Camille (Apiary)

Task

expert interview

Status

Completed

Date

Feb 28, 2024

Completion Time

Interviewer

Segment

consultant

User Details

Camille Canon's role is founder of Apiary.

Interviewer Observations

The conversation revolved around the intersections of ???, technology, and governance. Camille shared her passion for nature and how it influences her work, while Artem discussed his involvement in a community focused on web3 governance. Camille emphasized the need to rethink governance systems in light of the attention economy, while Artem offered alternative solutions such as sortition and AI. The speakers debated the potential of these approaches and the importance of ensuring that any new system provides sufficient information and fidelity for good decision-making.

Insights

Need

The client expressed interest in tools for aggregating input from many members quickly.

Pain points

Camille identified low participation and engagement as a pain point.

Camille identified the time required for proper deliberation as a pain point.

Outline

Governance in web3.

Speaker 1 emphasizes the importance of understanding the context and nuances of a particular system when consulting on technology-enabled projects.

Speaker 3 seeks advice on how to build a product that is different from existing solutions, while also acknowledging the limitations of sharing information without proper context.

Using AI to facilitate group decision-making.

Artem discusses using large language models like GPT to facilitate deliberation and sense-making in web3 communities.

Facilitating group conversations and sharing perspectives.

Camille was co-founder of Purpose and that merged with investment fund, to facilitate organizational change through stakeholder inclusive models.

Facilitator highlights importance of being heard in group conversations.

Group decision-making and technology.

Speaker 1 wants to know if an AI bot can facilitate a group conversation while maintaining individual trust and safety.

Speaker 1 is interested in exploring how to navigate the relationship between individual and group conversations, ensuring that everyone feels heard and their opinions matter.

Speaker 1 argues that LLMs are often misunderstood as being able to aggregate individual contributions and create group sense making, but this is a falsehood.

The actual process of group decision-making often involves extreme ideas and outliers, rather than the middle ground or consensus, and algorithms may actually hinder this process.

Speaker 1 highlights the importance of recognizing people's agency and decision-making abilities, rather than treating them as mere inputs.

Speaker 1 also cautions against the assumption that making decisions faster and more efficiently will lead to better outcomes, as this can result in a delayed cost to the system and a lack of engagement from stakeholders.

Designing governance systems for intervention.

Speaker 1 proposes designing systems for intervention rather than participation, citing Nathan Schneider's book on governing spaces.

Speaker 1 discusses the importance of structuring organizations to make localized decisions with localized information, ensuring that token holders understand how resources are being used and that decisions align with objectives.

Speaker 1 emphasizes the need for guardrails or systems to detect and address potential threats to objectives, such as governance proposals that could undermine them.

Governance and decision-making in the context of blockchain technology.

Speaker 1: Defining purpose and success criteria is crucial for decision making in decentralized organizations.

Speaker 1 and Speaker 3 discuss the challenges of creating a normative culture around intervention and decision making, including the limitations of knowledge management and the need for success criteria.

Speaker 3 expresses interest in turning research into prototypes or solutions, while Speaker 1 is noncommittal.

The speaker is reevaluating their focus on tech development due to concerns about replicating known problems in governance and the need for more resources to support innovation.

Leveraging technology for democratic decision-making.

Speaker 3 mentions several organizations focused on building tooling for web3, including Metagov, Changemakers Accelerator in Berlin, and a startup studio in Switzerland called Octree.

Speaker 1 expresses skepticism towards current approaches to democratic decision-making, citing a lack of scaled resources and innovative ideas.

Speaker 3 agrees, noting that participation-based metrics may not lead to meaningful democratic outcomes, and that citizen assemblies have not been successful in the past.

Decision-making in digital spaces.

Speaker 1: Interventions needed to shift paradigm on participation in digital spaces.

Speakers discuss using AI or sortition (random selection) to address attention in decision-making, with potential pros and cons.

Speaker 1 highlights the importance of considering the long-term viability of decisions made in a group context, as optimizing for information and efficiency can lead to unintended consequences.

Using AI to improve group decision-making.

Speaker 1 expresses concern about the lack of human insight in AI-driven decision-making, citing the importance of understanding human complexity and chaos in group decision-making.

Speaker 1 suggests that builders of AI systems may not fully grasp the nuances of human decision-making in practice, which could lead to unintended consequences.

Speaker 3 suggests using AI to quickly gather inputs from a large group of people, rather than relying on traditional deliberation methods.

Speaker 1 agrees, highlighting the potential for AI to facilitate ideation and collaboration, rather than just decision making.

Speaker 3 shares their insights on how to make workshops more engaging, including using murals and inviting participants to add stickers to a board.

Notes

web3 as a petri dish

tooling is ???

success is not about information

before Apiary I started Purpose (now merged with VC fund) — legal frameworks for ??? (ownership transition)

ripple effect of org change that occurs as a result

became a good facilitator inadvertently because things I wanted to achieve…

you can’t get people to arrive at conclusions faster unless you create that space for facilitation

investigate what exactly facilitation means?

sensemaking = information sharing — people are hearing other people say things they didn’t hear or understand before (e.g. direction for org to go)

sensemaking = individual — “I’ve been carrying around this weight because of information flows, etc. I didn’t have a place to share this” (psychological ??? of relief), probably the most important part of the whole thing

I can’t share my opinion (defensive) → collaborative ???, deeply human thing, probably not possible with a bot?

they say the same thing: “thank you so much for listening to me” RELIEF

the problem with trust — we’re really clear what the communication boundary lines are, all of this anonymous, rules are upfront, the feeling of psychological safety

can you have the info feed into LLM to create sensemaking, and simultaneously make the people feel trusted

psychological relationship of individual speaking to the bot might be… something I would PAUSE on

how you go from individual safety and trust and feeling heard to group?

individual can hear and share info…

then you go into group convo, two things

everyone participating needs to feel safe and secure

you need to navigate…

…

… LLMs think they can help with group sensemaking

one of the big mistakes is thinking there’s conversion between individual contribution that can be summarized into group thinking

really counter intuitive: if everyone say the same then we can… that nets out as a reflection of … but it doesn’t work like this!

with LLMs it’s often not what’s in the middle of the group what is actually the point of most interest and can help it move forward… it’s often outliers!

we ironically get further and further away from the actual truth of as a whole group can

there’s a fundamental tension that

one of the things that’s important to hold and this has been the evolution in web3 (Vitalik posting I’m no longer a child) is shift from people are inputs to people make decisions in our bodies than just inputs of information

the other point is around attention — how much of it people want to spend on a given topic, people ???, not sure it’s true in practice, sometimes the desire to move faster leads to delayed costs in the system (6 months later they ask why TF we’re doing it this way)

people were going too fast without explaining enough to make the decisions in the right way with context

if I was a Governing God, I would design a system based not on participation, but intervention… how do we allow people to intervene when something is not going well

the question depends on the context… if intervention is the primary metric, two things are needed:

information flow: how do you structure the org or teams so they can make localized decisions with localized info, and that feeds into the systems, so token holders can understand how resources are being used, if teams are on track, etc

localized power, enabling people to act autonomously

doing sensemaking well, and then ensuring that information that’s necessary for the objectives of the whole are fed into the whole

I think you need to figure out what is actually the field of what needs to be governed, what decisions have to be made for the system to function well

and then when decisions need to be made, how ??? it should be, looking on macro level, what would danger and success would look like, what do we want the system to achieve

…

you can’t hold people accountable for the decision made (in DAOs)

why are we collaborating, what does success look like and what we’re trying to achieve

Action Items