by the bots due to excessive message editing. This unexpected hiccup introduced unnecessary friction into the final submission process, potentially impacting the fairness of the vote.

Proposed Solution: In hindsight, we recognize the importance of accounting for such factors beforehand. To address this, we propose a solution that involves assigning moderators to collect information and ensure anonymization of the facts

Problem #4: We unfortunately overlooked the influence of framing effect bias in our process. To mitigate this bias, a more effective approach would have been to exclude the names of specific projects associated with the finalists. This is crucial because the presentation of information can significantly impact how it is perceived by individuals.

Here’s an example:
Submission by finalist: I am the Founder of Constitution DAO and I want to help the community grow.

Edited submission: I am a Founder of a DAO that raised 43 million dollars and I want to help the community grow.

Proposed Solution: Before publicly announcing the finalist have the moderators anonymize the names and filtering finalist responses based on objective facts, we could have ensured a fairer and more impartial evaluation.

Part II: Validation from an outside perspective

As we’ve covered in prior posts, determining who can vote in DAO is important, but as discussed above, equally important is how a vote happens. Most voters, whether consciously or unconsciously, have inherent biases and blindspots. Some are innocent and do not greatly impact outcomes, but several do. Among the long list of cognitive biases, the following tend to be associated with voting biases and can be particularly problematic:

• Confirmation Bias — focusing only on the information that confirms what you already believe
• Coverage Bias — bias driven by the focus on selected topics relative to many others
• Concision Bias — losing nuance
• Authority Bias — the likelihood of being influenced by statements stemming from individuals with authority
• Dunning-Kruger Effect — “the less you know, the more confident you are”
• Availability Cascade — bias where something becomes more valid the more it circulates
• Halo effect — when our perception of someone is influenced by our impression of them
• Declinism — longing for a past perceived to be superior than the present
• Framing effect — perceiving information differently depending on the way in which it is presented
• Groupthink — following the herd mentality in order to not stand out
• False consensus — overestimating how many people actually agree on a particular viewpoint

For our purposes, given that our DAO is still relatively young, some of the above will not have as much influence in our upcoming vote for the DAO Council. In particular, coverage bias, the halo effect, declinism, groupthink, and false consensus should be reduced by our newness.

Further, as we are purposely not enabling candidates to campaign and we will present the 2–3 sentence pitch in an anonymized fashion, we aim to further reduce the impacts associated with concision bias, availability cascade, and framing effect. While not perfect, by presenting each candidate with the same conditions for lobbying for votes, we are at least providing an equitable starting point for all.

Confirmation bias and the Dunning-Kruger Effect, however, are much harder to minimize. We all have histories that we brought with us when we joined the DAO. It is not realistic to think that the participants of the DAO arrived as blank slates — we all arrived having lived lives and accumulated knowledge that influence how we think. In some cases, we may have acquired a habit to “keep doing what works” that was reinforced by seeking information that validates that view — confirmation bias in action. Additionally, that knowledge with which we arrived, when remotely applicable to a topic presented in the DAO, could easily feed into a Dunning-Kruger Effect. The most that we can do under these situations is reinforce to community members to keep an open mind, seek contradictory views, and recognize that we are all on a journey of education where no one person has all the necessary knowledge.

Lastly, authority bias can present itself in one of two ways, one of which is easier for us to mitigate and another which is harder. The first way in which it can manifest is an inadvertent “endorsement” of a candidate by someone with established authority. This situation could occur if a founder, team member, or moderator of My First DAO speaks positively of a candidate or outright endorses them. In this case, the community would treat it as an endorsement and would be subjected to a bias that benefits a candidate, or candidates, over others. This not only influences the vote, but places “unendorsed” candidates at a disadvantage despite previously cited efforts to ensure as equitable a starting point as possible for all. This is easier to mitigate if the team adopts a careful stance in its communication around the candidates.

The other manner in which authority bias can present itself is through the way that candidates have been presenting themselves in the DAO’s Discord prior to submitting their candidacy and consequently being selected as finalists. Over time, the way the candidates communicate and the experience that they cite, could establish them as authority figures within the DAO. In isolation, this is a benefit for the DAO as we benefit from their talents and they help form a community. The reason this can be a problem is that their experience may be different than that which is required by the DAO Council and their perceived authority in one area might not be beneficial. Our approach to present candidates in an anonymized manner, in theory, helps mitigate this risk, but a careful voter might discern the candidate’s identity and disclose to the rest of the DAO. Even if each candidate were presented as a different anonymized number for each voter, it would only address half the risk associated with the bias.

In a DAO where we have been previously affiliated, the establishment of a DAO Council-equivalent was done through an open, token-weighted vote. In this instance, where there were several hundred potential voters, most did not have full visibility into the actions of the candidates within their working groups. As you might imagine, voting in this situation might be remedied by campaigning, but a DAO does not lend itself to that approach. The end result was that the vote devolved into a popularity contest driven in part by name recognition. Here, voters recognized certain names from being active members throughout the Discord and other communication channels. This touched on several of the aforementioned biases including coverage bias, availability cascade, and halo effect.

In that instance, we then layered on the fact that many of the then-candidates were also leaders of the working groups, and we see that the authority bias was also at play. The most pernicious problem, however, turned out to be the framing effect. What wound up happening in this DAO was that certain individuals were adept at presenting themselves as altruistic individuals who had the community’s best interests at heart when nothing could have been further from the truth. It was known to other senior leaders that some individuals were either selfish, difficult to work with, or intellectually inconsistent in their approaches; none of which are qualities that would serve the best interests of the community.

A key mistake, in retrospect, was that the transparency espoused by the DAO was not sufficiently transparent to give all members visibility into the true nature of these individuals. That is why la.monirojas24@gmail.com and I are producing this series — as a way to provide additional transparency to the community. The other crucial mistake was that the DAO Council did not have a mechanism through which the community, or the majority of the Council, could remove a bad actor or one who was acting counter to the DAO’s objectives. We will cover this topic in next week’s post.

Conclusion:

This week at My First DAO, the our community is voting on which members they want to elect for the DAO Council. Engaging the community, seeking diverse perspectives, and continuously improving voting mechanisms are essential steps in creating a more equitable and unbiased decision-making environment within a DAO. In order to mitigate the risks of bias, a careful understanding of the most common biases is necessary in order to plan accordingly. There is no perfect solution, because we as humans are imperfect, but there are many opportunities to learn from the mistakes of others in order to maximize My First DAO’s chance of success, achieved through the mitigation of common biases.

Disclaimer: The above represent the personal opinions of the authors and should not be ascribed to any affiliations thereof. Further, the above does not constitute a recommendation or solicitation to purchase or sell any assets that may be referenced herein. As of the time of this writing, the authors may or may not have positions in any above mentioned assets.

This article was collaboratively written by Eduardo Abreu and Monica Rojas