On governance
In which the game that explains all of civilization is introduced.
For some reason, I spent far more time researching the history of castes, estates, and other stratified social orders over the last two or three years than in my professional life before.
Discussing economics from a perspective of class struggle has that semi-nostalgic air of 19th-century political economy as a lesser branch of the moral sciences, of solemn waistcoated establishment figures defending the prevailing order against finger-wagging radicals in leather-elbowed tweed coats, of bespectacled men with beards pontificating in smoke-filled Victorian parlors. Of the bygone world Heilbroner’s worldly philosophers inhabit.
It’s as far removed from the things I usually do as one could imagine.
There’s a bunch of reasons why my perspective has changed. One big reason is this chart (originally from Oliver Williamson’s 2000 presidential address), which escaped my attention at the time, but which closed a lot of gaps when I rediscovered it a few years ago.
To explain this I have to fill in some biographical background: Before returning to grad school I worked in telecom R&D, on what was then in Bill Gates’s view the coming “information superhighway” based on the 7-layer OSI protocol, for which in the eyes of Gates and other visionaries the academic Arpanet-cum-Internet and its ramshackle TCP/IP protocol stack was at best a low-bandwidth precursor.
At the time I witnessed the conflicts between layer 2 and layer 3 developers at close quarters, a clash of incompatible mindsets exacerbated by a language barrier — between programming languages, that is. I also witnessed our own R&D program and the global quest for the “information superhighway” go belly-up right after Mark Andreesen had dragged TCP/IP out of its academic cubbyhole and pushed it towards homespun tabloids, flame wars, and world domination. The rest is history.
With this anecdote in mind I made the point in Williamson’s now eponymous seminar on institutional economics that this interlayer friction in software development bears a striking resemblance to the methodenstreit in economics (in both forms, not only on the question how much formalization is adequate to encapsulate economic problems, but also which formalization), and that a useful way to resolve this strife would be to propose a layer model of economics, since presumably not all subfields in economics benefit equally from the same level (and kind) of formalization.
I don’t remember the timing of my comment, so I can’t claim credit for inspiring his work. But I regret that it didn’t catch my attention, since it might have led to some fruitful discussion and elaboration. So take the following discussion as the belated attempt to atone for this.
A simple pattern language of relationships
Governance is decision making for others, on behalf of others.
The connection to the realm of this newsletter, the cybernetic economy, should be readily apparent since governance and cybernetics share the same etymology: kŭbernắō (κυβερνάω), the pilot or helmsman of a ship. Both are about steering.
This connection still needs fleshing out, especially in light of the goal of EconPatterns to introduce an economic design framework.
The reason why I’m starting with Williamson is, beside the coincidental fact that his final book was called “mechanisms of governance”, that the fundamental dichotomy between markets and hierarchies in the first book of his famous trilogy sets the scene.
This dichotomy, that transactions can be conducted either within or between entities, drives the split into business vs economics, decision theory vs game theory, bounded rationality vs “full” rationality, organization design vs market design, and many others.
This is not what I’m taking issue with. As a conceptual dichotomy, indeed as a pattern, it has proven to be extremely useful, even if the boundary between the two is inevitably fuzzy. Market vs hierarchy is a powerful economic pattern at a certain level of abstraction.
My issue is that both organizational hierarchies and markets are highly complex mechanisms with many moving and interlocking parts. Also, that markets themselves are organizations.
A typical corporation operates as a multilayer, stratified, one-to-many reporting relationship (the eponymous hierarchy), with an executive suite, a supervisory board, and a population of shareholders. The reporting relationships tend to be steep, the board and shareholder meetings comparatively flat.
A market is a mechanism to facilitate transfers, typically in the form of transactions, aka exchanges of goods, money, information, promises, and other things economic. It could be one-to-one, one-to-many (auctions), or many-to-many.
These are all parts to play around with, to reconfigure, and trying to find the perfect combination of interlocking parts seems to be the everlasting pastime of business schools and economics departments, not to mention strategy consultants.
These “parts to play around with” are the Lego bricks for designers, including economic designers.
But what exactly are these parts and how do they interlock?
And relatedly, are there any parts that combine in different ways to different ends? Are markets, hierarchies, states, religions, and other organizations just the same elements put together in different ways? If so, what are these universal basic building blocks and how do they fit together?
Where Williamson sees two, I see four: vertical, mutual, competitive, and random, with mutual and competitive forming the horizontal relationships.
The vertical relationship is the canonical reporting relationship, or, as Ronald Coase called it, command-and-control. A task is assigned, performed, evaluated, and compensated. The participants in such a relationship are known as the principal who assigns, evaluates, and compensates, and the agent who performs.
The mutual relationship is a simple exchange relationship where two things are swapped, two transfers forming a transaction, on an even keel. The things could include economic goods or payments but also promises, well-wishes, goodwill, or any other expressions. Once we're outside the economic realm, we’ll likely speak about interactions rather than transactions, but it's the even keel that matters.
The competitive relationship would be the same horizontal exchange, but now we're exchanging blows. While the horizontal relationship should at least in principle, and to the extent it's voluntary, make both sides better off, the competitive relationship is one or the other.
Random is random.
The first intriguing thing is that a market is now a mixture of the two horizontal relationships: competitive and mutual. First there is an allocation decision, which is (at least for private goods) competitive on both sides, followed by a exchange of transfers (the transaction), which is mutual. This is the fundamental idea of a pattern, breaking seemingly obvious things down into their constituent parts and putting them back together again.
A first objection will be that these things rarely come in their pure forms. This is true, there's such a thing as market power which typically determines who gets how much of a seemingly equal horizontal exchange relationship. Supply chains are horizontal in theory, but highly stratified in practice.
So already we're in a world that's skewed.
If horizontal and vertical relationships are separated by precedence, mutual and competitive relationships are separated by intent.
In other words, direction (or order of precedence) is one parameter, and intent is the other parameter, and they can certainly take on other values than 0, 1, or minus 1. If there’s precedence, the relationship is vertical. In a hierarchy that's the direct reporting relationship. It can be horizontal, like with direct colleagues. But it can also be something in between like a relationship with someone who is higher up in the hierarchy but not in a direct line.
Similarly, the relationship between two colleagues in the same team can be both mutual and competitive, because in part the compete for resources, and in part they work together to achieve mutual goals. It's never simple.
Pretty much every relationship that points into the future will have a random element. But as we know, randomness is rarely ever perfectly random.
Protectors vs providers, the civilization game
This finally brings us to the opening statement of the last section. Governance is the paradigmatic vertical relationship, but with a twist. It's directing someone but at least in intent with that someone's interests in mind. In short, it does, or should, contain a feedback loop.
That twist seems to be the reason why governance has become the buzzword du jour, replacing former biz school buzzwords like administration or management which mean pretty much the same thing but don't have that caring-for-others vibe. Sadly, in the harsh world of modern corporatespeak, “governance” still mostly means bogging down subordinates with paperwork. It just sounds nicer than compliance.
But I’ve been tweeting about governance since before it was cool, so I can go ahead and disassemble the original Williamsonian meaning of the word rather than its buzzword bingo stepsibling.
Speaking of tweets, a couple of years ago I tweeted about the “least known game in game theory”, the protector-provider game, aka swords vs ploughshares.
One fairly compelling reason why it's so unknown is that I made it up. But it also has very little chance to make it into a game theory textbook because it's utterly pedestrian and I wouldn't even have an idea how to express it in mathematical form.
It simply goes like this.
Two equally matched settlers (twin brothers if you want) decide to start a tiny economy (an oikos) and split tasks. They flip a proto-coin to randomly assign two roles: the protector who scans the perimeter for hostile man or beast, and the provider who tills the field. It's all good as long as there's no conflict, since they’re mutually dependent on each other. But as soon as there is, protector will come out on top, and will eventually end up owning the oikos.
As I said, it's as simple as that, and I can't even claim that the denouement has any kind of proof. It's simply because if it comes to strife it's faster to hit your adversary over the head than to starve him. Even if it comes to armed combat, a sword will almost always beat a ploughshare — or, for that matter, a pitchfork.
As with much of what I write in this newsletter, I invite you to take this little game, apply it to everyday situations you encounter, in this case any kind of vertical relationship, and ask yourself if it works as some kind of risk transfer upwards — the risk transfer we call protection.
If it reshapes the way you think about things even in this purely informal form, it's a pattern.
Maybe someone else finds the time to express it in a game-theoretic formalization. Until then, what’s so intriguing about it even in its current form is that it encapsulates almost the entire civilizational history in five or six lines.
Almost. And that's the reason why I started investigating caste systems, class struggle, and other economic topics with that Victorian parlor vibe.
There's more to it because protection isn't just protecting against man and beast. There's also evil spirits, divine wrath, ill omens, falling skies, floods, famines, and other adversaries, human or superhuman, past, present or future, asking to be warded off. Suddenly that coveted job doesn't look that easy anymore.
It's time to split the workload and, in keeping with Adam Smith, bundle the tasks by skillset to prevent death by war, pestilence, or famine. By job description, the roles should fall to specialists for handling hostile acts by man and beast, by nature, and by… misallocation?
Clearly famine caused by man or nature falls into the realms of the first two, which leaves the somewhat lesser role of feeding everyone by storing or moving victuals around to a less elevated position in the protector class. It's one step up from the provider, but not by much.
Which quickly gives us the whole “standard pattern” class system of the feudal era.
Admittedly, this is a crass oversimplification because feudal systems are notoriously complex, byzantine even. But the standard pattern is remarkably similar across cultures, no matter if they’re monotheistic or polytheistic, propped up by worldly or divine power. Division of labor in the protection racket seems to produce the same job descriptions everywhere.
Let's give the protection professionals the names we're familiar with: the priestly class first, the warrior class second, the merchant and administrator class third, and, depending on if the local caste system even bothered to list them, the peasantry, our poor original provider class fourth.
(Sometimes there's also a fifth class, the untouchables, which comprised those who gave the top two classes too much trouble and had to be put away before the crime.)
So in a few paragraphs, we captured the essence of much of human history, and we even managed to give our basic relationships a bow: the horizontal, mutually dependent protector-and-provider oikos turned into a competitive and then vertical economy once interests were no longer aligned. Sounds familiar? And we even managed to include a coin flip!
And for those who have been paying attention from the very beginning: in the roles of the bottom two classes we recognize the fundamental tasks of the cybernetic economy: transforming, storing, and distributing, along with extracting and consuming.
Now that we have our players lined up (vertically), we can pursue the question how they ended up in that particular order.
The lower two are easily dispensed with (as they were for much of history). The operational classes had to stick to the bottom of the order because they suffered from what Thorstein Veblen dubbed “menial contamination”.
Doing was clearly considered inferior to directing, aka telling others what to do, for the reason we established in the third law of organization, namely that the task of organization is to resolve the conflict on which direction is forward.
So even the two denizens of the third class, merchant and administrator (market dude and hierarchy dude), which combined we might dub the professional class, only translate the orders from higher up to all the way below. Even if we think of the lower two classes holding monopolies on direction and labor, the “direction” is really constrained by what serves the upper two.
How this split that defined the pre-modern era came about is again exemplified in the tools. It's still sword against ploughshare, and even if we assign, say, the compass to the professional class, it's still no tool to beat up a sword-wielding knight with — up until a juncture in human history: the enlightenment.
But to answer the question how the priestly class came out on top of this racket, we turn again to Williamson.
The TCE/IE protocol stack of economic institutions
Oliver Williamson’s four-level protocol stack of economic inquiry fits neatly into our four-layer standard pattern of human civilization, but only if we tweak them a little. Even the time spans he assigned to each level roughly pan out.
The need for tweaking is the reason why I don't present the original here, but my tweaked reinterpretation. But regardless, I recommend scanning both the original chart and the original paper.
His first level, Embeddedness and informal rules touch on Mark Granovetter’s famous eponymous paper and accordingly assigns that level to sociology/social theory. The time frame of centuries to millennia matches that of a world religion.
The second level, Institutional Environment and formal rules matches the state. Decades to centuries as timeframe, political theory as the matching field, both look like good fits.
His third level, Governance, playing the game, years to decades, transaction cost economics as analytical lens: We reached Oliver Williamson’s home turf: the firm.
Finally, ongoing Resource Allocation using the time-tested but worn-out marginalist framework is left to the minions of the trade, the traditional economists— “orthodoxy” in his terminology, the unswept shopfloor of the economic production function.
Beyond the identical number of layers, which might very well be a coincidence, can we draw any more connections between Williamson’s four-layer protocol and our standard pattern of a feudal society, can we find any further connections? Indeed our first starting point should be the realization that “hazards”, “safeguards”, and similar terms are part of Williamson’s standard vocabulary. His contention that layer four, “orthodoxy”, has little to say about how longitudinal contractual arrangements are secured simply because it assumes contracting hazards away, dots his published output.
Oliver Williamson was clearly in the protection business.
So let me roll out my own interpretation of the four-level model based on the “standard” feudal classes and their canonical output.
Belief and the creation of rules, assigned to the priestly class and world religions.
Violence and the enforcement of rules, assigned to the warrior class and the (nation) state.
Direction and the implementation of rules, assigned to the professional class, to firm and market.
Labor and the fulfillment of rules using physical effort, assigned to our poor provider class.
World religion, the nation state, the firm, and the individual remain. The time frames as well. Sociology, political science, and economics (microeconomics really) as the home sciences for levels 1, 2, and 4 are also still in play, with business administration more generally at level 3, Williamson’s home turf. (Williamson, while calling himself an economist, was entirely a product of business schools, as was Ronald Coase.)
Just like this, we built ourselves a top-down four-stage production function for the standard civilizational ruleset: persuasion, enforcement, direction, doing.
Which leaves me with the promise to answer the question why the priestly class came out on top of the pile, even if the warrior class as sword carriers epitomizes the entire protector class. What's the tool that keeps swords at bay? Whatever it is, it must’ve been powerful enough to force emperor Henry IV to kneel down before pope Gregory VII outside Canossa Castle.
The answer, again, comes down to transaction costs. The top two classes were mostly aligned in their mutual interests in fleecing the lower two, but they frequently clashed throughout feudal history trying to get most of the spoils.
Directing the populace, that core element of governance again, depends on the willingness of the populace to follow. Between the two options we have, beliefs and violence, which one is the one that wins out in the end, should they come to blows?
Again, I could go ahead and mathematize this question, but instead I simply submit that the answer is “violence in the short run, belief in the long run”.
Suppressing an unbelieving population with sheer physical force is certainly doable. It's been done many times in history. But it comes at a high level of friction.
To reduce that friction worldly rulers throughout history have recognized the value of divine blessing, of divine origin even, no matter how rough and tumble the actual ancestral lineage was.
World religions, the ultimate certifiers of good conduct, were all too happy to provide that service most of the time. But it came at a cost to the warrior class, namely the acceptance of the priestly class at the top of the pecking order.
So if we assign a characteristic tool to the priestly class, it behooves their status as controllers of beliefs to pick an immaterial one. If it's ploughshares for peasants, compasses for the bourgeoisie, swords for the fighting nobles, it should be the spell in the hand of the priest that symbolizes their unique powers.
Governance is decision making for others, on behalf of others. The feudal era, or the pre-modern era in generally, was mostly decision making for themselves, with only scraps left for the lower classes.
How did we ever manage to get out of a world where the twin threat of spell and sword was enough to compel almost everyone to give up almost everything?
The short answer: feedback. The somewhat longer answer: bureaucratization of feedback.
The feedback loop of (good) governance
Governance is decision making for others. Good governance is decision making for others, on behalf of others.
If almost everything up until now was about the “decision making for others” part, we should devote the remainder of this newsletter to the “on behalf of others” part.
Collectively, we should certainly feel happy to have escaped that feudal trap, no matter how nostalgic medieval markets and renaissance fairs appear to someone with a salaried office job and a comprehensive dental plan.
Using the toolkit we have built up so far:
the cybernetic economy and its triple of stocks, flows, and transformations
organizations as belief clusters to resolve the conflict between moving forward and staying together
attention as the means to reconcile the conflict between beliefs and observations
and the new tools from this newsletter:
vertical, horizontal (mutual or competitive), and random relationships
persuasion, enforcement, direction, and application of effort.
How do we get from the feudal world to the modern world?
How do we build governance structures that come close enough to rewarding individual contributions to collective efforts so that we can minimize friction in a world where everyone enjoys freedom of belief, of movement, of expression, of profession, and other freedoms we now take as unalienable?
In other words, how to build governance structures that rely more on voluntary than on enforced participation, on monopolization, captured rents, barriers to entry or exit, or other restrictions?
The answer lies with Max Weber: the bureaucratization of the protective functions.
The short version, and we’ll stay with the short version here because there is so much to unravel that it deserves its own newsletter, is that the professional class, step by step from renaissance to enlightenment to industrialization and the American civil war (the first industrial war) to the end of the feudal world after World War I and its aftershock in World War II, took over the roles of the protective classes, simply because it was better at it.
We had this before in feudal history, when towards the end of the völkerwanderung, the burly professional-class Carolingians pushed out the timid warrior-class Merovingians to produce Charlemagne, the Holy Roman empire, and the high middle age.
But with incipient industrialization, scientific observations pushing out devout observances at universities, and the emergence of well-drilled standing armies, the professional class not only took over managing the affairs of the state and its protective obligations, it also started to make insistent claims that it's better at governing it.
A key element of this hostile takeover was establishment and bureaucratization of feedback in the form of general elections. Ballots replaced ploughshare and pitchfork as weapons of choice for the governed.
The idea of “general” elections wasn't the impetus for the push to have more of a say in steering the nation, the fourth class was generally not considered competent and qualified by the other three to adjudicate between different proposals to move the commonwealth forward.
Many intermediate forms were considered and tried out to restrict suffrage to male members of the landed gentry and the civic merchant class — men of means and heft. But in the end, “adult citizen”, with arbitrary cutoff points at 21 or 18 years of age, was found to be the solution that minimized transaction costs.
But humans don't all turn qualified overnight on their 18th birthdays thanks to some perfectly aligned internal clock or celestial intervention. It's just easier to look at a birth certificate to answer the simple yes-or-no question “Are you qualified to vote?” In countries without central registers even that task is hard enough.
With the means of bureaucratized feedback, even in the bare-bones form of general suffrage, the “for others” and “on behalf of others” gradually became more aligned.
The inherently vertical governance relationship was tempered by shifting the nominally horizontal relationship among the populace into the direction of a mutual rather than adversarial “each man for himself” relationship.
The industrial equivalent of general elections: collective bargaining also arguably played a big role in shifting the power center and society-defining conflict from first against second class to third against fourth — the Victorian class struggle of capitalists against workers that opened this newsletter, and that ended in November 1989 with the fall of the Berlin Wall.
In the process of bureaucratization of state functions, the professional class usurped the roles of both protector classes. For a while, as Thorstein Veblen so keenly observed, it also absorbed their uppity habits.
With the protective classes now defanged and the professional class in full control, civilization lived through a short gilded age where it was in full control and governance was again deciding for others but not on behalf of others.
A somewhat more equitable game emerged when the new priestly class, the academics, gradually shifted allegiance to the working class, less in the guise of salon radicals like Marx or Sombart, and more via the erudition of liberal establishment figures like John Maynard Keynes or Paul Samuelson.
This newly aligned patronage of the feudal classes also shaped the demeanor of the new-world combatants in the class struggle: doctrinary vs belligerent.
Joseph Schumpeter presciently recognized this emergent realignment — no small feat in 1942, when universities were still ivory towers of the bourgeoisie — and he also predicted that this mismatched coalition between neo-priestly academic class and working class would prove to be temporary. A keen foresight given the ongoing collapse of the traditional left-of-center labor parties (in favor of doctrinary, purely academic parties) in the Western world.
The post-feudal realignment of the tectonic plates, which led to a civil war, two world wars, a cold war, and 35 years of unilateral American dominance, seems to be shifting backwards.
Governance vs good governance
Governance is decision making for others, good governance is if these decisions are also made on behalf of others.
EconPatterns is not supposed to produce grand societal narratives or even civilization-spanning game theory. It is supposed to be a practical guide for the everyday conceptual work needed to tackle economic design problems.
It is supposed to think about steering around challenges from nature and other actors.
It is also, and I reserve this right as its sole author, an outlet for my thoughts, knots, and insights I have while working on everyday design problems.
One of the most fundamental insights, and a source of personal regret, is the late recognition that the Williamsonian lens, or maybe the Coase-Williamsonian lens, is absolutely crucial for thinking about everyday economic design questions — even and especially in the digital age.
Steve Levitt claimed that the pre-millennial buzz about transaction cost economics had dissipated by the time Williamson received the Nobel in 2009.
I thought and still think that he was dead wrong at the time, but he is right that Coasean and Williamsonian transaction cost reasoning was deeply rooted in the industrial age, with their lighthouse, Fisher Body, and CATV war stories.
At the dawn of the Internet age it seemed at risk of falling by the wayside, being pushed out of bounds by edgy new MIT microeconomics with its dazzling machinery perfectly attuned to the coming information age — even for me.
I’ve completely changed course on this. The simple reason is that when designing economic engines, especially online engines, my time spent on MIT-type economics is comparatively little compared to the time spent on safeguarding the economic exchange.
I no longer spend much time thinking about the information economy with its MIT toolset. Most of what I do is rooted in the attention economy, which is marked by information overload, not information deprivation. More Simon than Samuelson. Everything in EconPatterns reflects this.
The Williamson box (or maybe Coase-Williamson box) that maps economic transaction into contracting phases (matching, agreement, transfers, recourse, reputation) and their respective specialized engines has become a standard design tool to think about online commerce.
The Williamson box does for horizontal (contractual) relationships what the 4-level protocol stack does for vertical relationships. Which means that with the two we’ve gone full circle, back to Williamson’s original dichotomy of markets vs hierarchies, just now in design language.
But that’s for another newsletter.