A Knotwork of Bureaucracies
The many complex challenges that face the technocratic regulatory model
This post on Best Of A Great Lot is a part of a series on the subject of designing a new form of governance. Each piece aims to stand alone, but fits together on the Table of Contents.
This is also part 2 of a mini-series about the core idea of technocratic/bureaucratic governance. Part 1: The Case of the Like Button that Ate Society.
Last time we reviewed a small piece of the current debate over why social media is corroding our politics, and we defined an outcome scale (A-F) for judging policies. One possible Congressional response to these challenges is the technocratic one: pass a fairly broad regulatory mandate and leave the details to a federal agency. This is the choice of anyone who believes that technocratic experts make better choices than legislators, whether because of greater knowledge, skill, focus or depth of experience in the topic, or simply because they are insulated from day-to-day politics.
In an earlier section I wrote about how this model is built on the idea that self-governance is important, but doesn't deliver as much effectiveness as experts. As I said then, Matt Levine, in his review of this model, posits these downsides:
There are also objections though. Congress is elected, and the SEC isn't. Letting agencies write rules is more technocratic but less democratic; the agencies might be more captured by industry or just by longtime staffers who are less politically accountable.
I think this is underselling the challenges. Remember that we’re imagining an agency tasked with solving the social corrosion that social media seems to have introduced and that Jonathan Haidt claims is being caused by “Like” buttons and the engagement algorithms behind them.
Bureaucracies are complicated things, and there are a wide variety of forces at play within them. As we go through them, my aim is to show how some of these forces push for better outcomes, but more push against.
Some challenges when talking about proposed policies
Despite the confidence of politicians and administrators claiming that their policy will of course solve the problem in front of them, with a policy which hasn't been implemented, there's always a degree of uncertainty about its outcomes. Reality is complicated, and governmental policy attempts to modify human behavior at scale, one of the hardest things to predict. Policy arguments are often framed in terms of certainty - we must do this to solve this problem - but if we were able to be more honest about them, we would frame them with a lot more uncertainty and need to experiment - we think if we try this, it'll improve in this way. An honest perspective on this would be experimental and paying careful attention to outcomes to navigate which changes we want to keep.
One way to frame this with our buckets above is to think of a policy proposal as having a probability number for each of the buckets. Perhaps it's probably a B, but could well be a C, and could even go very wrong and be a D. Or it has a chance of being an A but significant risk of D or worse if we've misunderstood the second order effects. We could put numbers to this and give a policy an imaginary scorecard, like a 45B/30C/15D.
Unfortunately, the hypothetical and the real don't stack up cleanly against each other. Rarely do we run well designed experiments to see multiple policies against each other, so when we're talking about a potential policy, we usually can only compare it to current reality with whatever policies are currently implemented. This is messy, of course, and allows for lots of argument. Right now we have Section 2301 which states:
No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.
Imagine a new proposal which has been made. It is hypothetical, and might well be better or worse than Section 230 in specific ways, but we know Section 230 and we’ve lived with it for the last quarter century and gotten used to it. Large companies have built their businesses around it. It’s tough to fairly compare a new proposal with the current law.
When we imagine the hypothetical alternative, we usually simplify it based on our biases about it. If we're in favor of the policy, we ignore its potential downsides — instead of imagining a 30% chance of D, we round it down to zero — and play up its possibility of it being an A policy. Or if we like the current policy, we do the opposite. Without implementing it, it may be very difficult to tell who is right, and many of us prefer to follow the principle most easily summed as the devil you know beats the devil you don't.
The sunk cost fallacy comes for us all. Just as it's easier to accept the flaws of the house you currently live in, the job you currently do, the friends you currently spend time with than it is to go seeking the better house, job, friends, so the policy we currently have has understandable warts that we can live with. Once we've found a B level policy, the urge to seek an A is much lower. And many people are very capable of rationalizing D level policies as simply the best we are going to find.
Better policies seem to be harder to find. Perhaps A-level policies more often require collaboration between disciplines, or are radical or unusual, or both. Perhaps society is just so complex that A-level policies require understanding a lot more of the details to work out.
Bureaucrats are people too
Despite frequent rumors to the contrary, bureaucrats are not stamped from a single uniform mold. This may explain why there are such disparate views about who bureaucrats are at their core and how they affect bureaucracies. Here are a few of the views I've run across.
1. Dedicated, well-intentioned, even idealistic professionals.
Many people who want to make a real difference in the world decide that regulating the excesses of businesses and private actions is a good way to prevent disasters. Sometimes the corporate world says that they want to do good — ESG and corporate social responsibility are two such movements. But many people rightly view these movements with significant skepticism. Businesses have a long history of ignoring externalities — some particularly amoral theorists even argue that this is their responsibility as a profit-seeking entity. Meanwhile, academia often seems to be allergic to interacting with reality. For those with a strong urge to make the world a better place, figuring out and enforcing the rules for society — i.e. working in governmental bureaucracy — can seem like a solid choice. In this view, the public sector is where idealists congregate. Even some of the critics of governmental bureaucracy will concede that many bureaucrats are well-intentioned by responding with the catchphrase about the road to hell.
2. Selfish humans.
Alternatively, all humans are fundamentally selfish. Some (especially economists, whatever that says about them) imagine that whatever reasons a bureaucrat might say they have to do the job, their primary motivation on any given choice is a selfish one — usually they’re optimizing for the admiration of their peers or promotion within the bureaucracy. Others impute selfish motives more directly. They suggest that bureaucrats are in government because they want the benefits of the public purse: more difficult to fire, organizations that can't go bankrupt, a 9-5 guarantee and sometimes home by 4, etc. In the extreme are those who take advantage of the revolving door to help themselves: delivering business-friendly regulations in exchange for lavish thanks from their corporate masters. All of these are variations on the idea that a bureaucrat is primarily on the lookout for a benefit to themselves, whether relationships they can use later, a step up in their career, or a chance to goof off and not get fired.
3. Pencil Pushers and Red Tape Sadists
Lastly, the term bureaucrat is a dirty word in our culture because we have the story that this is where the hidebound, joyless, and actively sadistic go to enforce rules upon everyone else. Sticks-in-the-mud looking for every opportunity to ensure that none of us stray beyond their strictly enforced lines.
4. Incompetent
I could allow for a fourth category in here, the incompetent, but I think it's more useful to realize that this is an independent variable. If we imagine a control-loving power hungry pencil pusher, we should fear the capable one vastly more than the incompetent one. An incompetent bureaucrat can be frustrating or bumbling depending on the situation, but they're nothing compared to a capable control freak. A capable control freak with an encyclopedic knowledge of the rules and actual power to enforce them is a terrifying thing.
That same encyclopedic knowledge in the hands of an idealist can be put to tremendous good, or at least good intention. And there are plenty of incompetent idealists. The selfish and competent can become great corporate lawyers or rise quickly as a bureaucrat, while the selfish and incompetent can make great comedy by trying.
The categories above are a casting director's cheat sheet of Hollywood stereotypes, so let's take a step back and realize that actual people are a mix of character traits and can even behave differently on different days. Real bureaucracies are staffed with a mix of people, so overall you get a wide variety of behaviors that incorporates all admixtures of all three of these sets of behaviors, within a wide range of competence.
Whether we're talking about individuals as archetypes or how they're behaving on any given day, only the competent idealists will be usefully searching for A level policies by default. Unless the system forces the incompetent out and aligns the selfish's incentives with delivering A level policies, the selfish and pencil pushers will work with the incompetent to drive toward mediocrity. I would certainly hope that the experts brought in to staff our imaginary social media regulating agency would be competent and idealist, but there’s every reason to expect a flood of the selfish — hired from Meta, for example — as well as those who simply love enforcing rules on others.
None of this is specific to governmental bureaucracy. Most businesses, especially the big ones, function very similarly. Your rulebound pencil pushers and corporate ladder climbers and idealists all mix together in any bureaucracy. Some organizations will have more of one type or another, but no organization is pure.
Organizational Culture
Yet different organizations genuinely feel and operate differently, and it's not just that some manage to be full of idealists and others are all selfish jerks. A lot of those reasons get lumped together under the term culture.
On the one hand, organizational culture is simply the sum of all the individual decisions that people within the org make. At the same time, culture is real and strongly influences people's behavior, pressuring them to match the cultural expectations. It's ouroboric.
Culture is often difficult to describe and slippery, and as a result, some people want to refuse to believe it exists or minimize its importance. If it's just the sum of individual decisions, then individuals can change the culture by making different decisions, right? But culture in the aggregate has a strong feedback influence on its creators, because we are a social species: our expectations of what will be considered normal mostly determines what options we consider.
Individuals who run up against cultural norms have four choices: 1) ignore the culture and just keep being weird; 2) attempt to change the culture; 3) conform; or 4) leave. Most people are uncomfortable being the odd one out very often. The ability to successfully champion and create improvements to culture is rare for a mix of reasons: it requires bravery, charisma, energy, and communication skills, and that's just to start. You could say it's in the running to be a definition of the concept of leadership - not the vapid utterances that are usually passed off as leadership, but the real ability to improve an organization by changing its norms.
The upshot is that most people conform to the culture or leave. You can think of it as the boulder rolling down the hill, collecting an avalanche, or if you prefer, the katamari swooping you up. Small unintentional changes to culture accrete to enable it to become unstoppable. Occasionally someone brave enough manages to swing a lever at the right moment and shift its direction. Mostly people just go along with it.
Cultures of safety will drive toward conservatism and avoid change. Cultures of rule following will do well or poorly depending on whether the rulesets are well thought out or not. Cultures of fear will drive the idealists to become selfish and the selfish to become backstabbers. Cultures of excellence, though rare, will pursue improvement for as long as the leadership doesn't break them.
Cultures of safety will drive toward conservatism and avoid change. Cultures of rule following will do well or poorly depending on whether the rulesets are well thought out or not. Cultures of fear will drive the idealists to become selfish and the selfish to become backstabbers.
It's vastly easier for a culture to descend from a culture of excellence into a culture of rule following, safety or fear than it is to ascend up, unfortunately. There are old organizations that still have a good culture, but they're rarer than you'd like, and once a culture descends, it takes enormous effort to pull it back. Old habits die hard and all that. This is one of the core claims in favor of startups and skunkworks in the private sector: they are new, and have a hope of building a new culture that drives for excellence.
But startups and skunkworks are particularly unusual in government. It's not 100% true to say there's no cycle of Schumpeterian creative destruction in governmental agencies, but it's also pretty close. New mandates sometimes create either new agencies or new teams within agencies, which can have their own culture. Occasionally an old organization gets reorg'd out of existence. But it's vastly slower and rarer than in the private sector. Overall, the forces of culture in governmental bureaucracies are most likely to push against a culture of excellence — and the resulting A level policies.
Mission
In some organizations, the mission statement has a strong influence on the culture. Yet most who have worked in the corporate world roll their eyes when they hear that an organization is re-evaluating the mission statement. Why?
I postulate a law: the more powerful a social technology is when done well, the more commonly it will be done poorly.
So we should consider how commonly mission statements are abused as a sign of how important they are, but also a sign of how difficult a tool they are to wield.
Mission statements, written down, do nothing. They are worse than dead paper: they are dead hope and distraction rolled up in one. It is only when every leader in the organization has etched the mission statement so deeply in their brain that it drives their every decision that they actually affect the organization.
Because it seems to be de rigeur to do so in 2023, I asked Chat-GPT to write me an empty, meaningless corporate mission statement full of platitudes. Here it is for your amusement:
At our company, we are passionately committed to driving value-added synergies through our cutting-edge, best-in-class solutions. Our unwavering focus on stakeholder satisfaction, along with our culture of innovation and continuous improvement, allows us to deliver unparalleled value to our esteemed customers and partners. As thought leaders in our industry, we embrace diversity, equity, and inclusion as integral to our mission, and we strive to create a culture of inclusivity and collaboration that empowers all members of our team to reach their full potential. We are laser-focused on driving value creation and driving shareholder returns, while always acting with the highest level of integrity and ethical standards. Together, we are building a brighter future for all.
I can barely even read that, it's so painful. It's like when HR circles back on your synergy. For comparison, here's US Air Traffic Control's:
The Air Traffic Control mission is to provide rapid deployment, operation and maintenance of Air Traffic Control and Landing systems to ensure the safe, efficient flying operations in both deployed and non-deployed environments. Flight tracking and control of airspace anywhere in the world allows for safe reliable routes for U.S. military and allied nations' aircraft, utilizing the latest in tower and radar services.
Given the success of ATC over the past 50 years at making commercial flying safe, and comments from people who have worked there, I think they have taken their mission seriously. Real missions drive real actions: everyone changes their behavior to fit into the culture that grows out of the mission. If you get the right mission, you can get amazing outcomes.
Of course, if you get the wrong mission and wield it strongly, the organization will follow you into the wilderness too. Consider the FDA's mission:
The Food and Drug Administration is responsible for protecting the public health by ensuring the safety, efficacy, and security of human and veterinary drugs, biological products, and medical devices; and by ensuring the safety of our nation's food supply, cosmetics, and products that emit radiation.
But wait! That's not the FDA's entire mission. They've had several additional things tacked on. Here's one of them:
FDA is responsible for advancing the public health by helping to speed innovations that make medical products more effective, safer, and more affordable...
The FDA regularly makes the choice to focus on the first part — protecting the public health — over this bullet point about speeding innovation that is much farther down their list. It's hard to imagine anyone calling the FDA process something that speeds innovation or makes it affordable. Consider the fact that getting through human trials costs so many millions of dollars that we cannot afford to invent new antibiotics. A new antibiotic would be a literal lifesaver, but also would be set upon a high shelf for use only in the most extreme circumstances, and drugs that are used only in extreme circumstances don’t earn profits. Consider that some organizations are trying to get drugs approved in dogs because it's so much cheaper than in humans, hoping they'll be able to come back to humans later. They also know the risk: even if they succeed, they will run out their time on their patent and after the patent expires, it likely won't be worth putting the drug through FDA approval for humans. Consider the unpatentable old drugs that nobody will pay to get approved for new uses.
If the FDA prioritized innovation and affordability, its entire process and structure would be unrecognizably different from its current one.
It's not surprising that they don't. The FDA was born in a time when the critical need was to protect the citizenry from a wide array of scams. Its mission reflects that, and its organizational culture is solidly on that side of the fence. No one in the early 20th century foresaw the current cost spiral or culture of fear that would take over medicine in the US.
The challenge to anyone attempting to fix this is that changing the mission is the biggest of culture changes. Appending a completely different mission like "speed innovation", as someone clearly tried to do, won't cut it. It takes a complete rewiring of the leadership and the culture and usually a long time for the new cultural expectations to seep through the organization for something so fundamental to change, and it's extremely rare that organizations succeed at change like this.
In discussing the challenges today of the Homeland Security department, one journalist retells the story of how the Department of Defense, which was created by stuffing the branches of the military together after WWII, took 40 years for its on-paper culture change to take root. It's not unreasonable to expect a generation to pass through an organization before we see a mission change take effect. Just as many mission changes simply never take root.
To return to our imaginary regulatory agency in charge of social networks and Like buttons, it's easy to imagine that whether we got the mission right could have a strong effect on whether it was successful. An organization set up to prevent online abuse and misinformation will do vastly different things than an organization set up to ensure the safety of citizens while they engage in their right to free speech. And that's only two ways the mission could be slanted — there are many more possibilities.
I hesitate to even try to imagine how an organization set up today, even getting its mission right for today's environment, will affect society by attempting to regulate the technology of a hundred years from now as if it was made of Like buttons.
Public Choice Economics
A lot of the academic literature discussing the reasons bureaucracies do poorly for us goes under the term Public Choice Theory within the field of Economics. Most of the time when I see Public Choice Theory referred to in non-academic circles, it seems to be a shorthand for the idea that bureaucrats are of the selfish type above. In this imagination, many of them are even villainous, deliberately choosing policies that are good for them and bad for the rest of us.
Stories of villains are always popular, so this caricature of antiheroic bureaucrats isn't surprising. A fairer characterization of the literature would be that it describes many ways in which public servants are about as selfish as the rest of us, and how incentives within bureaucracies support that selfishness over the public good.
A full analysis of human selfishness includes a large range, from people who deliberately sabotage good outcomes for their own personal gain to the extremely common situation where people are most focused on the internal politics of their own advancement and have lost sight of the broader purpose that their organization is intended to promote. But hey, villains are fun, so let's start at the villainous end of the selfish pool.
If there's enough money involved, some particularly soulless people take offers from big companies for later advancement in exchange for delivering policy that benefits the company or its clients. For example, we have stories like the big accounting firms rewarding employees who promise to work in tax regulation and deliver rules the accountants can use for their clients. This is the basis of the laws against a revolving door between government and lobbyists, but the laws don't prevent the wide variety of abuse that clever lawyers are able to think up. For example, food megacorporations largely own the Academy for Nutrition, which regulators rely on. This version of the relationship allows for a level of plausible deniability that can prevent anyone from being prosecutable.
In some areas of our government this sort of problem is endemic. It seems you can't turn over a rock at the treasury department without finding former and future Goldman Sachs employees wriggling around underneath. Unsurprisingly, Goldman Sachs keeps doing really well no matter what market or regulatory environment it finds itself in.
Occasionally someone gets caught for being a little too blatant in their for-profit public service. One of the DHS inspector generals was indicted for stealing data from DHS with the intent to use it to set up his own for-profit shop.
If we return to our Like button controversy, we can imagine that a new agency would want to hire the experts in the subject. Many of the people who know the most about how Like buttons work became experts through professional experience at one of the giant social media companies, like Meta, that has Like buttons. Perhaps some of them want to see good rules put in place because they're tired of seeing the results of Meta's current behavior, but just as likely some of them are going to be looking out for Meta in trade for future wealth.
In 2022, Meta asked Congress to regulate the social media industry. One can certainly read their request as an acknowledgement of the role of government. Or one can read it as further attempt by them to cement a monopoly in the industry. There's a term in the industry that's worth knowing here: a moat. A moat is a competitive defense, it's something that competitors will struggle to match and thus makes it easier for a business to earn outsize profits. Apple has spent the last decade preventing third party app stores on their iPhones because control of the app store is a moat against competitor phones, which cannot access the Apple app store. Google gives away Chrome because it defaults users into Google's search and ads ecosystem.
Legislation demanding that companies in the market do something expensive that Meta already has set up is a powerful moat for them. A brand new social network would have to go deep in debt to match it. The EFF calls this out.
The proposal calls for lawmakers to condition Section 230’s legal protections on whether services can show “that they have systems in place for identifying unlawful content and removing it.” According to Zuckerberg, this revised law would not create liability if a particular piece of unlawful content fell through the cracks. Instead, the law would impose a duty of care on platforms to have adequate “systems in place” with respect to how they review, moderate, and remove user-generated content.
This sort of idea is standard fare for the idea of regulatory capture. On the corporate side, it is often deliberate, and villainous — Meta executives surely know that their smaller competitors — and especially new entrants into the social media market — would struggle to be able to afford the "systems in place" tax. The ones who manage to get users but struggle to make a profit will make great acquisition targets to reinforce Meta's dominance.
The grayer the area and the more money that can be made, the more we should expect to find this sort of malfeasance. Plenty of people are willing to operate in gray zones to help themselves out, as the sheer number of corporate lawyers in existence attests. Many fewer are willing to out-and-out break the law. But with the scale of our government, even if the more egregious form is rare, we should expect to find enough of these stories every year to keep journalists happy.
Much of Public Choice Theory describes forces that drive governmental bureaucracies toward producing B-F outcomes, with the villains pushing even more strongly than the merely incompetent or mildly selfish.
Problems shared by private bureacracies
But enough with villains! As appealing as they are, they may well be relatively rare, and even without them, we still have plenty of reason to expect poor outcomes from governmental bureaucracies. We get plenty of mediocrity from the other set of bureaucracies we live with: big businesses.
The sociological causes of dysfunction within large bureaucracies have been extensively documented for businesses. Moral Mazes and the Peter Principle are two landmark descriptions of the challenges they face that are applicable regardless of governmental or for-profit. Oversimplifying, Moral Mazes acknowledges that people within a bureaucracy have stronger incentive to do what their peers and managers expect than to be ethical. The Peter Principle describes how a competent individual continues to be promoted until they reach a role where they are no longer competent, but then cannot be demoted or fired because of their history of competence.2 Many other books have been written on the problems of management and leadership within business, and these problems work to push government bureaucracies toward B-F outcomes too.
Here are just a few of the details. Bureaucracies are made of humans, who make mistakes but don't want to admit to them and will have to find ways to live with them instead. Sometimes they'll make mistakes about hiring and promoting people, and they'll want to cover up those mistakes too. Many of these mistakes will persist and cause follow-on mistakes. Incompetent people will stay because of their history of service or because it's just too embarrassing to fire them. Incompetent people will hire more incompetent people — in business circles the phrase is "A players hire A players. B players hire C players." Playing politics to get promotions will often trump doing what’s best for the citizens or even being right. Firing in government contexts is harder than in for-profit contexts, so poor employees will last longer and can wreak more damage. Competence is harder to tell in more abstract work, and a lot of policy work is the most abstract work available.
Most of these forces push toward C or worse policies. But there are also forces that push back, trying to drive better policies. Governmental work happens a little more in the open than even the most public of businesses, so sometimes journalists, academics or others can notice problems and push back on them. Goal setting, though much despised in industry, can at least work to minimize the worst of outcomes, pushing back against the F policies. Congressional committees can, at times, exert a level of accountability and this can be useful pressure for the competent idealists to use to push their less competent or less idealistic colleagues.
Overall, the forces of dysfunction are often stronger, but we get some successes, especially when the job of the bureaucrat is to stop something fairly straightforwardly bad from happening. For example, since the 1970s the regulations around the Clean Water Act have taken us from rivers and harbors being mostly toxic sludge to being a place that can harbor real ecologies. This is an enormous success driven almost entirely by regulatory bureaucracy given a mandate to (literally) clean up the world.
Amplified when having to work across agencies
One bureaucracy is a challenge. I think the medieval monks who gave us a “shiver of sharks” would suggest a “knotwork of bureaucracies” as a term for the tangled regulatory agencies that we have evolved. When agencies have to work together, all of the above problems are magnified.
Whether idealist or selfish ladder-climber, whether serving a purpose with an obvious reason or a subtle one, bureaucrats who have to coordinate across departments or agencies face intense challenges. Since every agency has its own culture, and culture often feels invisible to us, navigating another organization's culture is usually particularly fraught. Working against the cross-agency go-getter is also the very process of goal setting. It's rare to see carefully worked out cross-agency goals, and it's difficult to line up the work you want them to do with the goals they're (at least ostensibly) pursuing. The go-getter's goals just aren't as important to most bureaucrats as their own. Tribalism also plays a role: I'll do anything for anyone on my team, but those other bureaucrats over there are playing on a different team.
Maybe there's a simpler world where either agencies don't need to work together as much, or the lines between agencies could be easier to bridge. In our complex world, the best solutions are multidisciplinary, and stepping on toes or missing each others learnings are common.
To summarize: we should expect that any agency set up to regulate Like buttons will have to navigate through a wide variety of challenges, including: finding more competent idealists than other types of people; getting their mission right now and into the future; building a culture of excellence; avoiding regulatory capture; and limiting the normal problems of any large bureaucracy. Part three will discuss why regulating Like buttons isn’t likely to go as well as cleaning up pollutants from rivers or limiting the amount of arsenic in medicines.
If you find my work interesting, please share and subscribe. It helps tremendously.
Section 230 of the Communications Decency Act is actually a law, and a hypothetical agency in charge of regulating Like buttons would not be able to override it. In this specific instance I’m imagining a case where Congress allowed the new hypothetical internet communications agency the right to overturn or keep Section 230 based on its expert opinion of what was best for the internet.
One of Toyota’s surprising differences that drives their remarkable effectiveness is that they encourage demotions. They view a demotion as a sign that someone tried out a leadership position and discovered that they weren’t ready for it, and that both steps are good for the organization.