Eric Alston presents a framework of five recurring 'governance axioms' for protocol design—incomplete rules, governance-as-conflict, mechanical concentration of authority, asymmetric incentives from shared resources, and the transparency-privacy tradeoff—arguing these tensions can never be designed away, only managed through mechanisms like killswitches, veto-gates, and tiered amendment rules. Drawing on his experience advising Cardano's constitution, DAOs, and peace processes, Alston contends that 'incomplete rules' mean norms and discretion inevitably fill gaps left by any formally encoded protocol, and that durable systems require 'dynamic fidelity'—credible commitments combined with constrained capacity for adaptation rather than either permanent rigidity or unchecked flux.
0:23 Val going to try to figure out how to populate the chat. Um, bingo.
0:32 I’ll keep this tiny by the side. So, if anyone has clarifying questions, feel free to jump in. Um, yeah. So, this is just, you know, it it’s kind of a synthesis of a lot of my work in both studying the rules that we use to coordinate our behavior in large complex impersonal social groups. And so experiences ranged from, you know, advising constitutional reform and peace processes to helping to design the constitutional process for Cardano, advising on governance design and a huge range of cryptocurrency related contexts as well as others including chartering one of the world’s first special purpose data trusts. And so a lot of this is based on a ream of academic publications that I may occasionally allude to, but to me this is kind of a intended to be a primer for those interested in designing governance for a particular system. In an increasingly networkcoordinated world, chances are such governance will entail choice of protocols, kind of hardened engineered arguments baked into network rules that participants can typically observe but ideally can’t unilaterally change. More on that in a moment. But to me, I view the mantra of, you know, hey, we’ve awarded several Nobel prizes in economics for the study of institutions. My view is it’s based in a recognition that you can do so much more in context where it’s, you know, commitments are credible. The rules that are coordinating outcomes are credible to all participants and their change is visible to and ideally accessible to all participants within a particular system.
2:29 Often though those ideal types are not realized for a set of reasons I will be getting into throughout the course of this talk. Um, just as a very brief preview, this is kind of a road map and the shortest form articulation of what I call governance axioms. These are things you cannot ignore if you’re designing rules to coordinate large groups of humans at impersonal scale. we have time. We could also get into the chat in terms of what what this might have to do with coordinating agentic behavior because one of my summer research projects was all about how might we deal with pseudonmous agents and emergent harms resultant from the interaction of many many AI agents in a particular networkcoordinated context but TBD if I’ll even get through my talk because this is very much the first outing of this I gave kind of a proteian version of this at Zoo Village, Georgia. If Jesse Catz is in the audience, he came out of it being like, “My brain hurts and I’m not sure what happened.” And so, my goal is to make this much more legible than that three-hour extravaganza of a talk I gave at Zoo Village, Georgia. But to me, this is a pretty common uh hypothetical to start with. Protocol works. So what happens next which is it is coordinating large scales of you know voluntary human activity.
3:57 My contention is is the unforeseen springs eternal meaning we can’t fully program for all downstream outcomes. Some participants in a particular system say we need to pause. Imagine a large-scale hack. Imagine serious government concerns animating an enforcement action against a sense a set of validators. I could spin up, you know, existing examples of this such as the DAO hack as well as other interventions on blockchain networks, but this is a generalizable issue associated with governing complex human social orders. Our foresight is imperfect goes one narrative. So, some participants are like, pause the system. Others are like, “No, let it keep running because it was designed to do exactly that and we need to rely on its finality.” A team of devs proposes a repair and needs authority and funding. And I’m very sympathetic to Trent Vaneps’s, you know, ongoing arguments of in order to provide reliable security services and protocol updates for the Ethereum network, your core devs need a stable basis of funding. So this is a microcosm of that more general problem that Trent has, I think, correctly advocated uh for at some length on Twitter. But this poses the question, who decides? Who gets to decide on what comes next? Who pays the cost of any intervention? And who, if anyone reviews that intervention for consistency with shared principles animating the group being coordinated by that particular set of rules? And so to me this is such a quintessential issue which is governance exists decide in real time especially in the case of unforeseen circumstances. More on that in just a second right here as a matter of fact. And so my Twitter handle my website and my first axiom all are the same which is incomplete rules. And this is sometimes you know put in legal scholarship as all rules are over and underinclusive.
6:14 So that which you specify in advance is actually not the whole order the whole social order that a particular set of rules are are creating. This is Kelenian material constitutionalism in European scholarship. This is big C versus little C constitution in US constitutional scholarship. But all of it is saying and I’ve I’ve expanded on this particular point and its implications for protocol design in a piece entitled norms, institutions and digital veils of uncertainty. Do network protocols need trust anyway? And so there is that which you can specify in advance. If you can think through it with sufficient completeness, with sufficient certainty, you should specify it in advance. But nonetheless, what guides action in times of the manifestation of truly uncertain outcomes for a particular context is norms, a shared sense of purpose, values, all of the things that cannot be perfectly defined in an institution and cannot be perfectly encoded into a given protocol. This is not an argument against the exercise of doing so, which is if you can specify it, you probably should because there’s still going to be a gap. And nonetheless, a determinant response can still be inadequate on numerous margins. Think of, you know, somebody obeying the speed limit but considering ambient conditions or driving very dangerously or someone, you know, racing their pregnant wife to a hospital during a medical emergency. Our norms and values provide input to saying one is technically legal but highly undesirable and the other is arguably defensible despite being technically illegal. So in some automate that which you can specify in advance but governance exists to cover that which remains. Ergo, governance springs eternal is one way to view this. And the intro chapter to a handbook I edited last summer about rule choice under conditions of sufficient complexity, indeterminacy, or uncertainty. The intro chapter basically makes this argument at considerable length. It also is open source and previews all of the other chapters in the book. And for any of this, if you’re like, I’d like to learn more about that, get at me. I’m happy to flog any of my long- winded academic rants um in you know to anyone who’s actually that interested. Second is this is based in a piece I published in the MIT computational law report a few years back after advising numerous DAOs and hearing a kind of consistent refrain which was what’s going on where did we mess up what’s wrong with our governance design because we all came together and we’re like we want to do this cool thing using this new organizational technology that has a variety of affordances that haven’t existed previously when it comes to organiz ganizational design. We all came together and we’re like, “Let’s constitute a DAO. Let’s do the thing.”
9:25 And we’re all like, “Yeah, let’s do that awesome thing.” We’re so excited. And so they constitute the DAO. And then their first stream of governance decisions tend to be conflict and they’re like, “Why is it broken? Why can’t we agree?” And I’m like actually and credit to Michael Zaram of Dynamical Systems Group for this particular analogy. He’s like governance is a dumpster fire. It’s actually working when you know ultimately it keeps the fire within the confines of the dumpster. That means to me that conflict is not actually itself detrimental to any organization. It can be, especially if the organization’s rules or protocol channels conflict in unproductive ways, tends towards deadlock, doesn’t have residual authority specified, etc. And nonetheless though, my argument is often it’s kind of spicy, although I’m a little bit surprised by that, but I’m like actually governing is all about channeling conflict. And this just emerges naturally despite people coming together for a shared purpose. That does not imply agreement about every decision. And it’s structural. This is mechanical. What do I mean by that? If every group member agrees what should be done next, who should do it, what resources should be devoted to doing it, that will emerge endogenously by definition. But the more a particular outcome associated with a group is you know people have good faith disagreement because of different beliefs, different values, different preferences, different interests. To me they will fight over anything that isn’t an obvious answer.
11:16 The way I motivate this to to my former business school students, I’m now at the University of Wyoming, but down at CU, was you want the head of your marketing department if you come to them and say, “Hey, if I increased your budget by 10%, could you do more amazing things?” Similarly, your R&D division, you want the head of that particular division to say, if you gave them 10% more resources, they’re like, I would do really, really cool things. Go Cowboys indeed. I’ll be at the homecoming uh homecoming game tomorrow or Saturday. But back to governance as conflict. Ultimately, people should be disagreeing for good faith reasons within a particular organizational context. Back to the marketing versus R&D division. If each of them says, “I could do amazing things with a 10% budget increase and there isn’t enough shared resources for every division to have that level of increase.” Naturally, the role of governance authority is to resolve which unit within a particular organization should receive more resources. And guess what? If you think it’s contentious when it comes to a resource increase, now flip the hypothetical and say, should we kept cut 10% from marketing’s budget or R&D’s budget? And that happens all the time in economic downturns for many, many, many organizations in society. And so that’s but one example and I can spin up many many many others in terms of what is why governance is all about the decisions over which there is substantive disagreement on the part of organization members and so in some disagreement is not by itself evidence of bad faith.
13:08 Third, and this one’s a little bit spicy in uh, you know, blockchain communities that I often advised on governance, which was to argue that authority concentrates for mechanical reasons because attention is scarce and knowledge is specialized at a fundamental level. This leads to good faith institutional reasons for why we concentrate authority. Again, the example I gave to my B school students, finance majors in particular, was I can’t possibly have the time in my day to govern every last trunch of my retirement investments. They’re spread across public authorities, US treasuries, you know, all kinds of equities, foreign investments. I don’t even know the names of people managing many of my investment accounts. Yeah, there’s problems that come with that to be sure, but it’s also the case that I obtain a level of diversification in my investments that that’s thought to be just common wisdom for any financial plan. I obtain that level of diversification because I’m giving up paying attention to the management of my many investments. This applies in numerous contexts. Think of delegated representative authority within political systems. We elect people to represent our interests. We do not direct democracy that stuff. And I’m not a full believer in liquid democracy for things like say nuclear regulatory authority, electricity grids, air quality, you name the hypers specialized thing that I don’t know a damn thing about. Why would I want to have input to choosing what nuclear safety standards should look like? And so because of specialization in knowledge, I might be able to eventually learn about those things, but I don’t know a thing about them right now. We rely on certain, you know, specialized experts to make certain decisions. And in general, we delegate authority both because of attention costs, but also because of concentrated incentives.
15:18 I think that the CEO with stock options will stay up longer, think longer, and be far more concerned about the long run viability of the company that they lead relative to me. And so my being passive enables concentrated interests to act on my behalf. I’m not arguing that in every instance these result in good outcomes because each of the margins of concentrated authority that I’ve delineated results in serious capture or you know rent extraction concerns which is another design principle I’ll be coming to and so participation can authorize contest and reconsider delegation think of delegated representative authority and constitutional democracies but to me some measure of concentrated authority is emergent in nearly every single human organization at impersonal scale, which is what interests me. And so next, a direct segue from what I was just alluding to is that shared resources create asymmetric incentives.
16:21 Concentrated benefits disperse costs goes the mantra of public choice. Because of concentrated benefits, smaller groups, special interests as they’re called in that relevant literature, create it. This creates an incentive for them to have organized claims while the average voter doesn’t pay attention to what’s happening with the government budget. Thus, disperse costs weaken individual scrutiny in these circumstances. And because of a common expenditure out of commonly raised funds, whether it be transaction fees leading into a DAO treasury or public budgets in constitutional democracies, what you see in practice is small groups receiving a lion share of the benefits because of this mechanical margin. So this creates a structural pressure toward capture as opposed to arguing that capture is the de facto reality. There are ways to tame this particular tension within governance systems. But the big problem is and you know and this is especially familiar in blockchain context. A common treasury does not create uniform incentives to observe, monitor and adjudicate whether the uses of those treasury funds went well.
17:36 Next, and this is some work I’ve done with Sophia Kosar on the institutional privacy dilemma as well as Federica Carugati on opacity as constitutional capacity. But my argument is is we can overindex on the benefits of transparency to the detriment of protected spheres of autonomy that are thought to be a bedrock of civil liberties and associational freedoms. And so there’s a tension which is more visibility means less privacy.
18:06 And so on the legibility side, auditable votes and spending, but this exposure can itself enable vote buying and coercion and chill participation. You for many reasons, you want private votes in most democratic contexts. And I do I’m bullish on certain zero knowledge proof applications for taming this particular tension. But this tension never goes away and therefore a critical design choice for anyone engaging in protocol design intending to govern a complex human-coordinated system is you know it it to me this is a major one to take into account. Pseudonmous voting can expose ballots while obscuring who controls them. And you know, think of the many identity challenges that have created different fixes within blockchain voting contexts such as quadratic voting or you know, identity solutions that were presented upon yesterday by Kalia and I’m forgetting her uh compatriate whose insights I enjoy. And this poses the fundamental question of what must be verifiable and what should remain private.
19:15 Next, the dreaded constitution. So my view on constitutions is they are rules about making rules of necessity. Collective action requires you to first create a set of secondary rules. Otherwise, there are emergent cycling problems if you allow this the same people making a decision to at any time decide on their decision rule. And so you entrench certain rules that define authorities that can collectively decide, collectively implement, and collectively review. Not coincidentally, these correspond to the legislative, executive, and judicial functions in constitutional democracies. But because these will necessarily entail concentration of authorities in any human social group above a certain size or level of complexity, you also have to think about how you constrain this inevitably concentrated exercise of governance authority. And therefore you entrench decision rules, constrain the scope of authority. And one of my favorite ones is follow the money.
20:25 figure out where revenue raising uh authority resides and where expenditure authority resides because that is one of the shortcuts for figuring out within a constitution where the power actually is situated. And the other shortcut to understanding the dynamics of power that a particular set of secondary rules creates figure out who gets to change the second rule secondary rules. That’s amendment rules in constitutional design theory. And so to me, careful thought about all of these choices is a first order consideration. But often I’ve encountered this understandable tendency among participants in uh a particular context saying we want to formalize our governance. We want to constitute the thing because the thing is cool and we all agree it’s cool and we want to do it. to me often especially those engaged in kind of it it it computer programming are used to looking for a particular final solution just tell me which governance modules I need to insert and let’s hit go on the system and start doing the cool thing and we don’t have to talk to boring professors like you who are interested in these strange abstruse governance questions and I’m like no the axioms I’ve provided all have to be fundamentally tailored to a particular context. These are general principles and specific choices that emanate there from must always be consistent with an animating purpose and the governance context that pursuit of that animating purpose creates. And so which decisions within a system might require speed or specialized knowledge?
22:16 You’re probably going to concentrate authority there. Which choices can be localized? When are review and exit practical? And to me in particular, another major consideration is your governance regime will interact immensely with how easily or how you know it high cost it is for a particular participant to exit. These are tradeoffs not universal design principles when it comes to higher or lower exit costs. In one sense, governance is more representative when exit costs are lower. But in another really important sense, if no one has skin in the game because they can exit at any given point, you have a variety of commitment problems that large-scale exit from a variety of tokenized contexts have made uh have have made quite plain. And while standardization enables scale, it leaves consequential differences unrepresented. And so assuming that there is a single decision that will operate to the good of the entire group is to entirely ignore my governance as conflict axial.
23:28 Instead, you should be considering what is the actual what are the margins of heterogeneity within a particular group because the problem is is raw majoritarianism in context where you can exit means your minorities the more they’re def disenfranchised by a particular majority decision the more they will exit with respect to any such disenfranchising decision. And so you’ve got to take extremely seriously the nature of heterogeneity that a particular group and its animating purpose create in terms of pursuing the realization of that same animating purpose. And so things to consider are that purpose, the scale of the group, its existing history, because path dependence is a grinding beast in terms of determining the Overton window for any group’s governance choices in a given moment, as well as the distribution of stakes within that system. Therefore, I’ve identified a number of recurring governance challenges or problems, but that is not an argument that identical solutions are what can resolve those same problems. And a lot of my interest is about how we formalize mechanisms to deal with uncertainty. And so, the dreaded 2x two matrix of, you know, it they’re only becoming more common now with the advent of all of these LLMs. But for me, this is how we deal with the material emergence of uncertainty in governance contexts. And in particular, the more we protocolize, the more we create hard entrenched rules that are outside the control of any participant subject to them. To me, the more there is a really strong technical margin associated with this. Sidebar, you know, I’m giving this talk because I was fortunate enough to be selected for to be one of the participants in the first summer of protocols program. And this is kind of a 2x two matrix based upon my kill switch protocols article that I wrote with uh Seth Killian and Garrett David, other um other participants in that first summer of protocols. But this is all about how do you intervene in the context of an unforeseen event. And the margins are whether they’re technical or legal organizational or if there’s automated activation of a particular kill switch or discretionary activation. In the interest of time and getting through all my slides, I’m going to, you know, rip past the laboringness, but I feel like these are fairly self-evident. The big question is if you need an emergency authority and gosh if blockchain networks proven to need that despite being unstoppable money or unstoppable code execution so to speak. To me the fundamental question for any kill switch protocol is who gets to define the trigger and who uses the trigger and ultimately can anyone change what should fundamentally be an entrenched rule. You should be very worried if anyone’s amending the emergency provisions of your constitution.
26:37 To give but one example, what do you think Donald Trump would do to being able to declare a state of emergency and the executive authority he would assume under such a state of emergency as president to give but one hopefully not too startling of example. So an emergency mechanism of necessity allocates authority. pause poses the question of who may act, how quickly and how narrowly, who needs to spend in order to resolve the the situation that created the pause. Can this can temporary authority make permanent changes to the system elsewhere? Often it does so whether you like it or not. and who decides that we’re exiting that state of exception or emergency authority. And therefore, authority to stop ideally is not also authority to spend, amend, compensate, or restart the system. These are all means that you can still cabin this inevitable need to intervene in a running system. So ultimately, what we’re driving for here, the goal is dynamic fidelity, as I call it. This is in a piece of mind called digital constitutionalism on the problem of constraining concentrated network authority. To me, what we want is credible commitments without permanent rigidity. If you posit uncertainty and the need to decide in real time certain parts of your system more should be more in flux relative to others. And so what you don’t want is adaptation completely rewriting the commitments that users rely upon. But you also, you know, need sufficient adaptation because of these same unforeseen circumstances that I’ve posited spring eternal within sufficiently complex human coordinated systems. So observable commitments, constrained amendment processes and checking and balancing the exercise of concentrated governance authority.
28:34 That’s, you know, separation of powers in constitutional design. Another way to put it is no one can unilaterally control the system, let alone change it rules. And so ultimately, you want to govern amendment without entrenching every existing arrangement because some measure of dynamism is critical for any complex social group above a certain scale. So govern institutional commitments and also govern the authority to depart from those same institutional commitments because uncertainty makes the capacity to respond necessary and you’ve got to design for that even as public choice makes constraints on any such concentrated governance capacity also equally necessary. And so, miraculously, I’m actually on my final slide. And so, if I’ve piqued your interest about any of this, feel free to check out my new website. Um, hat tip to Jesse and Kate for actually making it because God Lord knows I am not an especially uh techsavvy person. But if you want to read a lot more incomplete rules, my email’s down there at the bottom. And you know, thank you so much for your attention and the opportunity to present here to the organizers.
Simone Cicero of Boundaryless presents O2A (Open Organisation Alliance), a proposed open standard for making organizations 'composable' by replacing implicit coordination mechanisms like hierarchy and meetings with an explicit, machine-readable contract graph of units, offerings, and relationships. He argues the protocol's compact grammar—three economic primitives (purchase, revenue split, investment) plus composition mechanisms—can represent bespoke institutional arrangements like Haier's Rendanheyi model or joint ventures as 'sentences' in a common organizational language, becoming essential as AI agents enter firms and transaction costs collapse.
Kaliya Young presents her Protocol Institute-funded research on the IETF as a 'group brain' — a durable collective cognition combining a memory substrate (the data tracker, RFC series, mailing lists) with a human 'knowing' culture (rough consensus, birds-of-a-feather sessions, sortition-based leader selection) that together produce capacities no individual member holds. She extends the framework to Wikipedia and DAOs as comparative cases, introduces the concept of a 'group membrane' (who's in/out and how boundaries harden toward power), and argues AI should augment rather than replace human group cognition, positioning group-level alignment as a missing middle between individual and humanity-wide AI alignment debates.
Open discussion session following the 'Inventing Psychohistory' symposium talks, with participants including Andre, Florian, Shawn, Venkatesh Rao, Sachin, and Mike Travers debating pace-layer time theory, fashion as an interface layer, and Shawn's concept of 'new nature' as technological systems with no one to appeal to. Venkatesh Rao connects this to Paul Ford's 'why wasn't I consulted' essay and Dan Davies' unaccountability machine thesis, arguing complex systems becoming unaccountable makes them paradoxically more predictable as autonomous machines; Mike Travers questions how this relates to historical disenchantment and Descola's animism framework.