50 Comments

"On the other hand, perhaps a patient could die because a qualified doctor or nurse is not around to administer life-saving medication that a less-qualified hospital worker might have provided. That would be a Type B error. I imagine that Type E errors are more of a problem than type B errors, so that a hospital is better run with bureaucratic rules than as a free-for-all where anyone can administer medication."

What's "more of a problem" depends on your perspective. From the perspective of the patient the type of error is irrelevant. They're dead either way.

The main benefit to the hospital avoiding type E errors while allowing type B errors is that it can generally avoid any liability for type B errors since it's almost impossible to prove that a patient died of inaction rather than action. This is generally the biggest problem in any bureaucracy. Harm caused by action results in concrete and actionable liability while any harm resulting from inaction is generally diffuse and inactionable, giving all bureaucracies a bias towards inaction and avoidance rather than problem-solving.

Expand full comment

Great phrasing, Empowered vs Bureaucracy problems. With a trade off, of getting more of one whenever a system changes to get less of the other. Very analogous to Type I & Type II errors of false positives & false negatives.

Also important to note that bureaucracy is required for organizations, to coordinate actions. Not noted explicitly is that org bureaucracy is far more needed in order to build something big, than to merely protest or oppose (or write comments). But bureaucracies get slower over time, so their cost slowly increases.

In IT software dev, the waterfall method is quite bureaucratic, and slow, yet was built up to avoid errors that are costly to fix. The agile idea is to empower the small team of developers to more quickly crank out something that works, also seen in hackathons.

Society needs more work on identifying ways to make bureaucracies, not just organizations, work better. 8 year term limits on US govt bureaucrats, about 13% per year turnover, seems like a change for the better (Vivek).

Expand full comment

Terms limits on the gov bureaucracy are simply a bad idea hard stop. The government has many niche positions and you aren't going to develop expertise in eight years. Do you really want 4 star generals commanding a war to only have seven years of experience and be short timing the eighth? Do you honestly think firing the PhD geophysicist working in the national tsunami warning center every eight years is going to make the field of studying tsunami wave propagation better it better? Etc

Term limits aren't the problem, the lack of oversight of middle management by all three branches to ensure they are faithfully executing their assigned duties is. Bureaucrats at the top and bottom are usually competent, it just gets lost in the middle.

Expand full comment

Lack of oversight of middle management is certainly a key problem of all bureaucracies, and especially without good metrics to judge outcomes. The profit metric is the primary reason profit making orgs do so much better over time than other orgs, which do not have bottom lines to evaluate performance of top line revenue, which depends on customers, minus the internal costs that depend on management decisions, whose job is to get the most work out of workers for the lowest price.

This doesn’t, and can’t, exist in government orgs—whose middle management becomes expert at CYA (cover your ass) to avoid blame for mistakes, while bootlicking (ass kissing) their bosses. Bootlicking is also in private orgs, but usually is dominated by cost & revenue & innovation metrics.

For tsunami warning, I certainly prefer govt subsidies to competing orgs, like research institutes where the workers are not govt employees, preferably with less govt funding. Had Fauci been replaced after 8 years, with constant searching for top health experts getting 4 more over the last 40 years, I’m sure we would have had a better, more measured response to COVID.

FBI and CIA agents should certainly be cycling thru state and local police offices, with the police chiefs who have been doing the best jobs becoming hired to lead the FBI. Everybody should be disgusted that 51 top officials were willing to lie about H Biden’s laptop, with none of the experienced officials in the DoJ willing to tell the truth before the 2020 election, even tho they had the info since 2019. Had all of them been limited to 8 years, it’s far more likely some less experienced top officials would have been honest.

The special case military would likely be an exception to an 8 year limit. I’m not sure if they still have an “up or out “, but they leave too many generals and admirals, but not enough ships or tanks or ammo, nor killer robot dogs & drones.

In all of the US wars, when fighting started there were lots of lousy generals: Civil War, WW I, WW II, Vietnam War. It remains the case that most generals are focused on fighting and winning the previous war.

I want the active reserve general who correctly predicted that Putin would invade, and also correctly predicted that Ukraine would survive the first invasion, I want that guy to be recalled to active commanding duty.

But reform of the peace to war time military org is beyond the scope of a comment.

The claim that 8 year limits is simply a bad idea, is false. If tried, it will have lots of good churn, and some problems. It might even have more negative effects than positive effects, especially at first. But once implemented, the orgs will develop rules, red-tape, to reduce the problems. Probably also reduce the benefits (as Handle previously noted some weeks ago), but given our current govt bureaucrats, replacing all of them in the next few years remains a far better option than praying for magical better oversight.

Expand full comment

So I think you and I are using your vision of term limits different, you seem to be using them "8 years in the same job" which will have absolutely zero affect as opposed to how I was using it "8 years government service period". Very few government people stay in the same job for eight years, people are people and they like to get promoted, change pace, etc. For example GS-9 admin assistant Gary for Deputy Chief Richards in an eight year period will most likely become GS-10 admin assistant for Chief Smith. GS-12 Contract Officer Shaniqua for the US Army will most likely become GS-13 Contract Office for Social Security Administration. It's basically the Putin/Medeved swap game which accomplishes nothing in practice.

And bouncing someone between state/municipal jobs doesn't change that as government bureaucracy is the same at all levels, maybe even worse at state and municipal to be honest. Plus the "move up/out" leads to the problem of you FORCE highly competent people out because they don't want to be managers, it's a common problem in the US military among the officer corp as you have to promote excellent people to a unwanted level where they will fail.

Likewise every organization will just game it. For example there is a five year rule for OCONUS DOD employees, they game it so they can spend 20 years OCONUS straight by simply changing there job classification every 5 years between exempted and non-exempted so it continually resets or by their "home" organization just sending them back stateside to a "get paid to sleep on the job" assignment for 365 days and then rehiring them right back in the same position they left open just for them.

And that goes for private sector positions that interact in the government. Contractors, etc are just mercenary bureaucrats and folk already bounce back and forth between them. Private groups which get government grants, NGO's, etc are the exact same including effective government only on paper but not in practice policy and rule formulation, interpretation, etc as the government employee at the top just pencil whips it all signing whatever as she also will eventually enter that same cycle to double/triple dip on pay/retirement.

Also while churn can be beneficial, it also introduces issues such as limit competition as you get people that won't commit while increasing the amount of people of government benefits as they can't find meaning employment whereas before they had a made work program. You forget a large number of government employee exist purely for make work and that is a policy a goal, they intentionally aren't effective. For example we had a GS-5 mail clerk who was mentally traumatized in his early 20's to the point he was basically not able to function in society and work. He was hired on as part of a program, now defunct I believe, back in the 1980's, to give disabled people a meaningful life, i.e. either way us taxpayer were going to give him $50K taxes a year so why not just give him a job where he can sort literally seven pieces of a mail a week and feel like he contributed to society. Deadweight is is not uncommon in the USG and it's intentional, remember public policy goals are not about efficiency.

Oversight is really the only thing that will fix this and that is hard. Downsizing doesn't really solve the issue, it just lowers the number of interactions you will have with them but you still run into the same problem if you have to. You are always going to have dead weight and bureaucracy is going to always protect it's own even if that means as even more increased gov/private sector revolving door where the "private" is gov in all but name, i.e. Budget Officer Smith (USGS) and Budget Officer Smith (USGS "Contractor") are the same thing functionally.

All that said one practical thing which would help and it's related to your churn, would be enforced heterogeneity not so much in hirings but local office staffing and it' not be by "protected categories" but by socioenomic background and geographic subcultures. Many of the bureaucratic problems folk see is simply locals get hired and stay there for thirty years implementing policy not as Washington direct but to whatever their local community says even in direct violation of actual policy. A practical example for example would be something like the SCOTUS where each nine members are effectively identical on all but irrelevant wedge issues hence groupthink is rampant, i.e. all were lawyers, all graduated the same couple schools, and all outside of protected categories basically have the same lives / career background / paths and are interchangable. Imagine if we required to SCOTUS to have nine objectively independent dissimilar people, i.e. Yale Prosecutor from New York, a high school drop out and former meth head from California, a dirt farmer from Kansas, etc. You could have a low bar like "must be literate" or "understand English" but that is a bar too high for even our current members who don't understand words like "not" or "no".

Expand full comment

Just because someone is a government employee and thus technically part of the "civil service" doesn't make one a "bureaucrat". A state hospital may directly employ an oncologist who is better thought of as primarily an expert professional service provider where years of experience and job security matter, but they are only incidentally participating in the bureaucratic roles of planning, administering, collecting and analyzing information, writing reports, procuring, budgeting, making rules and coordinating enforcement.

Expand full comment

Except they aren't. The overwhelming majority of people's complaints with the bureaucracy are not "Deputy Director for XYZ" as they don't interact with that. Their complaint is the customer facing government employee, i.e. bureaucrat, they interact with that arbitrarily and capriciously interpret and apply said policies. For example when you are at the VA Clinic it's not SES3 career non-appointment Vice Deputy Jim who wrote the policy you are getting mad at, especially because the policy you have in your hand with you probably authorizes whatever it is you want, it's GS-13 career non-appointment Dr. Beth who is refusing to follow said policy or is it interpreting in a bad faith as a way to deny you care.

All government employees are bureaucrats or, if you want to narrow is slightly, all government employees that have the ability to make a decision which can impact any member of the public is which is overwhelming nearly all of them. Sure you can find an exception for "PhD cryptanalysis genius living in his NSA bat cave doing nothing but hard math and not interacting with policy in any way at all" but that's the irrelevant unicorn which doesn't invalidate the rule.

Expand full comment

You might be stretching it a bit I basically agree.

Expand full comment

Type E error avoidance is the main benefit of bureaucracy, whereas Type B errors are not the only cost and may not be the main cost. Other costs include the time spent on navigating bureaucratic processes that could have been spent on directly productive work, and the morale cost of living with a bureaucracy-heavy institution. You could frame these as errors, and some of them do spawn errors indirectly, as when people are too psychically beaten down/non-agentic to even consider innovating. But intuitively they seem like a different, additional kind of cost.

I would assume that Philip Howard's new short book which Tyler just raved about on MR touches on these issues. Maybe it should go in your review queue?

Expand full comment

A good time to reup Arnold's COO/CA framework. An institutional check on bureaucratic inertia and cultural drift without losing the benefits of Type E error mitigation.

Expand full comment

We already have great auditing, go read any report from GAO, CRS, or a departmental-level OIG. It doesn't matter, because auditors are powerless and without power no one cares what they have to report.

Expand full comment

"I imagine that Type E errors are more of a problem than type B errors, so that a hospital is better run with bureaucratic rules than as a free-for-all where anyone can administer medication."

Whenever an institution pays for Type E errors but doesn't pay for Type B errors it will have a bias. For instance, the FDA's incentive structure on approving new drugs is clearly biased towards avoiding Type E errors while it pays no price for its Type B errors (in fact it is a type B cost center self interested in creating type B costs).

Of course one could imagine the opposite bias, where it can keep the gains from good type E successes but avoid the costs of bad type E errors. This is how limited liability works, let alone too big to fail.

Expand full comment

"For every middle manager with a fantastic idea for a new product or a new process, there are dozens of middle managers whose ideas will cost the business a lot of time and effort to try and will fail to pan out. The bad ideas that a business tries are Type E errors. The bureaucracy helps to filter out the risks that are not worth taking, even though it misses out on some of the good ideas."

So when is this actually a simple trade-off, and when do you end up with the "Marvel movie problem"? In Hollywood in the last 15 years or so, the industry as a whole is incentivized to play it very safe, not take risks, and pretty much all of the business is remakes and comic book universe blockbusters. You end up with, if not a race to the bottom, a sinking to a far lower level of equilibrium with no way up. (I actually think the descent hasn't stopped in this instance but that's a personal judgment)

Is modern Hollywood more unique, or a particularly obvious example of a pattern that's a lot more widespread, both in larger corporations and in government, universities and non-profits? If the latter, is the intuition that it's been getting significantly worse in the last decade or two credible?

Expand full comment

The French-Greek root of the word is not promising. The biggest mistake is to think that this function can deliver economies of scale via centralization (a classical French approach). It normally becomes a self referencing body with the main goal of perpetrating and broadening its power. Normally also highly correlated with high taxation at federal level. Much better to look at places like Switzerland where the system is highly decentralized consistently with high taxation at municipal and cantonal level. There you see a direct control exercised by citizens and guess what, bureaucracy works quite well.

Expand full comment

The Bureau itself has become the boo word. That’s too bad. Noah Smith has offered recent pieces of the puzzle recently arguing for more (better) bureaucracy. His attack on the ~$1 trillion non-profit sector as being the endlessly grifting heir-apparent to erstwhile governmental services (in the U.S.) offers a unique angle of understanding. In retrospect regarding decades of abuse of the word “bureaucracy”, these more universal (less politicized) arguments seem to be making headway. Entering the next decade characterized by our current age demographic dynamic, the discussion will only get louder.

Expand full comment

As someone who is surrounded by both government workers and contractors, it's not clear to me that either is superior to the other. The fundamental issue is what we are paying them to do, not whether they are on the GS schedule or not.

Expand full comment

Sorry - that’s a lot of words to say we’re gonna need a better bureaucracy or, none at all. I don’t think the later is an option.

Expand full comment

Let’s summarize the basics and then propose solutions.

1. Bureaucracies perform poorly because they aren’t facing the right set of incentives.

2. The solution isn’t to destroy them, so says Kling, but rather to?

3. Create a better set of incentives. Drive out ineffective bureaucracy by making the processes serve the users - all users, all stakeholders.

4. But the devil is in the details, so how exactly will this work?

5. We need a for-profit business venture or another branch of government that “works for all stakeholders” to create a better set of incentives for bureaucracies. Its sole mission is to improve bureaucracy. This would be different than a think tank and different than a politician running for office. Different than the Supreme Court. The business venture or new branch of government needs a longterm vision and must bring all stakeholders into consideration. It can’t be captured by concentrated special interests.

6. How to make this happen? What form would it take? Many small firms/branches located in each state? Or one big firm/branch serving everyone?

Considered this a continuation of our republican form of government in which a balance of power is maintained by competing organizations.

Expand full comment

Competing organizations doesn't accomplish anything in government anymore. Maintaining the administrative state while restoring checks and balances requires a radically different approach to the fundamental way we structure organizations, control information, and assign personnel.

Expand full comment

It’s hard to be positive when inundated with negative news and dialogue. But let me try to lift your spirits a bit. When you get a chance, sit down and ponder the history of concealed carry here at Wikipedia. Then look at other countries. Ask yourself some questions. Am I being pessimistic? What am I grateful for? How well does the separation of powers work?

https://en.m.wikipedia.org/wiki/History_of_concealed_carry_in_the_United_States

Expand full comment

Ask yourself whether you're being too optimistic. I have a permit, but if you ask someone who works in firearms law for the real deal, they'll say that in many jurisdictions if you actually fire your weapon in self-defense, you are increasingly likely to be treated worse than your attacker by the local criminal justice system. Some right that is.

Expand full comment

It helps to live in a red county within a red state. Are you in red for both? I’m red for both and that makes a huge difference, but I admit, it’s a full-time job trying to be optimistic. If you tell me what your biggest concerns are, I might be able to point you in a better direction. And I’m not saying red is a panacea, but it is better than blue.

Expand full comment

This post also reminds me of a recent Econtalk with Mike Munger, “The Perfect vs The Good.” In this case how are we going to get to a better bureaucracy? Are you a directionalist or a destinationist?

Here’s the episode description.

Is the perfect really the enemy of the good? Or is it the other way around? In 2008, Duke University economist Michael Munger ran for governor and proposed increasing school choice through vouchers for the state's poorest counties. But some lovers of liberty argued that it's better to fight for eliminating public schools instead of trying to improve them. Munger realized his fellow free-marketers come in two flavors: directionalists--who take our political realities as given and try to move outcomes closer to the ideal--and destinationists--who want no compromises with what they see as the perfect outcome. Listen as Munger talks to EconTalk's Russ Roberts about two different strategies for achieving political goals. Along the way, they discuss rent control, the minimum wage, and why free-market policies are so rare.

https://www.econtalk.org/michael-munger-on-the-perfect-vs-the-good/

Expand full comment

Here’s what I’d recommend after working with the bureaucracy for 20 years. First Merit based civil service test like we used to have. Second, salary scales geared to Silicon Valley for technical talent. Third, immediate hire authority for anyone who passes the merit test. Fourth, end of any hiring preference programs. Fifth, creation of a parallel procurement system for new tech that allows for automated certification for meeting cyber and privacy requirements and then allows for no protest bids up to some limit. I’d also probably vastly expand the digital service and run it like DARPA. Private sector oriented innovators who come into government for 2-3 years. Help agencies stand up new tech and then have to leave. It would also create a cadre of experience innovators to start businesses outside of gov.

Expand full comment

Why would we want good people (high merit) to waste their talents in bureaucracy? My gut reaction isn’t that you will get better bureaucracy, you will just get more powerful bureaucracy and less talent where it is really needed in production, innovation and service.

Wouldn’t it just be easier to require civil service bureaucrats to be outsourced and thus hire from better firms?

Expand full comment

Believe me, you can't successfully outsource what you aren't smart enough to understand or assess. The government actually does this a lot, and it goes about as poorly as you'd expect.

Expand full comment

But then doesn’t the challenge shift from creating a bureaucracy to creating the organizational capacity to understand/oversee/assess? I don’t see these as the same things.

Expand full comment

To use a worn-out phrase I heard more often than I liked during my working career, the trick is “right-sizing” the organization’s bureaucracy (or the less distasteful term “administration”). Finding the Goldilocks zone, of course, hinges on a host of variables. Overall scale and scope of the organization, type of business, private enterprise or governmental, required skill sets to function, etc., will also dictate more or less bureaucracy. IMHO, while admitting it is necessary, I’d tend to favor less rather than more, since I agree with Christopher B’s comment that it commonly has “a bias towards inaction…rather than problem-solving” (and propagating busywork, I’d add). As for how to reduce established bureaucracy, selective tactics seem in order to me. For example, nuclear is the clear choice for DEI departments, while HR in general could be selective.

Expand full comment

"Getting rid of bureaucracy is not a shortcut to utopia." No of course not....nothing is either a shortcut nor a long-haul to Utopia. But since you have focussed on upsides of bureaucracy, it is only fair to flag some downsides:

* governmental bureaucracy has emasculated democratic pluralism by entrenching a permanent Progressive regimen irrespective of who is nominally 'in power'.

* the main agenda of any bureaucracy when it gets too large eventually becomes....itself

*bureaucracies are a magnet for the lazy and work shy among us.

Expand full comment

China is an example of bureaucracy gone wrong.

Expand full comment

As you say, there is definitely a balance between bureaucratic control and empowerment. I see similarities with federal vs state policy and state vs local policy. Implementing at the lower level can be a little less bureaucratic and more empowering but more importantly allows easier experimentation and tailoring to regional or local needs. Of course that also means some places will do it really badly too. And it also suffers from a gap in services between wealthy and poor locales as well as generous services attracting too many and ungenerous being a way of chasing away those same people. Liberals tend to be on one side of this, conservatives the other, but neither is clearly better.

Expand full comment

Government bureaucracy is different.

Most organizations have "internal governance" bureaucracies, indeed, in what is not as much as a mere metaphor as you might think, all complex organisms have internal biochemical bureaucracy tasked with regulation and coordination and doing things for the cell, organ, or body that bureaucracies do for subject populations. Indeed, one talks of a corps or corporation as a metaphor for a 'body' with head to think, mouth to eat, arms to execute, etc. "Organization" and bodily "organ" (and organelle) come from the same root as the musical organ in terms of being a instrument for a specific function.

The strictly internal governance bureaucracy of the state is its own problem, but what people here seem to mostly be complaining about are the parts of the state bureaucracy trying to govern -them- that is, the external population and economy.

"You are doing a bad job being a tree" is a lot different from "you are doing a bad job at forestry management".

Expand full comment