A fully meta-rational workplace

(What would that be like?)


Here’s a description of a utopian workplace:

There is no overall plan for anything. No one even knows what the organization is supposed to be trying to do. It’s completely unclear who is responsible for what. People frequently take on tasks they have no qualifications for, just because they feel like it. No one understands the work they are doing, and no one thinks that’s a problem. There are frequent arguments about what to do, with no procedure for coming to a decision, and no one has the authority to conclude one.

To you, that sounds like a nightmare, because you are rational. But for many pre-rational people, it’s obviously the way a workplace should work.

This is a utopian anarchist collective! Everyone does what they think is best at the time, there’s no boss, no schedule or arbitrary top-down job assignments, and problems are hashed out by consensus discussion instead of someone issuing commands.

The anarchist collective is an intuitive attempt to recreate the archaic choiceless mode, the natural way of human being, but without understanding that timeless mythology and fixed social roles are prerequisites to the mode’s functioning.

As a rational person, you have noticed that anarchist collectives hardly ever get anything done. But, maybe they are at least pleasant to work in? No, typically they spiral down into hellish political infighting. “Consensus decision making” becomes a contest to see who can be most theatrically emotional. Eventually someone commits suicide or murder or has a psychotic break, and the collective splits up. Then everyone goes and joins or starts another one, and the cycle repeats.1


Thank god for rational, well-managed, modern workplaces. There’s a documented strategy, derived from clear institutional objectives. A standardized planning process sets performance goals for each team. For individual contributors, there’s clear accountability for well-defined responsibilities. There’s an org chart with specific qualifications for each job. You don’t get put in a position unless you know how to do the work. Executives make decisions on the basis of quantitative risk and benefit models, ensuring the institutions’s continued survival and growth (so you’ll still have a job next year).

A fully meta-rational workplace

Here’s a description of a fully meta-rational workplace:

There is no overall plan for anything. No one even knows what the organization is supposed to be trying to do. It’s completely unclear who is responsible for what. People frequently take on tasks they have no qualifications for, just because they feel like it. No one understands the work they are doing, and no one thinks that’s a problem. There are frequent arguments about what to do, with no procedure for coming to a decision, and no one has the authority to conclude one.

To save you checking: that is, indeed, identical to the “nightmare” scenario.

Let’s say you are hired into that organization. Your first days there are a shock. “This place is insane; what’s it going to look like on my resume if I quit immediately?”

Before joining, you had read the job gossip on Glassdoor:

Yeah, I was there for a few months. They have a lot of smart, fun people, but it’s really chaotic. Project management is a disaster. No one would tell me what I was supposed to do. There’s no defined path to promotion. I want a job where I can keep my head down doing the work and meet my KPIs and get a raise at the end of the year. YMMV.

You joined because the company is known for repeatedly producing peculiar products that… well, the business press says “wtf is this, no one wants that, they’ve really screwed it up this time,” and a couple years later fanatical customers say they can’t imagine how they lived without it.

Deciding to stick it out for a while, after a few days you notice that despite apparent chaos, routine practical things run unusually smoothly. The employee intake paperwork is minimal and makes sense. You are used to employers taking ages to sort out your office and equipment and intranet accounts, due to communication snafus between the HR and Purchasing and Facilities and IT departments, so you couldn’t really get started working there for a couple weeks. Here, all that was already in place when you arrived.

The meta-rational workplace superficially resembles an anarchist collective because neither runs on rational principles. But all the systematic bureaucratic stuff, which a well-managed rational organization puts most of its attention and effort into, just works here with minimal fuss, because everyone is fully competent at rationality, so it goes without saying.

You aren’t “assigned to a team,” and in fact someone tells you on the first day that there aren’t any. But during the first week, you spend most of your time talking with people who are interested in your previous work. They’re fun discussions, because they’re also interested in their own work, and love talking about it, which gets you interested too. You begin to notice a pattern, as well: they are particularly interested in how you think and feel about types of work. And by the second week, you’re offering your advice, which they take seriously, and you’ve gotten sucked into helping with several different projects that are actually quite cool, and the sort of thing you can do parts of well—and you’re confident you’ll learn how to do more parts through participation in the work.

“YMMV.” Your mileage did vary from the Glassdoor reviewer’s. A few months in, having the time of your life, you reflect on what makes this workplace work.

There is no overall plan for anything, because it’s impossible to plan for discontinuous innovation. But individual projects often get detailed, rational plans as they move toward realization. People working on a project construct plans as they go, and ignore or revise them when they aren’t working, and no one makes a fuss about it. Plans, of different sorts, are just tools for efficient work, sometimes useful and sometimes not.

No one knows what the organization is supposed to be trying to do, but its members constantly discuss possible directions, what purposes are worthwhile, and what might be feasible in practice.

The organization’s conception of itself has changed several times over the years. What it’s trying to do next year might be different again.

It’s completely unclear who is responsible for what, because the nature of the work is constantly changing, and which projects are most important fluctuates unpredictably. However, everyone can mostly trust—based on experience—that others will dive in to help when needed.

People frequently take on tasks they have no qualifications for, just because they feel like it. No one understands the work they are doing, and no one thinks that’s a problem.

General competence” is a result of meta-rationality. That is the attitude that you are willing and potentially able to do whatever needs to be done.

A new employee will arrive tomorrow, their office is bare, and the facilities person is out sick. You don’t say “I’m a professional software engineer; this isn’t my problem.” You find a spare desk and bookshelf and move them into the new person’s room and make sure their lights and thermostat are working. You do that because you can see it’s the most important thing you can do this afternoon; it takes priority over the program you are debugging. It’s important for new people to feel taken care of. It’s also important to start them developing the implicit understanding that the easy stuff just works here. Then they’ll feel safe about working on scary difficult stuff, without worrying about getting interrupted by time-wasting nonsense. And also, weeks later when they notice something not working, they’ll take the initiative to point it out or just fix it, instead of letting an irritating problem fester. You can get the office set up, it needs doing, so you do it.

Someone on a project you’re working on realizes it would help to get an exact solution to a particular differential equation, but no one knows anything about those. You don’t say “I’m a software person; that’s not my kind of math.” You know that an undergraduate course in differential equations is months of hard work, so learning all that is right out. It would be better to hire a math consultant! But, “hmm, I’ll see what I can do,” you say.

You spend the afternoon reading stuff on the web. What is a differential equation, actually? What’s up with this particular differential equation? After a few hours, you realize that by rearranging it with some high school algebra, you can make it fit the form needed to apply one of the standard solution techniques.2 You verify the answer against the project problem, and take it back to the person who found it. “Oh, nice,” they say, and you both move on with the job.

General competence doesn’t mean you are expert at everything—no one can be. It’s based on a realistic assessment of what you can do now, and how long it may take you to figure out how to do something well enough outside your areas of expertise. The confidence of an expert is based on a self-definition as someone with complete knowledge of a circumscribed rational system. The confidence of the generally competent is based on absence of any self-definition. You do not identify as “a professional software engineer.” The boundaries of your self are nebulous, so you are not limited to any system. You are meta to all systems, and can juggle them as needed.

There are frequent arguments about what to do, because reflection on purposes is a major aspect of meta-rationality. That is often best done in conversation, especially in the context of a group project.

… with no procedure for coming to a decision, and no one has the authority to conclude one.

In a meta-rational organization, someone usually does have decision-making authority, but they exercise it only when absolutely necessary, so you rarely see it happen. A meta-rational leader introduces themselves like this:3

Hi, apparently I’m the boss here. Unfortunately, I don’t understand what’s going on, so I can’t make any decisions. I do understand that none of you understand what you are doing either, not more than partly.

When it comes to decision-making, it’s my job to help you talk to each other across specialisms, so you can all bring your partial understandings into the mix, and you can argue about it productively. It’s also my job to help you relate to each other meta-rationally,4 so you don’t let irrelevant feelings get in the way, and you also don’t try to win with rationalizing justifications or quantitative data of dubious relevance.

A meta-rational organization requires an unusual degree of interpersonal trust, because we hold principles and procedures lightly. We override rules when they don’t make sense. For you to feel safe operating meta-rationally, outside your zone of notional expertise and responsibility, you need confidence that decisions which affect you will get made for the benefit of our people and projects, not based on how I feel, nor some artificial A/B test. So it is up to you to figure out tough choices, together, meta-rationally.

If you can’t come to a decision together, it means I’ve failed at my job. Then I’ll have to make the decision, which will be no better than a coin flip, so the project may fail.

Fortunately, that would be fine, although a pity. There are endless cool things we can do together. Not all of them will work. Probably enough will succeed to keep the organization going. And if it doesn’t, that’s fine too.

Before utopia

A fully meta-rational workplace is a utopian fantasy—for now.5

The fantasy organization consists exclusively of fully meta-rational people. It’s not clear there are any of those currently; and if there are some, they are rare, so assembling a substantial number isn’t feasible. We might imagine a future in which everyone is meta-rational, but that’s not happening any time soon.

Even in an innovative research organization, there’s a lot of bureaucratic rational work, and practical circumrational work, to do. Someone has to administer the employee benefits plan, and someone has to keep the air conditioners running. Fully meta-rational people are also fully rational and fully reasonable, and therefore fully capable of both those sorts of work, but they find it tedious if it takes much of their time. It’s better done by people who enjoy it.

Conversely, few people would want to work in a fully meta-rational organization. You’d be constantly confronted with nebulosity: uncertainty, bafflement, and chaos—the negations of rationalism’s promises of certainty, understanding, and control.

Realistically, any organization’s members are diverse in many respects, including in their cognitive capacities. A good workplace provides both respect and enjoyable, meaningful work for all members, wherever they are at.

As a further step, a “deliberately developmental organization” can encourage pre-rational members to advance into rationality, and rational members to advance into meta-rationality. That takes a lot of support, which means expending considerable institutional resources. It can be a hugely valuable perk for members, perhaps inspiring loyalty. And given that people competent in technical rationality are expensive and difficult to attract, and the meta-rationally competent are extremely rare but can provide unique value, growing both internally may pay off handsomely.

  1. 1.Jo Freeman’s “The Tyranny of Structurelessness” is a brilliant analysis of this failure mode.
  2. 2.You vaguely remembered reading about programs that can solve differential equations, but your hunch was that finding one and buying and installing it and figuring out how to tell it to solve your equation might take longer than learning a tiny bit of math. Your hunch was correct.
  3. 3.There’s a substantial literature on meta-rational leadership in management theory. Some prominent theorists are Robert Kegan, John Seely Brown, Donald Schön and Chris Argyris, and Etienne Wenger.
  4. 4.See my story “The Cofounders” for an explanation of meta-systematic relationship in a business context. (In this essay, I’m blurring the distinction between “meta-systematicity” and “meta-rationality” for simplicity.)
  5. 5.I have experienced some workplaces that were partially meta-rational. Innovative tech startups and both academic and established industrial R&D labs have to be somewhat meta-rational, because their purpose is to do things that have never been done before, and which no one knows how to do. It’s difficult to keep that going, and even the best labs slide into routine mediocrity eventually.