The Collective DAO Archives: DAO Design Principles | Justine Humenansky - Optimism Foundation

welcome back eth Denver um next up I'm excited to announce uh Justine from optimism um who is going to be giving some interesting discussions uh on some Dow research um and those who don't know obviously you know optimism is running some of the most interesting experiments right now in the Dow space so give it a hand to Justine cool today we're going to go over optimism's dowo design principles so I'm the governance lead at the optimism Foundation one of the primary goals of the optimism Collective is to support open-source public goods so last year at eth Denver we open sourced some of the tools that we use in our Dow research uh via something we call the Collective dowo archives we've added new resources since last year to the archives which we're open sourcing today and those are the dowo blueprints and our dowo design principles so we have to stop saying that we're building a plane while we're flying it uh when we talk about Dows because it's not really true Dows have been around since 2016 and there actually is a blueprint for this uh because I built them so I mapped out the organizational blueprints for 17 Dows this is what gitcoin looks like you can see what roles were hired when which working groups were formed when they were formed how many people were part of those working groups and you can see how this evolved over time these are open source you can go in and look at them and excruciating detail yourself from this madness though comes a method um during by doing this research I derived 11 Dow design principles and we actually use these in practice today at optimism so these are not theoretical we're designing based on these so we're going to go through those principles today these are also open source obviously now the first one is to use magic numbers when you're designing teams I don't mean magic numbers in a mathematical sense so don't at me uh mean magic numbers is a term used to refer to this natural limit um in the number of people that can basically effectively propagate information between them and for teams that tends to be 7 to 14 members when you look at Dows you also see a similar natural limit at around 50 full-time or almost full-time contributors if you put those two data points together uh it would imply that Dows can support roughly five teams if you look at the evolution of dows looking at the blueprints you'll see that's what happens most Dows evolve to a final state of five critical business uh functions the second one is to delegate convex decisions and make concave decisions directly this is not my framework this is vitalic framework for convex versus concave decision making but we use it a lot at optimism so when decisions are convex decentral izing that decision- making process can lead to confusion and lowquality compromise that's when you want to delegate those decisions to a council or a board or a small group of experts when decisions are concave though uh they actually benefit from the wisdom of the crowd and those should still be made by the full Dow we have to distinguish between representatives and contributors I talked about this last year but it's super important so I'm going to say it again representatives are delegated rights that token holders would otherwise directly execute and they should be elected and um treated as public officials held publicly accountable contributors that are doing work for the Dao don't take on token holder rights and therefore don't need to be elected and held to that same level of public scrutiny a lot of AD Dows just elect all roles and we're publicly hiring and firing everybody which is a terrible contributor experience it's only necessary if you're a representative uh number four is we want to try to design for resilience to churn not resistance to it so D like churn is inevitable basically in Dows they're inherently parttime flexible work environments which means they're also going to be high turnover environments so we still want to try to retain our high context delegates but we need to do that within a system that doesn't break if any any one of them leaves some ways that we think about doing that is that team leads should never be a central point of failure Community rol should be spread among multiple contributors and not concentrated in one or two like you can think of moderation or governance facilitators if you only have one or two they have a lot of power in the system uh and then we have to establish Talent pipelines and scalable onboarding flows so that we could replace any Community leader in 7 to 14 days optimism's resignation policy is 7 days so we need to be able to do to do a find a replacement in that period of time five we need to avoid principal agent problems they're everywhere in Dows um the problem with that is that they make organizations work across Us Versus Them lines Dows are already managing tenuous relationships between many different stakeholders this exacerbates that problem we most commonly see this in water fall compensation schemes but you also see it in community-led initiatives that are dependent on core teams or when the core team is required to support Community mandated initiatives uh teams should have as much autonomy over their own operations and budgets as possible six is embarrassing to have to say but we need to make datadriven decisions most Dows are not really doing this um and that can help us make fewer better decision is so we need to reduce the number of subjective decisions voters are asked to make and instead Empower humans to override otherwise algorithmic decisions so what can that look like that can look like calculations that are run over attestation graphs or formulate strategies and then you have Smart defaults that are built on top of that which your voters can override we need to develop more objective Frameworks especially when it comes to economic parameters and that is partly to to reduce this natural self-preservation tendency to crew value for the Dao at the expense of the protocol this happens a lot actually um this principle is the most lacking in treasury allocation strategies that are um most of the time made subjectively so there's no concept of being in a growth phase or value phase or when or how you adjust economic parameters Seven's really important at optimism we want to Value outputs not inputs so one way we do that is to separate decision- making about the work to be done from decision- making about the team that completes it you want to try to avoid batch proposals because that doesn't let you value any specific outputs and then to the extent possible we try to Value work based on the output or the imput uh or the impact and not on hourly or any other type of input the impact profit framework um goes into more detail about this concept and it's a core part of retroactive public goods funding it's linked here if you want to see it number eight is internal redundancy creates waste not competition so you'll see a lot of dows support redundant teams thinking it's going to promote competition and the best approach will kind of rise to the top it's not what happens in practice uh what happens in practice is these groups have overlapping responsibilities and messy handovers and it's basically inefficient and a bad ux it tends to go on forever also because it's super unpopular to have to fire any one of those teams via public governance vote nobody wants to do that what works much better are competitive basically RFP processes we have multiple external teams competing for the ability to complete a piece of work and that's because it creates healthy competition via performance-based accountability scope should be narrow not broad most um working group mandates and Dows are super Broad and the problem with that is that it creates room for inefficient budgeting and also sometimes fraud because the ambiguity reduces accountability um you see sometimes like metal work streams which is a where you have multiple working groups within a larger work stream that represents the function that doesn't usually work because it introduces additional coordination costs when coordination costs are already high in decentralized environments for the same reason you don't usually see cross functional working groups so the thing to remember is to keep Scopes mandates rfps whatever uh is narrowly scoped as possible this one was one of the more interesting findings from the research I did and that was that you want to scale work within the Dow via public goods rather than public services and I'll explain what I mean you can imagine like a data team or data working group that provides data services to other teams within the Dow that suffers from dis economies of scale and also like a A variation of the principal agent problem where the Dow approves the budget for that team but the individual groups consume the services what works a lot better though is if that same data team within the Dow provides like public data infrastructure that the other teams can use to conduct their own analytics so that's what I mean by public goods public infrastructure versus public services this is the last one so we want to try to achieve alignment through action a lot of dows create structures without any consideration for what the Strategic objectives of those structures are and then retroactively try to to align or Justify the Dow's efforts usually attempts at strategic alignment are made via paper policy and the problem with that is it results in bureaucracy not alignment instead we want to try to design systems that um in which merely taking action is inherently aligned so I can it's a uh it's hard to do but I can give an example so paper policy approach would be having everybody grade proposals on a scale of 1 to 10 based on how much they align with a stated goal um a more inherently aligned system would be to allocate each strategic goal uh budget and then have people rank order initiatives under them so just by casting a vote there's no way it can't be strategically aligned so why are we open sourcing uh these and why do the principles matter well because we're actually not just open- sourcing these resources we're open sourcing everything uh and what I mean by everything is the entire governance design process at optimism so this is the first step in a gradual path towards open meta governance right now we're in Phase zero which is to open source the thinking of of the foundation and the design team by sharing resources goals Milestones the collective archives so that we can build shared context with the governance participants fa one which will uh happen shortly will be to establish or formalize a feedback panel of high contacts governance participants so they can start to provide feedback during the design process right now we get feedback after designs are more less complete we will start to have governance participants involved during the design process uh phase two would be that that feedback panel evolves into a contribution path where we where um it's an open process and like kind of training to do metag governance design and then the Final Phase of this is that that contribution path um turns into a concept we're introducing called core delegates and much in the same way that core developers maintain develop and evolve the core protocol core delegates would maintain develop and evolve the core governance system through metag governance proposal rights so there's more details on this path it's going to be experimental like everything we do and um open to your feedback but you can read the full Forum post which is linked here and if you haven't already joined the collective you can register to be a delegate anybody whether or not you're uh involved in our ecosystem can use the collective da archives they're open- Source public good that are free for anybody to use and they were also made using entirely public information h