Previously in sequence: Moloch Hasn’t WonPerfect CompetitionImperfect CompetitionDoes Big Business Hate Your Family?What is Life in an Immoral Maze?Stripping Away the ProtectionsWhat is Success in an Immoral Maze?

Immoral mazes (hereafter mazes), as laid out in the book Moral Mazesare toxic organizations. Working for them puts tremendous pressure on you to prioritize getting ahead in the organization over everything else. Middle managers are particularly affected – they are pushed to sacrifice not only all of their time, but also things such as their morality, family and ability to think clearly. Only those who go all-in doing this get ahead, and even most of them fail.

Even successfully getting ahead is little consolation.

Mazes exert similar pressures on those who do business with them or work in non-managerial roles, to a lesser but substantial degree.

The best defense is to identify mazes before you agree to work for or do business with them, and choose to work or do business elsewhere. At a minimum, one’s eyes should be open, and the costs involved must be fully factored in before making such decisions.

This makes it important to figure out what parts of what organizations are mazes, and to what extent. This is hard to get exactly right.

What is easier is using simple heuristics to get a good approximation, then keeping an eye out for and updating on new evidence.

I offer seven heuristics, the first two of which will do the bulk of the work on their own. You benefit from the ‘right’ answer to all of them even absent concerns about mazes, so they are good questions to get into the habit of asking.

1. How many levels of hierarchy exist?

Full mazes require at least three levels of hierarchy, without which one cannot have middle management.

Each level beyond that makes things worse. The fourth and fifth levels both make things much worse.

With only one level, there’s nothing to worry about.

With only two levels, a boss and those who report to the boss, the boss has skin in the game, no boss causing problems for them, and not enough reason to reward bad outcomes.

With three levels, there are middle managers in the second layer, so one should be wary. But things are unlikely to be too bad. No middle manager has a boss or underling who is also a middle manager. This means that in any interaction between non-equals either involves the head of the company, or it involves someone ‘on the line’ who doesn’t have anyone reporting to them, and must deal with object-level reality. Either of them has reason to keep things grounded. Since there is only one person at the top, every conversation includes someone who interacts regularly with object level reality.

With four levels, we start to have interactions between middle managers in charge of each other. These dynamics start to get serious, but everyone still interacts with someone on the top or bottom.

At five levels, we have people who never interact directly with either the boss or anyone dealing with the object level.

At six levels, those people interact with each other.

And so on.

Meanwhile, the boss has less and less need or ability to comprehend the object level, and we get more and more problems with lack of skin in the game, which is question two.

At least one of the corporations in Moral Mazes had more than twenty ranks. That is way, way too many. By that point, it would be surprising if you weren’t doomed. I have actual no idea how to have twenty ranks and keep things sane.

Note that those outside the company, such as investors or regulators, seem like they should effectively count as a level under some circumstances, but not under others.

As a spot check, I looked back on the jobs I’ve had. This matches my experience.

Most impressive is that I can observe what happened when several of those jobs added new layers of hierarchy. This led in every case to traceable ways to additional maze-like behavior. In every case, that made life much worse for me and other employees, and hurt our productivity. In one case I was running the company at the time, and it still happened. 

I would be very wary of any organization that had four levels of hierarchy. I would be progressively more skeptical of any organization with more than that, to the point of assuming it was a maze until proven otherwise.

2. Do people have skin in the game?

Skin in the game is a robust defense against mazes, if it can be distributed widely enough and in the right ways. That can be tough. There’s only 100% total equity to go around.

One can only reward what can be observed or often only what can be quantified and measured. Something about Goodhart’s Law, and so on. The problem with levels of hierarchy and middle management is in large part a problem of inability to provide skin in the game.

For sufficiently large organizations, as described in Moral Mazes, skin in the game is not so much spread thin as deliberately destroyed. The successful keep enough momentum to run away from the consequences of their problems. This alone is fatal. 

If an organization has solved these problems for real, it likely isn’t a maze. 

If an organization lacks skin in the game and also has many levels of hierarchy, you’re almost certainly dealing with a maze. 

If it lacks skin in the game but also lacks levels of hierarchy, maze levels can differ. But also keep in mind that lack of skin in the game causes a whole host of problems. Only some of those are the problems of mazes. Detailing these issues is beyond the scope here, but be highly skeptical whenever skin in the game is lacking.

3. Do people have soul in the game?

What’s better than having skin in the game? Having soul in the game. Caring deeply about the outcome for reasons other than money, or your own liability, or being potentially scapegoated. Caring for existential reasons, not commercial ones.

Soul in the game is incompatible with mazes. Mazes will eliminate anyone with soul in the game. Therefore, if the people you work for have soul in the game, you’re safe. If you have it too, you’ll be a lot happier, and likely doing something worthwhile. Things will be much better on most fronts. 

It’s worth prioritizing soul in the game, above and beyond skin in the game.

4. How do people describe their job when you ask?

Remember this quote:

When managers describe their work to an outsider, they almost always first say: “I work for [Bill James]” or “I report to [Harry Mills]” or “I’m in [Joe Bell’s] group,”* and only then proceed to describe their actual work functions. (Location 387, Quote 2)

You want them to say almost anything else. Anything that does not make you recoil in horror a different way. Hopefully something worthwhile and interesting. I don’t know how good this rule is, but I suspect it’s quite powerful.

5. Is there diversity of skill levels? Is excellence possible and rewarded?

The belief that all middle managers have the same skills, and are all equally capable of doing any managerial job aside from the politics involved, is a lot of what makes mazes so bad. If there is no good reason to diverge from standard practice, if everybody knows that you cannot do better, then any divergence is blameworthy, and shows you are not doing your job. There’s no need to ask why, or what advantages it might have. 

It also all but ensures the wrong answer to the next question.

6. Is there slack?

A world without slack is not a place one wants to be. Mazes systematically erase all slack. Slack is evidence of not being fully committed, and given that everyone’s skills are equal and competition is perfect, holding anything back means losing even if undetected.

7. Pay Attention

Sounds silly, but it works. Observe people and what they do and how they do it. If you work in a maze for long enough, you’re not going to shout it from the rooftops, but every sentence you speak will reflect it.

And as always, when people tell you who they are, believe them.

Other Notes

These questions do not differentiate between corporations, non-profits, governments, parties, clubs or other organizational forms. That’s not a good indicator. Corporations are only the original observed case.

Asking how proposed or expected changes will change the answers to these questions is a good way to know if those changes will raise the maze level of an organization. 

Like most puzzles, there are multiple solutions, and the pieces reinforce each other. Most of the time, hardcore mazes will give alarm-bell level answers to all seven heuristics. 

Are there any other good simple heuristics? 

Next is How to Work With Moral Mazes, providing my best advice in detail to those dealing with the threat of mazes on a personal level.

New Comment
19 comments, sorted by Click to highlight new comments since:

What’s better than having skin in the game? Having soul in the game. Caring deeply about the outcome for reasons other than money, or your own liability, or being potentially scapegoated. Caring for existential reasons, not commercial ones.
Soul in the game is incompatible with mazes. Mazes will eliminate anyone with soul in the game. Therefore, if the people you work for have soul in the game, you’re safe. If you have it too, you’ll be a lot happier, and likely doing something worthwhile. Things will be much better on most fronts. 

In general I worry that the advice you are giving is phrased too confidently. This quote about soul in particular stood out to me. I have a few friends who have worked for big hierarchical non-profits, and their experience seems to contradict it. Plenty of people who do seem pretty passionate about 'the cause' and yet lots of dysfunction, bureaucracy, office politics, metric-gaming, etc. Maybe these problems didn't rise to the level of a true moral maze, or maybe the people weren't actually passionate but really were just posturing. But maybe not, and at any rate how do you tell at a glance?


Overconfidence is a reasonable thing to worry about. This in particular, that finding people with soul makes you safe, does seem likely to be too strong a claim. It certainly greatly helps your odds, versus the alternative. But yes, it does seem plausible that some passionate people surviving could be compatible with remarkably high maze levels, especially if those people provide good out-facing looks or are willing to do absurd levels of grunt work for low pay as a result of their passions.

One key is that this asks about the people you work for, not the people you work with. That's an important distinction. Mazes are fine with object-level workers having passion, and even prefer it, since they'll work harder and more reliably for less pay, and complain less, and so on.

Even if they were posturing and not truly passionate, that leaves the question of how to tell from the outside before entering an organization.

This is mostly a terminology quibble, but I guess it also means I've been mildly misreading previous posts: I wouldn't say you have middle management until you reach four levels. That is, my usage (and also Wikipedia's) of the term would be "people who manage managers, and are also managed" while you seem to just mean "people who manage, and are also managed".

At least one of the corporations in Moral Mazes had more than twenty ranks.

But per the comments on the previous post, that's not necessarily (and probably not) twenty levels of hierarchy, right? In that someone might be given a promotion to a new rank and pay grade, while continuing to manage exactly the same set of people who themselves have exactly the same set of responsibilities.

In which case it's not clear how relevant it is, and it seems misleading.

Do you have thoughts on "how to find/create acceptable modern corporate paths"? Or how to set some Schelling lines for those of us who want to avoid the worst, but do find value in working for large companies? Your list is good in terms of characterization (and thank you for it), but might benefit from some mechanisms/measurements.

My thoughts:

  • Have 1:1s with your manager AND your skip-level manager (and in some cases, higher), which include both personal and team/org/company goal contributions.
  • Encourage your reports to talk to your boss, and to hold you accountable for the direction you're giving.
  • Have written goals for the company, division, group, team, sub-team, etc. Document and get agreement (in the same way you do for the goals) how the sub-goals aggregate to the total goals.
  • Do a lot of writing, both in reports and metrics up and down the chain, and in longer-term narratives (annual or semi-) about why these are the right metrics for your shared goals. Demand this of your reports and managers as well.
  • Be open with yourself and management chain (and at least some peers) about the cases where goals are contradictory or divergent, and the fact that you are picking weights between them.

These are difficult and uncomfortable, that part is normal. If they're impossible or harmful, you're probably in a maze and you should get out. I think a one-liner is "be wary of things that you believe but can't say to your boss, peers, or repots").

How does this work in the military? They have a very deep hierarchy: is life in the army above private and below commander-in-chief also a maze?

Zvi listed multiple points. I think that it's possible that there are militarizes where the soldiers have a strong belief that they are fighting for the right thing and thus have a lot of soul in the game. A good chunk of the military hierarchy also has skin in the game. Very few people in the military hierarchy would first tell you about there supervisor if you ask them to describe their job.

One factor is that the military has a pretty consistent policy of moving officers around to different postings every few years. You never work with the same people very long, except maybe at the very top. This might help enable some of the outrunning-your-mistakes phenomenon mentioned above, but it also probably means you can't develop the kind of interpersonal politics you might see in a big corporation.

When you say "the military" do you mean "the US military" here? I would be surprised if that's a consistent phenomena over the different militarizes that exist.

From the long-term military personnel I personally know, it seems very similar to my corporate experience: there are elements of it, but not to the degree described in the book or these posts. I suspect it varies as widely as corporate experiences do - some paths are horrible, some are pretty good.

Having extremely specific goals and expectations goes a LONG way toward avoiding the maze aspects of management.

Is there some reason you don't think it would be?

[+][comment deleted]10

This was the post I was most personally looking forward to. I think it lays out of a number of gears that are easy to reason about, that seem useful regardless of whether the entire thesis hangs together. 

The issue with layers-of-management was my most important update from this sequence, and "what to do about that?" seems like one of the most important questions for groups of people trying to put a dent in the universe.

An open question in my mind is something like "how much better is it to found a new organization rather than open up a new department at an existing institution", in particular because it's not obviously better if you end up with an ecosystem that has less legible "management layers" in between organizations. (Zvi notes that investors may-or-may-not count as a management layer and it depends. "What does it depend on?" is my next question)

Something I note: I think this post almost-but-not-quite-stands alone. I think the rest of the sequence was necessary to make the overall points Zvi is meaning to make, but I think there's a narrower set of points that this post is making that only depend on you roughly having a sense that there's a special kind of middle-management-hell that can exist. 

I think you could summarize that in a few paragraphs and then have a pretty good standalone post, with pointers to the rest of the sequence for people that want to delve into the broader argument.

When counting levels of hierarchy, it's not necessarily clear where to stop. For DeepMind, would you stop counting at the CEO of DeepMind, or continue through to the CEO of Alphabet? Or to take my previous job, I'd say there were four plausible stopping points (the respective heads of Universal Pictures International, Universal Pictures, NBCUniversal, and Comcast; though I could be misremembering the structure such that either of the first two is not plausible).

My impression is that when a startup gets bought by a large company, the startup typically turns to hell; and this would point towards counting the levels in the parent. But I also suspect that would be too pessimistic on average (i.e. adding a level above the CEO counts for some amount, but less than a level below the CEO). This is super not confident.

Curated, after chatting a bit with Zvi about a better intro.

I noted previously that this lay out gears that seemed clear and easy to reason about, which seemed quite useful. The concepts here seem quite important to consider for anyone building an organization, or organizational ecosystem. This seemed like a significant hack-away-at-the-edges of the questions that originally prompted Inadequate Equilibria

I'm interested in followup work that checks into the "Moral Mazes are Common and Damaging" hypothesis a bit more empirically (how common? how damaging?)

More sanity-checking "20 layers of hierarchy"

If a each person at each layer of hierarchy reports up to one person and has two people that report to them, 20 levels of hierarchy means 2^20 -1 people, over a million. A quick Google suggests... Wal-Mart, McDonalds (including franchise employees), China National Petroleum, and some government actors as potential groups of that size. If we call "team lead" and "Shift manager" different levels of hierarchy, then we get a ratio of about 4:1 across the first middle manager (who is an hourly employee that doesn't make business decisions about who to hire). If we assume that ratio holds, Wal-Mart could have 11 levels of hierarchy with their employee total. To have 20 levels of heirarchy with each level having 4 direct reports would require 4^19+1 people in the organization, which would require a larger population. (while it isn't strictly necessary that every branch be fully populated, the termination of a branch represents people 'on the front lines', which affects how distant from the front lines their immediate superior is)

In US government hierarchies, the GS payscale has 15 grades, and each position within the GS structure reports to a higher paygrade individual, and not all of the grades are actually usable. If we add the Senior Executive Service employees, political appointees, and elected officials, there are almost 18 possible paygrades above the blue collar FWS payscale employees. Outside of the US government, it is of course possible for people on different hierarchical levels to have the same paygrade, so a paygrade count is not always an upper limit on the number of levels.

I'm worried that people will see "One of the organizations studied reached 20 levels of hierarchy without falling over under it's own middle management, therefore this organization that has merely seven is not particularly bad". Instead of counting layers, I would instead ask "What responsibilities does this layer actually have- what buck stops here?" In the retail sphere: A shift lead decides when everyone's breaks are; the shift manager decides whether to approve overtime; the store manager makes hiring decisions; the district manager decides which stores to close; the regional manager... might lack specific purveiw, but there are too many districts for each of them to report directly to the CEO, and there are a few tiers there. My alternative would be "If this position serves mostly to consolidate several lines of a spreadsheet into one line on a different sheet, they can and should be automated with a cell formula"... except that policy would result in every middle manager instantly finding something that they can do to justify their position, and start actively doing that thing, resulting in massive shock to an already stable system.

This was a great followup, thanks!

I don't like the specific description of levels of hierarchy for reasons I'm not quite certain about. This is at least partly just the phrasing, not the deeper point.

One piece of this is that, as is mentioned in some other comments, not every level of management has the same size. For example, if I regularly speak with my manager's manager and they have both explicit awareness of the object level work I do and share some responsibility for the outcome of that work, are they really two levels above me? What if I only talk to them once a month? Once a quarter? There may be some calculus that one can use to compute whether these count as 2 levels or 1.27 levels, but it seems like it has a strong interaction with the other criteria, such as how much slack I'm given, whether excellence is measurable, and who has skin in the game.

Maybe this is an optimistic take based on good luck with the team I currently work on, but my expectation is that usually there will be more levels of hierarchy than there are levels of non-interaction; that is, I expect most actual ranked titles to correspond to 1/2 or less of a level of hierarchy, which makes it a bit difficult to measure the depth of an organization.

Maybe this is just pessimistic of me though, and it's easy to find an up to date, readable org chart when joining an organization? That doesn't match my experience though.

Anyway I don't really feel satisfied that I've found my true objection but maybe this will help someone else or future me identify something.