One of the distinctions I make throughout my work is between binding and non-binding expectations.

Binding expectations constrain what a participant may legitimately do. A rule, for example, might require something, prohibit something, or establish a limit that must be honored. Non-binding information may still matter enormously. It might guide judgment, communicate a preference, express an aspiration, or call attention to something worth considering. But ultimately it leaves the participant free to decide what to do.

The distinction is intentionally binary.

Something either binds you or it doesn’t.

And this tends to make people nervous.

Human life, after all, is not binary. Situations are complicated. Context matters. Obligations can conflict. Language is ambiguous. People exercise judgment. Any attempt to divide the messy landscape of human expectations into two categories can therefore seem hopelessly mechanical—another Procrustean bed that achieves conceptual tidiness only by chopping off the parts of reality that don’t fit.

I understand the objection. I just think it gets the relationship between the binary and the complexity almost exactly backward.

The fact that reality is not binary is not an argument against making binary distinctions.

It is one of the reasons we need them.

Counting and Measuring

I found an unexpected way of understanding this in Norbert Wiener’s The Human Use of Human Beings.

Writing in the early days of cybernetics, Wiener distinguishes between what he calls digital and “analogy” machines—the latter being what we would now ordinarily call analog machines. His simplest distinction between them is also the most useful: digital machines count; analog machines measure.

An analog machine represents one continuously varying quantity through another. A slide rule, Wiener’s example, represents numerical relationships through physical distances. An old thermometer represents temperature through the height of a column of mercury. An analog speedometer represents speed through the position of a needle.

These systems preserve something of the continuous character of what they represent. There are always positions between positions.

But this creates a problem: measurement has limits.

Make finer marks on a slide rule and eventually the width of the marks matters. Make the rule longer and its physical rigidity becomes a problem. Eventually accuracy depends upon manufacturing tolerances, the steadiness of the instrument, and the ability of the human eye to distinguish one position from another.

Digital machines take a radically different approach. Rather than continually asking how much?, they repeatedly ask questions with discrete answers:

Yes or no.

On or off.

One or zero.

This sounds like an enormous loss of nuance. And at one level, it is. A switch capable only of distinguishing YES from NO contains less information than a measurement capable of occupying indefinitely many intermediate positions.

But Wiener points out the strange advantage created by precisely this limitation. In a digital machine, he writes, “the sharpness of the decision between ‘yes’ and ‘no’” allows information to accumulate in ways that permit the machine “to discriminate very small differences in very large numbers.”

That is the trick.

Individually crude distinctions can be combined to produce extraordinary precision.

The binary is powerful not because reality itself comes neatly divided into two categories, but because a sharp distinction gives subsequent operations something dependable to build upon.

Subscribe now

A Binary is Not an Ontology

Wiener is equally careful about the other side of the argument.

Neurons appeared to offer an appealing analogy to digital switches because they exhibit something like all-or-none firing. But Wiener warns against concluding from this that the brain itself must therefore be fundamentally digital.

Whether a neuron fires can depend upon thresholds that are themselves affected by continuously varying phenomena such as temperature and blood chemistry. A discrete output can therefore emerge from a much messier underlying system.

That leads to a distinction worth preserving far beyond computer science:

A binary distinction can be useful without being a claim that the underlying reality is binary.

We can use an all-or-none distinction without adopting an all-or-none theory of the world.

This matters because many objections to binary social structures quietly confuse these two things.

Suppose an organization establishes that purchases above $5,000 require approval.

The difference between a $4,999 purchase and a $5,001 purchase is not metaphysically significant. Nothing in the underlying economic reality suddenly changes as we cross $5,000. A $4,999 purchase might be reckless while a $5,001 purchase might be entirely sensible.

The world around the boundary remains continuous.

But that is not a defect in the boundary. The boundary was never intended to describe the ontology of purchasing.

It was intended to help someone navigate it.

The relevant question for the participant is no longer:

Is this purchase large enough, risky enough, unusual enough, or consequential enough that perhaps I ought to seek someone else’s approval?

It becomes:

Is it above $5,000?

The first question requires measurement in Wiener’s broader sense: an indefinitely contextual assessment of degree. The second permits something closer to counting: the participant can determine which side of a defined distinction they occupy.

Something has certainly been lost.

But something enormously valuable has been gained.

Binding or Not Binding

The same logic applies to the distinction between binding and non-binding expectations.

Imagine someone tells you:

Customer impact should be taken seriously when making this decision.

What does that mean for your agency?

Perhaps customer impact is merely something worth considering. Perhaps you are expected to weigh it heavily. Perhaps you may disregard it if other considerations seem more important. Perhaps certain forms of customer harm are absolutely prohibited. Perhaps you are expected to get someone else’s permission before proceeding despite them.

Those are substantially different structural conditions.

Yet the sentence itself does not necessarily tell you which one you are in.

The participant is left to measure.

How serious is “seriously”? How much discretion do I have? At what point does departing from this expectation become a violation rather than merely a judgment someone disagrees with?

Distinguishing binding from non-binding expectations introduces a sharp first cut through that complexity:

Must I honor this, or does the final judgment remain mine?

That distinction does not tell us everything.

It isn’t supposed to.

There can still be enormous complexity on either side. Binding rules can contain exceptions, thresholds, interpretive terms, and discretionary provisions. Non-binding guidance can be extraordinarily influential. Context still matters. Judgment still matters. Human beings remain human beings.

The binary answers one particular question because that particular question benefits enormously from having a determinate answer.

Where does my agency stop?

The Procrustean Objection Gets it Backward

This is why I now think the standard objection to binary distinctions accidentally supplies part of the argument for them.

Yes, social reality is continuous, contextual, ambiguous, and multidimensional.

Exactly.

A participant operating within that reality cannot continually reconstruct every relevant expectation from first principles. If every boundary remains merely a matter of degree, participants must perpetually measure: How important is this? How strongly do they mean it? How far may I go? When should I ask? What will happen if I don’t?

Binary distinctions provide footholds.

They turn selected questions of degree into questions of kind—not because the underlying world naturally contains those kinds, but because people trying to act within that world sometimes need a reliable distinction more than they need a perfect representation of its complexity.

This also helps explain something otherwise puzzling about explicit structure.

From the outside, a sharp rule can look more mechanical. From the inside, it can make action considerably easier.

The person designing the $5,000 boundary sees everything it fails to capture. The person trying to buy something sees something else: I know what I am authorized to do.

That is not necessarily the elimination of judgment. It can be the infrastructure that makes judgment possible without requiring every judgment to reopen every question.

And this is where Wiener offers an unexpectedly powerful defense of binary structure. The value of the binary does not depend upon pretending that reality resembles the binary.

Quite the opposite.

We introduce sharp distinctions because reality does not reliably provide them for us.

The mistake is not drawing a line through a world that contains gradients.

The mistake is forgetting that we drew the line—and then mistaking our navigational instrument for a description of reality itself.

A binary can be a terrible ontology and an excellent tool.

And when the question is whether an expectation actually binds the person expected to navigate it, that may be exactly the kind of tool we need.

There are two pieces of advice about rules that seem to contradict each other.

Don’t just tell people what not to do. Tell them what to do instead.

“Don’t leave dirty dishes in the sink” is less useful than “Put dirty dishes in the dishwasher.”

“Don’t send me a long explanation” is less useful than “Send me the recommendation and the three facts I need to evaluate it.”

A prohibition removes an option. It does not necessarily tell anyone where to go next. If the point of a rule is to help people successfully navigate toward something, saying no may not be enough.

So far, so good.

But there is another piece of advice I believe just as strongly:

When defining someone’s authority, say what they can’t do—and be very careful about saying what they can.

Suppose I tell an employee:

You may spend up to $500 on software, hire freelancers for design work, and schedule meetings with customers without getting my approval.

That sounds empowering. I have just explicitly given them three permissions.

But I may also have accidentally taken away a thousand others.

Can they spend $200 on books?

Can they hire a freelance copyeditor?

Can they interview a customer without calling it a meeting?

Nothing I said prohibits any of those things. But the moment I bothered to specify what they may do, the obvious question becomes: What about everything I didn’t mention?

This is one of the strange things about explicitness: making one thing explicit can change the meaning of everything left implicit.

There are names for pieces of this phenomenon.

In ordinary language, we routinely infer more from a statement than its literal content entails. Linguists and philosophers call this conversational implicature. If someone tells me, “You can have the soup or the salad,” I am unlikely to conclude that the steak is equally available but simply went unmentioned.

Law has an especially wonderful name for a related interpretive principle: expressio unius est exclusio alterius—roughly, the expression of one thing implies the exclusion of another. If a rule specifically enumerates A, B, and C, there may be reason to believe D was deliberately left out.

Computer science has another useful cousin: the closed-world assumption. What has not been established as true is treated as false.

Each describes something slightly different, but they point toward the same practical danger.

A list of permissions can quietly turn an open world into a closed one.

Before the list, I might reason:

I can do anything unless there is some reason I can’t.

After the list:

I can do the things I have been told I can do.

That is a profound change.

And it can happen without anyone intending it.

In fact, the more conscientious someone is about respecting rules, the more susceptible they may be to it. If my boss has explicitly told me I can spend $500 on software, I have to decide what that statement means. Why software? Why $500? Why tell me at all? If I assume the statement was meaningful, then its boundaries probably mean something too.

Soon I am asking whether books count as software expenses. My boss says yes.

Now I know books are allowed.

Next week I need a microphone.

“Can I buy a microphone?”

“Yes.”

Nothing particularly authoritarian has happened. My boss keeps saying yes!

And yet, somehow, we have built a permission culture.

That is because giving permission and establishing freedom are not the same thing.

Every time my boss answers one of these questions, they may reinforce the premise underneath it: their yes is what makes the action permissible.

This can become self-reinforcing. The more permissions an authority explicitly grants, the more significant the absence of permission appears. The more significant that absence appears, the more people ask for permission. And the more often the authority answers, the more firmly established the permission-seeking relationship becomes.

Which brings us back to our apparent contradiction.

Sometimes a prohibition leaves people stranded.

Don’t do X.

Okay. What should I do?

In those situations, positive specification can increase navigability. If I care not merely that you avoid one outcome but that you successfully reach another, telling you what successful action looks like can be enormously useful.

But sometimes the openness of the remaining territory is precisely what we are trying to preserve.

In those situations, positive specification can decrease navigability by making an open possibility space look like a menu.

Fortunately, we don’t have to choose between the two.

If You’re Writing the Expectation

Keep the List Open

If you are naming possibilities without intending to exclude everything else, say so.

Consider:

You may store the files in Dropbox or the company drive.

That sounds suspiciously like a complete list. But a tiny change does a surprising amount of work:

You may store the files in Dropbox, the company drive, etc.

Or:

You may use a secure storage service, such as Dropbox or the company drive.

Or even:

You may use any secure storage service. Dropbox and the company drive are two examples.

“Etc.” may not be the most profound contribution to organizational design, but sometimes it is exactly the tool the job requires.

Words like such as, for example, including, among other options, and or another appropriate option all perform a similar function. They tell the reader something important about the list itself:

This is illustrative, not exhaustive.

Say Who Gets to Decide

Sometimes keeping the list open isn’t enough.

People may understand perfectly well that there are other possibilities while remaining unsure who gets to decide whether one of them is acceptable.

This is where phrases like “use your best judgment” can do real structural work.

Don’t share confidential information outside the company. Otherwise, use your best judgment about how to communicate with clients.

Or:

These are examples, not an exhaustive list. For situations not covered here, use your best judgment.

Or, when the authority really is that broad:

Unless something is explicitly prohibited, you don’t need permission to do it.

These aren’t motivational slogans about trusting people. They clarify an expectation that would otherwise remain ambiguous. They tell someone where discretion resides.

There is an important difference between saying:

Yes, you may do this.

and saying:

This is yours to decide.

The first answers a permission request. The second may prevent the next one.

Don’t Let an Example Become the Default

Even when people understand that a list is not exhaustive, the examples we provide can acquire a special status.

If the policy says, “Use a secure storage service, such as Dropbox,” people may reasonably conclude not only that Dropbox is permitted, but that Dropbox is preferred.

Sometimes that’s useful. Sometimes it isn’t.

If it isn’t, say that too:

Dropbox is one example, but there is no preference for it over other options that meet the requirement.

Or:

These are examples, not recommendations.

Explicit examples create salience. If you don’t intend that salience to carry normative weight, it can help to strip that weight away explicitly.

Give People Somewhere to Go

Of course, sometimes the original advice remains exactly right: don’t just tell people what not to do.

“Don’t send a long explanation” may leave me genuinely unsure what you want from me.

But even here, positive guidance doesn’t have to become another rigid requirement.

You can say:

Don’t send a long explanation. A good default is a short recommendation followed by the key supporting facts.

Or:

Don’t send a long explanation. Consider leading with the recommendation and adding detail only where necessary.

Words like consider, a good default, typically, and one option is provide orientation without necessarily converting that orientation into a mandate.

Sometimes people need a rule.

Sometimes they just need a signpost.

It helps if they can tell which one you’ve given them.

Specify the Criterion, Not the Answer

There is one more possibility.

Sometimes people don’t actually need to be told what to do instead. They need to understand what makes an alternative acceptable.

Consider:

Don’t make the presentation too long.

We could replace that with:

Keep the presentation to 15 slides.

But perhaps 15 slides isn’t really what matters. What matters is having enough time for discussion.

If so, say that:

Keep the presentation short enough to leave at least half the meeting for discussion.

Now I know what success requires without being told exactly how to achieve it.

This is often a better alternative to enumerating acceptable actions. Instead of trying to list everything someone may do, specify the condition their choice needs to satisfy.

If You’re the One Encountering the Expectation

There is another side to all of this.

Navigability isn’t solely the responsibility of whoever wrote the rule. We also have agency in how we interpret and respond to the expectations we encounter.

Once we know that specification can have these side effects, we can become more careful about not manufacturing constraints that aren’t actually there.

Don’t Turn Examples Into Boundaries

If a policy says:

You may use Dropbox, Google Drive, or Box.

pause before concluding that OneDrive is forbidden.

Ask yourself what kind of statement you are actually looking at.

Is this intended to define the complete set of acceptable choices? Or is someone giving me examples of acceptable choices and failing to signal that clearly?

Those are very different things.

The mere fact that something has not been explicitly permitted does not necessarily mean it is prohibited. Yet it is surprisingly easy to read rules that way, especially when some permissions have been explicitly named.

Knowing about this tendency gives us another navigational option: don’t automatically infer a boundary from the edge of a list.

Don’t Give Away Judgment You Already Have

This matters even more when the rule is genuinely ambiguous.

The conventional advice is simple:

If you’re not sure, ask.

Sometimes that’s exactly right.

But sometimes asking for clarification can actually make things worse.

Suppose I read the Dropbox policy and ask:

Am I allowed to use OneDrive?

My boss says yes.

Problem solved.

Except something else has happened too. I have implicitly treated my boss as the person who must authorize my choice. An ambiguity in the rule has become a decision for an authority.

I may have just given away discretion I already possessed.

And my boss’s answer can create a little more precedent. Now OneDrive joins the list of things we know are permitted. When somebody encounters the next unnamed option, there is even more reason to ask.

In other words, permission culture isn’t produced only by people who insist on granting permission. It can also be produced by people who unnecessarily seek it.

Sometimes the responsible response to ambiguity is not to escalate it.

If the stakes are modest, no rule prohibits the action, and the decision is reasonably yours to make, make a reasonable interpretation and act.

Ambiguity does not automatically create an obligation to ask somebody else what to do.

If You Clarify, Clarify the Structure

Of course, sometimes the ambiguity really does need to be resolved.

But even then, there is a useful distinction between asking:

Am I allowed to use OneDrive?

and asking:

Is this list intended to be exhaustive, or are these examples?

The first asks someone else to decide my case.

The second asks what the rule means.

That distinction matters. The first may resolve one person’s uncertainty while reinforcing a permission relationship. The second can improve navigability for everyone who encounters the rule.

Better yet, if the answer is that the list was only illustrative, fix the rule:

You may use any secure storage service, such as Dropbox, Google Drive, or Box.

Now the next person doesn’t have to ask at all.

That is part of what good citizenship in a rule system looks like. It isn’t simply following the structures we have been given as carefully as possible. It also means noticing when those structures create unnecessary uncertainty and helping make them more navigable.

The Problem Isn’t Positive or Negative Rules

So the lesson isn’t “always tell people what to do.” Nor is it “only tell people what they can’t do.”

Both are occasionally excellent advice. Both are occasionally terrible advice. The deeper problem is that specification has side effects.

Naming some options can make unnamed options appear forbidden.

Giving permission can teach people that permission is required.

Giving examples can turn them into unofficial defaults.

Giving only a prohibition can leave people without enough information to navigate.

And sometimes specifying an answer unnecessarily replaces a judgment someone was perfectly capable of making.

But responsibility for those effects exists on both sides.

If you’re writing an expectation, don’t accidentally close possibilities you mean to leave open.

If you’re encountering an expectation, don’t accidentally close possibilities the rule itself leaves open.

Say etc.

Say for example.

Say use your best judgment.

Say this is yours to decide.

Call something a default when it is a default.

Call something an example when it is an example.

And when you encounter an ambiguity, don’t automatically turn it into a request for permission. Sometimes use your judgment. Sometimes clarify what the rule means. And when the ambiguity is in the structure itself, help fix the structure.

The language of rules does more than tell us what we can and cannot do. It shapes how we interpret everything the rules don’t say.

Learning to write that silence carefully is important.

Learning not to fill it unnecessarily may be just as important.

A lot of what counts for clear communication comes down to how well you can navigate the difference between information and obligation. And because this sounds abstract, this article breaks it down in detail.

The reason this is so important is because these two things are often mixed together, and they often work perfectly well that way. You give someone some context, drop a hint about something; they pick up on it, ask you a question, you answer it, they react to your answer, etc. There is a dance to it all that is hard to break down into discrete moves.

With that said, sometimes things go wrong. Communication almost never needs to be perfect, but sometimes it fails to be good enough. And when those failures happen, then what? Can we recognize what went wrong? Do we learn from those experiences, or maybe more importantly, do we learn the right things from those experiences?

Again the functional distinction I’m talking about here is the difference between just general-information and obligation-information.

Obligation-Information

If the information was meant to convey a request (e.g., I mention that the trash goes out Tuesday night, that it’s my week, and that I happen to be out of town), then the implied obligation would be to comply or at least answer it.

Or if the information was intended to set an expectation (e.g., I mention that we start these meetings at nine), then just like the request, the desire would be for compliance or some form of pushback at that moment.

The challenge is being able to sense the difference between obligation-informatrion and just regular ole information.

So, how do you tell? Well, the best question I know of is this: “If they hear this, think it over, and then decide to do nothing differently, is that totally fine with me?”

If the answer is yes, then you’re talking about “information” or “general information” or “reference material.” If the answer is no, then there is some kind of tension behind what you said. You want the recipient to actually DO something with the information. The trick here is that most of us don’t easily recognize this. So, to help clarify this, let’s go through some examples of these differences.

Here are some examples in which I don’t care what the person decides after hearing the information:

1) Sounds Like Information and There Is No Expectation

• I tell someone that if they want good pizza, Tony’s Pizza Shop is a good place to order from. [I don’t care if they get pizza somewhere else because maybe Tony’s doesn’t deliver to their area or something.]

• I mention to a coworker that the client we’re both working with tends to respond a lot faster to email than to Slack. [If they keep Slacking him, that’s their business, and they may have a read on that relationship that I don’t.]

• Someone tells me they’re driving up to Philly around four, and I mention that 95 is usually a mess right about then. [They can leave at four anyway; maybe they’d rather sit in traffic than move the whole afternoon around, which is a perfectly reasonable trade.]

You’ll notice that these examples are also contingent upon the idea that I’m trusting the other person to use their best judgment. If you don’t trust that another adult can do that—as an overbearing parent might not—then, sure, you may not be able to easily identify with this category of communication.

But the confusions that happen with communication are rarely because someone is acting infantile or overbearing; it’s because what we say is always only a version of what we mean. The map is not the territory. And even similar communications can lead to misunderstandings and failed expectations. Now, let’s look at those examples, but this time with some implicit expectation behind them.

2) Sounds Like Information, But There Is Actually an Expectation

• I’m paying for pizza for the team offsite and I tell whoever’s putting the order in that I like Tony’s Pizza Shop. [If the order shows up from somewhere else (maybe several people pitched a different place), I’m going to want to know why, because I’ve got reasons for Tony’s—the gluten-free crust and the fact that they’ll actually deliver on time—that I may not have bothered to explain.]

• I tell my coworker that this client responds a lot faster to email than to Slack, and that I think it makes sense for us both to use email with him from here on. [If they keep Slacking him, I’m going to raise it, because the whole point is that he gets one consistent channel from our side and a mixed approach undoes what I’m asking for.]

• My colleague and I are driving up to Philly together for a two o’clock meeting, and I tell her 95 is usually a mess in the afternoon, so we should probably leave by eleven. [If she shows up at noon, we have a problem, and it isn’t a problem about traffic—it’s that I was making a commitment about our departure and she heard a fact about the highway.]

In these examples, I have some kind of implicit expectation. It likely I’m not even fully aware of it myself. And I likely wouldn’t ever even become aware of it unless the other person fails to meet my expectation.

But let’s imagine that I am aware of the expectation I’d like to set. How might I communicate that more clearly? Well, let’s look at this third variation.

3) Sounds Like an Expectation and There Is an Expectation

• I’m ordering pizza for the team offsite and I tell whoever’s putting the order in, “Use Tony’s Pizza Shop since they’re the only place nearby that does a decent gluten-free crust and Marcus can’t eat the rest of it. If they can’t take the order for some reason, come find me before you go somewhere else.” [Notice here the impact of providing my reasoning. The trick to learning to do this well is to just tack-on a “because…” after conveying something. At which point your mind will search for a way to complete that sentence.]

• I tell my coworker, “This client answers email way faster than Slack, and I’d like us both to move him over to email entirely. I know that’s a little annoying since we Slack about everything else, but I think the mixed channels are part of why he’s slow to respond, and I’d rather we test it for a month than keep guessing. Can you go along with that?”

• My colleague and I are driving up to Philly for a two o’clock meeting, and I tell her, “I want to leave by eleven. That’s earlier than we need if the roads are clear, but 95 in the afternoon is usually a mess, and I’d rather sit in a parking lot near their office than walk in late. Does eleven work for you, or do you need me to make it later?”

Again, the point here isn’t that the explicit versions are better than the less explicit ones. Life is messy. And while explicitness has some advantages, it also has plenty of disadvantages too. Sometimes being explicit makes things more confusing. Sometimes you just don’t have the time, and if it’s a choice between communicating something vaguely and not communicating anything at all, then it’s totally worth the tradeoffs.

But even outside any of these more calculated explanations, communication just doesn’t work through calculation. We just do it. And most of the time it works just fine.

Now, with that said, two things can be true at the same time. Yes, it tends to work out just fine without calculation AND we are more likely to get our needs met when we make it less likely the other person would misunderstand what we meant. You can never say exactly what you mean, but some maps get closer to describing their territory than others.

But often enough the problem is the opposite one, where I’ve reached for a strong-sounding phrase out of enthusiasm or habit or just because that’s how the sentence came out, and the other person hears a requirement in it that I never put there. Anyone in a position of power is very familiar with this.

4) Sounds Like an Expectation, But There Isn’t an Expectation

• Someone mentions they’re getting pizza and I say, “Oh! You’ve gotta go to Tony’s Pizza Shop.” [I don’t actually mean they’ve got to. I mean it’s good and I like it. If they order from the place down the street, I won’t think about it again, but “you’ve gotta” has the shape of an instruction, so they may well hear one.]

• My coworker mentions the client’s been slow to get back to her, and I say, “Yeah, you should just email him—he never looks at Slack.” [I’ve said “should,” but I’m not making a request; I’m handing over something I noticed about how he works. She’s the one on that account, and if she has a reason to keep it in Slack, that reason beats mine.]

• Someone tells me they’re driving up to Philly around four, and I say, “You’ll want to leave earlier than that—95 is brutal in the afternoon.” [What I mean is that it will probably be slow. “You’ll want to” makes it sound like I’ve got a stake in when they leave, and I don’t—it’s their afternoon.]

In Conclusion

So that gives us four situations, and I think the interesting part is that the two failures fail in opposite directions and don’t feel remotely the same from the inside. When I sound like I’m sharing information and I’m actually expecting something, I’m the one who ends up disappointed, and the disappointment usually shows up with a little story attached about the other person not listening, or not caring, or dropping the ball.

That story feels true, because from where I’m sitting I did say the thing. But when I sound like I’m asking for something and I don’t actually want anything, I generally feel nothing at all, since the whole cost of that one lands on the other person—they rearrange something they didn’t need to rearrange, or they go along with an instruction I never meant to give and file away a small resentment about it that I am never going to hear about.

So the question doesn’t fix anything by itself. It just tells you which business you’re in: Am I handing something over, or am I asking for something?

And that turns out to be worth more than it sounds like, because those two things have pretty different requirements.

That is a good part of why explicit structure exists in the first place—not because being explicit is better, or more honest, or more respectful, since plenty of the time it’s none of those. It exists because a group needs some way to make their expectations legible and contestable.

We do not need to spell out every step of the dance. But when it matters whether something is merely information or actually an obligation, we need some way to tell whether we are at least dancing to the same music.

Subscribe now

Note: This is a longer article compared to some more recent ones.

When people talk about improving collaboration, they almost always mean one thing: getting better results. Did the team ship the product? Did the meeting produce a decision? Did the conflict get resolved? If yes, the collaboration “worked.” If not, it didn’t.

This makes sense as far as it goes. But it doesn’t go very far. Imagine evaluating a car solely on whether it gets you from point A to point B. By that measure, a rusting 1987 Corolla with no brakes and a Porsche 911 are equivalent on any trip that ends without a crash.

To actually compare them, you’d need to talk about reliability, safety, comfort, fuel efficiency, handling, maintenance costs—a whole set of dimensions that are invisible when your only question is “did it get there?”

Collaboration is the same way. “Effectiveness”—did it produce the intended result?—is just one dimension of success among many. And when we evaluate different approaches to collaboration—structural versus cultural, for instance—collapsing everything into that single dimension obscures most of the interesting and important differences between them.

I want to argue that we need a richer vocabulary for what “successful” or “better” collaboration looks like. And it turns out that other fields have already done much of this work for us.

Other Fields Already Think This Way

The idea that “success” is not a single thing but a bundle of independent dimensions is well established in fields far from organizational theory. The most influential articulation came from David Garvin, a Harvard Business School professor who published “Competing on the Eight Dimensions of Quality” in the Harvard Business Review in 1987.

Garvin’s argument was that companies were failing at quality because they treated it as a monolithic concept—something a product either had or didn’t. In reality, he showed, quality comprises at least eight distinct dimensions: performance, features, reliability, conformance, durability, serviceability, aesthetics, and perceived quality.

What made Garvin’s framework powerful was not just the taxonomy but the strategic implication. A product could be excellent on some dimensions and mediocre on others, and that was fine—as long as you knew which dimensions mattered to your customers and your competitive position. You didn’t have to win on all eight. You had to know which game you were playing.

A similar move shows up in the ISO/IEC 25010 standard for software quality, which is less well known outside engineering but arguably more sophisticated. The standard splits quality into two models. “Product quality” covers the inherent characteristics of the thing itself: functionality, reliability, usability, efficiency, maintainability, portability, and so on. But then there’s a separate model called “quality in use,” which evaluates how well the thing performs in the hands of actual people in actual contexts. Quality in use has five dimensions: effectiveness, efficiency, satisfaction, freedom from risk, and context coverage.

That last dimension—context coverage—is particularly interesting for our purposes. It asks: does the solution work only in the situation it was designed for, or does it generalize to novel situations? Its sub-dimension of “flexibility” specifically assesses whether a system functions well in contexts that were never originally specified. This is not the same as asking whether the system “works.” A system can work beautifully in the original context and fail catastrophically in a new one.

In manufacturing systems research, the same instinct appears in the distinction between metrics like resilience, flexibility, reconfigurability, and robustness. Researchers in this space have argued that every performance metric can be decomposed into primary dimensions including time, quality, cost, capacity, and system entropy—where entropy captures how much disorder the system introduces or tolerates. Two manufacturing lines might produce identical output, but if one of them does so only under tightly controlled conditions while the other handles variation gracefully, they differ enormously on dimensions that matter in practice.

The common thread across all of these frameworks is a refusal to let “does it work?” stand as the only question. They all insist on asking: works how? works for how long? works under what conditions? works at what cost? breaks in what way? And when you start asking those questions about collaboration, the landscape shifts dramatically.

Dimensions of Collaborative Quality

Read more

You may already know this, but I’ve been working on a book called How Values Work for a while now, and I’m happy to announce it will be available for purchase by late next month.

And I think it’s going to be good. Really good. There’s nothing like it currently available and I thought today I’d share some of the introduction. Let me know what you think.

Introduction

The secretary of state, the president, they all talk about “values.” A “value” is a subjective desire, not an objective truth. George Washington said: “The foundations of our national policy will be laid in the pure and immutable principles of private morality.” If you had said, “Oh, Mr. Washington, you mean in our ‘values?’” Washington would have replied, “What the hell are you talking about?” -Harry V. Jaffa, quoted in Peter Robinson, “Harry Jaffa’s Affair With The Lincoln-Douglas Debates,” Forbes, July 17, 2009.

In the late 1940s, the U.S. Air Force was struggling with an alarming problem: pilots were crashing planes and no one could figure out why. The aircraft themselves were sound, the accidents weren’t occurring during combat, and the pilots were highly trained. The suspicion eventually turned to the design of the cockpits. The engineers had done lots of research and measured the physical dimensions of thousands of pilots to determine the ideal layout for the average pilot.

But in 1950, a young researcher named Gilbert Daniels tested that assumption. He compared the body measurements of over 4,000 pilots against the cockpit dimensions. The result was surprising: not a single pilot actually fit the average. Designing for the average turned out to mean designing for nobody. The fix wasn’t complicated. Instead of trying to find the one ideal design, they introduced adjustable seats, pedals, and controls so each pilot could fit the cockpit to himself. Crash rates fell and the problem was solved.

Now, what strikes me about this story is that the engineers weren’t careless and they weren’t working from bad data—they took thousands of measurements—they were just answering a question that couldn’t be answered, and no amount of better measurement was going to change that. I fear the debate about shared values falls into the same trap.

On one side are those who insist that every organization must clearly define and live by its shared values. By “values,” they usually mean a small set of guiding principles that are meant to shape decisions and behaviors—statements like “we value integrity,” “we put the customer first,” or even just a word like “innovation.”

Salesforce is often held up as proof that explicit values can anchor a culture and sustain trust, since it publicly centers values like “Trust,” “Customer Success,” “Innovation,” and “Equality” as day-to-day decision guides, from product choices to hiring and policy.

On the other side, attempts to define shared values are seen as too vague or inauthentic, an effort to manufacture alignment through empty slogans. Take Enron. Its 2000 annual report listed four values: Communication, Respect, Integrity, Excellence. Employees signed statements confirming they had read and would comply. Within a year the company was gone, having been exposed for widespread corporate malfeasance. Clearly, Enron’s declared values weren’t doing the job they claimed to be.

Both sides of this values-debate are animated by real truths. But they also share a premise I think is worth pausing on—that a stated value is a promise about behavior, so the space between the wall poster and the Tuesday afternoon decision is straightforwardly a kind of dishonesty. One camp responds by trying to shrink that space through enforcement and repetition, the other by refusing to open it at all, and in my experience most of the practical advice on offer is some version of one or the other.

But what if the gap between saying and doing isn’t the problem? What if that gap actually exists for good reasons? And if we imagine that it does, then what does that mean for all of our efforts to make sure we are “living our values?” Groups often need to articulate ideals that they don’t always meet. Sometimes the gap motivates aspiration. Sometimes it shields flexibility. Sometimes it preserves cohesion in the face of difference. In other words: the gap can be functional.

The real issue is not the existence of a gap, but the failure to understand what kind of commitment a declared value actually represents. Is it a binding promise that carries consequences if broken? Or is it a non-binding orientation, something meant more to inspire rather than constrain?

This distinction between binding and non-binding expectations is what should be getting our collective attention. It is the cockpit adjustment that makes the whole system usable, and much like the adjustable seat, it does not require anyone to settle the argument about which values are the right ones. It only requires that each participant be able to tell what is being asked of them.

When values are treated as binding, they function like rules or deal-breakers: cross this line and you’re out. When they are non-binding, they act as aspirations or frames-of-reference: they orient but do not compel.

Both forms are legitimate, but confusing one for the other is where the trouble starts—not usually as a dramatic failure, but as accumulated friction, hurt feelings, wasted meetings, and decisions that get relitigated because nobody was sure what they were in the first place.

This book takes up the question of what function a declared value is serving. It doesn't take up the question of whether a group is living up to its values, which I do think matters and which is often the reason someone picks up a book like this one, but it approaches that question from underneath. My hope is that with a clearer sense of what a value is actually doing, you'll be in a better position to make those judgments for yourself.

The aim therefore is not to ignore or eliminate the saying-doing gap, but to manage it better. If we can learn to see values in this way, we know how to operationalize them. We stop chasing one-size-fits-all and start designing for variation. I don't think that fixes much on its own. But it does mean that when a group disagrees about its values, they can at least be disagreeing about the same thing.

Leave a comment

“It’s easier to hold your principles 100 percent of the time than it is to hold them 98 percent of the time.” -Clayton Christensen, How Will You Measure Your Life? (2012)

If you want to save yourself a lot of pain and wasted time when you’re working with other people, you need to understand the distinction I make between “information” and “obligation.”

A long time ago, I wrote a piece about why you can’t expect favors, which took an insight I got from my experience with Holacracy and expanded it. It’s probably the single most important thing to understand about all of my work.

It also reveals a clear divergence from Holacracy, but I’ll get to that later.

It is the distinction between descriptive expectations (i.e. information; or what one thinks is likely to happen based on past history) and prescriptive expectations (i.e. obligations; or what one thinks should happen based on a commitment or agreement).

For example, imagine a contractor is working on your house and you need to get some information from them. You call and the contractor says he’ll need to get it from his office and will call you tomorrow with the information.

The operative question is: what kind of expectation is “I’ll call you tomorrow?” Is it just a projection or a promise? The distinction matters because we can’t know what kind of response we should have until we know what kind of communication it is.

If it’s just a projection, that means:

Read more

I talk about different types of structures like “values,” “guidelines,” “rules,” and “deal-breakers,” and I’ve been assuming that these different functional definitions can make sense to people.

But I don’t think I did a good enough job clarifying something upstream. It’s more about how I’m approaching the very idea of definition itself, which I’ve been using intuitively, but never really realized.

And it turns out I’ve accidentally stumbled on a new way to define words.

And since words have been around a long time, and there has been no shortage of people thinking about language and how it works, any skepticism you’re feeling at this point is warranted.

Note: There are some close relatives to this idea (see the very bottom of this post), and several one-off examples in different fields, but none define it the way I do here.

How We Define Words

So, what do I mean “a new way to define words”? Well, let’s start with an example. How do you use your name? That is, your first and last name are abstractions that represent your “you-ness”—but how do you use the actual words? Well, there are a few things you might think of:

• You use it to sign legal documents.

• To hold a spot on a list.

• To label things you own.

• To make it easy for other people to refer to you when you’re not there.

And many more. But notice, this is a list of what your name does, not what it means.

Your name does have a meaning—an etymology, a family it came from, something your parents were going for. That’s a different list, and for most of the work your name actually does, it’s irrelevant.

So there are two lists available for any word: what it means, and what it does. Let’s call the first a dictionary definition and the second a functional definition.

This is already a break from the two people most associated with the question. Wittgenstein said the meaning of a word is its use in the language, and Peirce said our conception of a thing's effects is the whole of our conception of it—both of them, in other words, answered the question by collapsing one list into the other.

In their defense, most of the time the distinction doesn’t really matter.

A “bucket” can carry water, be a step stool, or be banged on like a drum. That’s a functional list. The dictionary says a bucket is a round, open-topped container with a handle used to carry liquids or solids. Everything on the functional list is based on that one dictionary definition (although of course there are others).

Some Words Represent Reality, Some Shape It

These two types of definition can be applied to any word. But again, for representational words—words that point at things—the uses don’t vary much, because the thing itself does the constraining.

But constitutive words—words that create binding obligations—don’t work that way.

For example, I could talk to you about “the different functional ways to use scissors” because you can hold onto a rough concept of what “scissors” are, completely separately from any specific definition or use of them.

But this isn’t true for words like “promise,” “rules” or “values.” The ways we define them and the way we use them are hard to pull apart since there isn’t an identifiable neutral object we can envision that exists independently.

And this is where the two types of definition become important.

Just like with your name, some words have shared meanings, but what they represent isn’t defined in a dictionary. However, they also aren’t purely subjective. Your name means YOU because it has a locatable and contestable shared meaning within the social systems it’s used in.

Similarly, when you run both lists on a word like “values,” they come apart. We can look up what “values” means in the dictionary, but none of those definitions will tell you how that group actually uses the term.

Or if it does, it only does so generically (“values are core beliefs and guiding principles that shape how people decide what is right, wrong, important, or desirable in life”), which still doesn’t tell you how they actually operationalize them.

Now, for most linguists who may have recognized some version of this distinction, at most they stop here. And it’s understandable. Once you get to the observation, “This particular group uses this particular term this way,” it seems so specific you can’t really do much with it. It’s like if an isolated group decided to start calling a bucket a “fish-holder” instead of “bucket”—so what?! How does that help explain anything happening with other groups? But when it comes to collaboration, it can.

And that’s because we are not trying to define the words so abstractly that they must be recognizable and useful for all possible speakers, but we are looking at collaborative systems as the unit of interest. And it turns out that there are recognizable patterns that exist from group-to-group.

Collaboration Was the Key

They may not use exactly the same terms (e.g. one group might call it a “mission statement” and another group may call it their “charter”), but when we look at how they are using that document, we can see that they are functionally equivalent (e.g. “…this includes our history and identity, and we evaluate individual performance each quarter through referencing the core values contained within it”).

The local uses recur. Groups that never spoke to each other end up using their values-language to do the same handful of jobs.

So it’s not a dictionary definition, because it doesn’t hold for the word in general. And it isn’t subjective, because it keeps showing up across groups with no contact between them. It’s a third thing: a localized lexicon that generalizes anyway.

That class of language is why I call Making Rules Matter (MRM) a grammar for collaboration, because it does not specify what a group should say or even what they can say, but it does specify patterns in the underlying structural mechanisms groups use to coordinate action.

This also means I’m not doing what a linguist or analyst does. An analyst is after the abstraction that holds everywhere, stripped of any particular context. That’s a legitimate project, but it’s not mine. In a collaborative system the local is the point. What the word does for these people, in this arrangement, is the thing worth knowing.

In Conclusion

This is why my rule/structure typologies look the way they do.

When I distinguish values from guidelines from rules from deal-breakers, I’m not synthesizing dictionary definitions—hunting for the common core across everyone’s usage. And I’m not stipulating, announcing what I think the words ought to mean and asking you to adopt it.

I’m doing something similar to what Carnap called “explication”: taking a concept people already handle competently and making its working structure explicit.

Most people could probably tell the difference between a loose guideline and an absolute deal-breaker in practice, but what if they both appear on the same list of “rules”?

What is the relationship between a “No Cellphones” sign and a promise you make to a friend?

What kind of “rule” should you establish with your roommates about washing the dishes?

There are lots of ways to approach answering those questions, but I’m starting with the definitions themselves. And not the ones you’ll find in the dictionary.

Appendix: Are Functional Definitions Really New!?

Plenty of people have noticed that words do things. What I haven’t found is anyone treating “what it does” and “what it means” as two separate lists you can run on any word, then reading the gap between them. Here is the nearest company, sorted by where each one stops.

Peirce comes closest

Charles Sanders Peirce divided the study of signs into three branches: one for what a sign is, one for whether reasoning is valid, and one for how signs affect whoever takes them up. That third branch is the room this idea would live in. He called it the highest and most living branch of logic—and never wrote it. By 1903 he had narrowed it into a theory of scientific method and left the original slot empty.

He also came within an inch of the point about names. In 1905 he admitted that his own theory of meaning fails to furnish any translation or meaning of a proper name. A name has no meaning in his sense, while obviously still doing plenty. He states the gap and does not walk through it.

Peirce, “What Pragmatism Is,” The Monist 15 (1905). The interpretant material is in Collected Papers 5.475–5.476, drawn from an unpublished 1907 manuscript (R318). For a close reading of that manuscript—and a warning that the published selections truncate it badly—see T.A. Jappy, “Peirce’s 1907 logical interpretants,” Language and Semiotic Studies, doi.org/10.1515/lass-2023-0043. The manuscript itself is digitized at Harvard: iiif.lib.harvard.edu/manifests/view/drs:48987092$1i

Ran the “what it does” list and stopped

• Matthieu Queloz asks what drove us to develop a concept, in order to discover what it does for us. A general method, applicable to any concept—but with no second list to set it against. Queloz, The Practical Origins of Ideas: Genealogy as Conceptual Reverse-Engineering (Oxford University Press, 2021).

• Geoffrey Bowker and Susan Leigh Star ask what work classifications and standards do. Same shape, applied to categories rather than words. Bowker and Star, Sorting Things Out: Classification and Its Consequences (MIT Press, 1999).

• Alexis Burgess and David Plunkett’s “conceptual ethics” asks which concepts we ought to be using. Forward-looking and normative, where this is diagnostic. Burgess and Plunkett, “Conceptual Ethics I” and “Conceptual Ethics II,” Philosophy Compass (2013).

Ran two lists that answer the same question

• Sally Haslanger distinguishes the concept you think you are applying from the one you are actually applying. Her example: the school rule says tardy means arriving after the 8:25 bell, but nobody is ever tardy on Wednesdays because the teacher turns in attendance late. Both of those are criteria for sorting students. A functional definition isn’t a criterion at all. Saying that it “creates a record that follows the kid” doesn’t sort the students. Haslanger, “What Are We Talking About? The Semantics and Politics of Social Kinds” (2005). Free PDF: sallyhaslanger.weebly.com/uploads/1/8/2/7/18272031/haslanger_2005_whatarewetalkingabout.pdf

• Elinor Ostrom, with Sue Crawford: rules-in-form versus rules-in-use is the same axis—what is written down against what is practiced. Crawford and Ostrom, “A Grammar of Institutions,” American Political Science Review 89:3 (1995), 582–600. Free PDF: wiki.santafe.edu/images/a/ab/Crawfordandostrom1995.pdf

• John Meyer and Brian Rowan’s decoupling, in organizational theory, is that axis again—formal structure as ceremony, with the actual work happening somewhere else. Meyer and Rowan, “Institutionalized Organizations: Formal Structure as Myth and Ceremony,” American Journal of Sociology (1977).

Ran two genuinely different lists, but not on standing words

• J.L. Austin’s performatives are the origin point: “I promise,” “I do,” “I name this ship.” Utterances that accomplish rather than describe. The unit is the utterance, not the word. His essay “The Meaning of a Word” goes further still, arguing the phrase itself is nonsense—there is no handy appendage of a word called its meaning. Austin, How to Do Things with Words (Oxford, 1962); “The Meaning of a Word,” in Philosophical Papers.

• Quentin Skinner built a whole method of intellectual history on recovering not just what an author meant but what he was doing in writing it. Two different questions, systematically applied—to texts. Skinner, “Meaning and Understanding in the History of Ideas,” History and Theory 8:1 (1969), 3–53. jstor.org/stable/2504188

• The ISO standard for dialogue annotation splits every conversational move into a “communicative function” and a “semantic content.” This is the distinction built to industrial spec, with annotated corpora and inter-annotator agreement studies. Per utterance, and with the two halves composed rather than compared. ISO 24617-2, “Semantic annotation framework, Part 2: Dialogue acts,” building on the DIT++ taxonomy.

Got it right, on one class of word

• Charles Stevenson’s work is the closest single piece of prior application. He argued that ethical terms carry emotive force largely independent of descriptive content, and named the move where you swap the definition while keeping the force: a persuasive definition. Restricted to ethical terms—and he calls both halves “meaning,” which is why the split gets reabsorbed into semantics. Stevenson, “Persuasive Definitions,” Mind 47:187 (1938), 331–350. doi.org/10.1093/mind/XLVII.187.331. See also “The Emotive Meaning of Ethical Terms,” Mind 46 (1937).

• H.L.A. Hart argued that legal sentences about rights don’t describe, they ascribe, and wrote a section on why that function gets overlooked. He later said he wished he’d had the distinction between meaning and force available at the time. Hart, “The Ascription of Responsibility and Rights,” Proceedings of the Aristotelian Society 49 (1949), 171–194. doi.org/10.1093/aristotelian/49.1.171

• Michael Calvin McGee’s “ideographs” are political abstractions—liberty, freedom, equality—that have no clear definition and do enormous work anyway. McGee, “The ‘Ideograph’: A Link Between Rhetoric and Ideology,” Quarterly Journal of Speech 66:1 (1980), 1–16. doi.org/10.1080/00335638009383499

• W.B. Gallie’s “essentially contested concepts” are ones where the argument is permanent and the permanence isn’t a defect. His opening observation is that ordinary disputes about a word assume agreement about the kind of use appropriate to it—and that some words are the exception. Gallie, “Essentially Contested Concepts,” Proceedings of the Aristotelian Society 56 (1956), 167–198. Free PDF: cooperism.law.columbia.edu/files/2023/12/Gallie-Essentially-Contested-Concepts-1955-CL.pdf

• John Stuart Mill, and later Saul Kripke, argued that proper names denote without connoting—they pick out without describing. Mill, A System of Logic (1843), Book I; Kripke, Naming and Necessity (Harvard, 1980).

Adjacent, doing something else

Eric Eisenberg’s “strategic ambiguity” is the organizational version: abstractions people can agree on precisely because they don’t pin down interpretation. Eisenberg, “Ambiguity as Strategy in Organizational Communication,” Communication Monographs 51 (1984), 227–242. doi.org/10.1080/03637758409390197

Susan Leigh Star and James Griesemer’s “boundary objects” supply the mechanism—things that carry different meanings in different groups while still coordinating them. Star and Griesemer, “Institutional Ecology, ‘Translations’ and Boundary Objects,” Social Studies of Science 19:3 (1989), 387–420. Free PDF: lchc.ucsd.edu/mca/Mail/xmcamail.2012_09.dir/pdfuaCxVBhVe5.pdf

Adam Kilgarriff argued that word senses exist only relative to a task. That is this territory approached from the opposite side: rather than pairing the two lists, he dissolves the meaning list into purpose. Kilgarriff, “I Don’t Believe in Word Senses,” Computers and the Humanities 31:2 (1997), 91–113. Free PDF: kilgarriff.co.uk/Publications/1997-K-CHum-believe.pdf

Uwe Pörksen’s “plastic words” are terms that migrated out of science into everyday speech with their content stripped off. Pörksen, Plastic Words: The Tyranny of a Modular Language, trans. Mason and Cayley (Penn State Press, 1995).

Bronisław Malinowski wrote, in 1923, that a phrase of politeness fulfills a function to which the meaning of its words is almost completely irrelevant. One sentence, one class of case, never developed. Malinowski, “The Problem of Meaning in Primitive Languages,” supplement to C.K. Ogden and I.A. Richards, The Meaning of Meaning (1923).

Historical linguistics has “semantic bleaching”—words losing their meaning while their function survives and even strengthens. Divergence as an ordinary documented process, though the function in question is grammatical rather than social. Hopper and Traugott, Grammaticalization (Cambridge).

In this video I share my framework defining the boundaries between information, relationship, and collaboration. I’ve written a reference article that goes into more detail, but the key for today is the definition of relationship.

Let me know what you think and thank you again for all of your support!!!

Holacracy’s approach to self-organization rests upon one single construct: a tension. It’s defined rather blandly as a gap you sense between current reality and a potential future,1 but it’s a definition that does a lot of heavy lifting.

One of the most important things it does—almost impossible to see unless you really get into it—is offer a way to connect felt bodily experience to cognitive reflection, without resorting purely to emotional states or personal history.

Much like how I emphasize “structure,” not “culture” or “mood.”

The key is the word “sense.” We may say a tension is “felt,” and that’s true, but what we call feelings can be broken up into at least two categories: sensations and emotions.

For example, can you feel your feet right now? Do you feel itchy, cold, or hot somewhere in your body? What about the chair you’re sitting on—can you feel its pressure against the weight of your body?

None of these are emotions. They are sensations. And just because we can make this distinction doesn’t mean the two can’t be linked. In fact, the space between a raw sensation and a nameable emotion is one of the most worked-over territories in the history of thought, and the word “sense” sits right in the middle of it.

Psychotherapy researcher Eugene Gendlin called it exactly that: the “felt sense.” Before you can say you’re anxious, or that a plan is somehow “off,” there is a bodily, murky, not-yet-nameable knowing of the whole situation—specific but implicit, and waiting to be articulated.

Gendlin’s entire method (he called it Focusing) is the practice of carrying that felt sense forward until it yields meaning. And the distinction from emotion is the whole point: an emotion is already categorized and recognizable, while a felt sense is the thing that comes before the category (Gendlin, 1978).

Antonio Damasio draws the same line from the other direction and makes it physiological. For Damasio, an emotion is a set of bodily reactions that mostly unfold outside our awareness, while a feeling is the mind perceiving that bodily state—the read-out of what the body is doing (Damasio, 1999).

Which gives a principled reason why “felt” and “emotional” are not synonyms: the feeling is the reading; the emotion is the event being read.

The phenomenologists have their own versions—Heidegger’s attunement, how you already find yourself situated in a world, which is not a mood; Merleau-Ponty’s “solicitation,” the felt pull of a situation calling for a response.

All of these different vocabularies have one throughline: there is a register that is genuinely felt and genuinely informative without being emotional.

Why is this important?

Because it explains the philosophical stance Holacracy takes toward humans and their systems—and once you see that stance, a lot of other things become clearer.

Too often a structure-oriented practice like Holacracy gets read in one of two opposite ways. The first is that it’s emotionally avoidant: all rules and structure, indifferent to how people actually feel.

The second reading, ironically, is the reverse—that as a non-hierarchical system, it must be a place where everyone’s feelings finally get a seat at the table. You can find evidence for both, but both are wrong.

A tension is not necessarily an emotion to be processed. It’s a felt sense—a signal that something is off.

Holacracy asks you to take that signal seriously, which is exactly why it is not cold or anti-feeling: the whole system starts in the body, with something sensed.

But it does not ask you to be governed by whatever story that sensation arrives wrapped in, which is why it is not a feelings-democracy: sensing a tension is not the same as being led by your emotions.

You register the gap, and rather than suppressing the feeling—or overriding it—you process it into something. A request. A change. Maybe just a new insight about yourself.

This is the difference between being structure-oriented and being mood-oriented. A mood-orientation lets an individual’s emotions or the collective mood override the group’s shared purpose and agreements. A structure-orientation begins with the very same felt data—it just refuses to stop there.

What’s the difference between a vow of silence and a gag order? Or fasting and starving? It’s all down to choice. The behavior is the same, but the relationship to the behavior is completely different.

And I think a similar phenomenon impacts almost everything we do together, and it’s largely self-imposed. It’s our assumption that “we must work together.” That is, how we relate to a relationship sits underneath the relationship itself.

Is it freely chosen? If not, must you stay in it? Or does it feel forced upon you? You agreed to join a fun progressive company of like-minded people, but now you’re forced to deal with Todd in accounting.

And this provides a great example of how this Making Rules Matter approach (MRM) aims to explain the deep mechanics shared across different relational or organizational practices. This one is called, “exit.”

Should I Stay or Should I Go?

Years ago at HolacracyOne, I reached a boiling point frustration about the direction of the company and I was determined to figure out what to do. As I reflected, I realized that the problem wasn’t just HolacracyOne.

It couldn’t be in fact. Because on its own “this place sucks” isn’t a problem. I was suffering because I was adding to that belief, “and I must be here.” It’s the old Buddhist thing, “pain + resistance = suffering.”

And once I realized that, I felt done relief. Maybe this isn’t the place for me anymore? Maybe the company is doing exactly what it should be, and this feeling is maybe a sign that it’s time to move on?

Read more

There are lots of ways to define your group’s “success,” but here is a simple one: do we do what we say we do?

It means saying what you mean and meaning what you say. It’s never perfect. But it also never needs to be.

The point isn’t perfection, but predictability. Do your words generally and consistently match your actions?

For the roles and rules you have defined together—do people follow them?

Are you disciplined enough so you can afford to be undisciplined?

Do you consistently distinguish your intentions, aspirations, and desires…from your obligations, your promises, and your commitments?

When there is congruity between a person’s words and their actions, we say they have “integrity,” or are being “authentic.”

The path towards being a more authentic organization is no easier for a group than an individual, but it’s a path worth pursuing.

And this definition of success is as straight forward as this line from Ben Franklin: “What you seem to be, be really.”

I’m finishing up the latest edits on an academic paper, and last week I came across a critical insight about why I think making rules matter is not only important, but why it seems important now. And it’s something worth knowing.

In the 1980s, “user-centered design” became a thing. This is when computer engineers started to actually care about the user experience. Well, that is an overstatement—of course they cared about the experience, but it wasn’t what they prioritized. Instead, they tended to prioritize the power and efficiency of the machines themselves.

At the University of California, San Diego, and driven forward by advances in cognitive psychology, Donald Norman’s group published User Centered System Design: New Perspectives on Human-Computer Interaction (1986).

Somewhat separately, participatory design grew out of Scandinavia in the early 80s from labor and workforce-democracy concerns rather than psychology, an approach distilled in Computers and Democracy: A Scandinavian Challenge (1987). This group shared the same observation and solution: they should build computer tools with the people in mind.

What Changed?

I had always thought of user-centered design as a perfect parallel for what I was trying to do with collaboration, governance, etc., with my best attempt to articulate that vision being what I call the principle of participant navigability.

Meaning that my distinctions, models, and labels are all angled toward the concerns of participants trying to work together, rather than for objective outside analysis, as is often prioritized in the social sciences.

Read more

W.L. Gore & Associates has been running without a conventional management hierarchy since 1958—decades before anyone had a name for it.

Founded in a Delaware basement, the company built a “lattice” structure in which people commit to projects rather than receive assignments, and leaders are recognized because others choose to follow them, not because a title says so.

Since its founding, W.L. Gore has employed thousands of people. My mother-in-law was one of them. And she hated it.

Of course, it was a long time ago, but it’s clear from the literature that there are recognizable patterns for what happens when you give people too much freedom without enough scaffolding.

Here are some:

1. Responsibilization

You’re handed “ownership” and “autonomy,” but what actually changes hands is accountability. Responsibility for outcomes slides from the organization onto the individual, and the transfer gets sold to you as freedom (Shamir, 2008; Rose, 1999).

2. Concertive control

Take away the manager and control doesn’t leave the room—it moves to your peers, and it gets tighter. In a well-known study of self-managing teams, the norms coworkers enforced on each other ended up binding them more completely than the hierarchy they had replaced (Barker, 1993).

3. Laissez-faire leadership

Sometimes “self-management” is just leadership that checked out. Absence isn’t neutral; it reliably produces role confusion, conflict, and stress that everyone downstream absorbs (Skogstad et al., 2007).

4. The tyranny of structurelessness

Getting rid of explicit structure doesn’t get rid of power. It hides it. Informal in-groups end up running things off the books, and the people without the right relationships pay for it (Freeman, 1972).

Conclusion

That’s four names for one lesson: self-management does not automatically equal more freedom.

Which is why this was never really an argument about hierarchy versus self-management. Self-management needs good structure to work well. So does a hierarchy. What you’re choosing between is good structure and bad—not triangles versus circles.

What people actually experience as freedom at work is almost never the absence of rules or structure.

It’s a good structure doing its job—clear roles, real support, authority that’s accountable, and a way to change the rules when they stop working. Strip those out and “good work” just becomes “good luck.”

References

Barker, J. R. (1993). Tightening the iron cage: Concertive control in self-managing teams. Administrative Science Quarterly, 38(3), 408–437.

Berenstain, N. (2020). White feminist gaslighting. Hypatia, 35(4).

Freeman, J. (1972). The tyranny of structurelessness. Berkeley Journal of Sociology, 17, 151–164.

Rose, N. (1999). Powers of freedom: Reframing political thought. Cambridge University Press.

Shamir, R. (2008). The age of responsibilization: On market-embedded morality. Economy and Society, 37(1), 1–19.

Skogstad, A., Einarsen, S., Torsheim, T., Aasland, M. S., & Hetland, H. (2007). The destructiveness of laissez-faire leadership behavior. Journal of Occupational Health Psychology, 12(1), 80–92.

Imagine you were asked to decide which kidney patients received life-saving treatment and which wouldn’t. It’s exactly the kind of moral dilemma used to generate debate in a philosophy class. And while it’s useful for that, imagine if you actually had to make the decision of who lived and died.

Well, in 1961 that’s exactly what happened to seven Seattle citizens who were tasked with choosing who got one of three scarce dialysis machines with no guidelines beyond their own conscience (called “the God Committee”).

And it wasn’t a one-time fluke: the whole scenario (as well as the backlash from its ad-hoc formulation) helped push U.S. hospitals toward formal medical ethics programs, which are now required for accreditation and present in 97% of American hospitals today.

I mention this because medical cases demonstrate the critical difference between discussion and formal obligation. Said more simply, it’s the distinction between comments and commitments.

Read more

I spend a lot of my time helping people untangle disagreements about rules, and one thing I’ve noticed is that the disagreement is rarely where everyone thinks it is.

People assume they’re fighting about whether a rule is good or bad, fair or unfair. More often they’re standing in completely different places, looking at the same rule, and seeing different things entirely.

That’s not a small point. If you can’t see the different vantage points a rule can be viewed from, you’ll keep having arguments that can’t be won—because the people in them aren’t actually talking about the same thing.

And integrating these differences isn’t necessarily just reconciling differences in interpretation, but often it’s about different fundamental perspectives.

For example, a lawyer, a priest, a game designer, and a meditation teacher would give you four different answers to the question of what a rule is, and they would barely seem to be describing the same thing.

The lawyer points to a codified constraint with a defined scope and an enforcement mechanism.

The priest points to a sacred obligation, an inherited “should” that holds a community together across generations.

The game designer points to a protocol that shapes behavior through incentives and feedback, revised as the system learns.

The meditation teacher points to an inner command arising in awareness—a felt “I must”—something to be noticed rather than automatically obeyed. None of them is wrong.

This is the peculiar thing about rules. The word does not name a single object so much as a whole family of experiences, each one real from where the observer happens to be standing.

Recently I ran the concept through Integral Life’s GigaGlossary, an AI tool built on Ken Wilber’s AQAL framework that shows how the same phenomenon discloses differently depending on the perspective you take on it.

Asked to enact “rules,” it produced a set of views that lined up almost exactly with the four quadrants of experience: interior and exterior, individual and collective.

Seen from inside an individual, a rule is a felt injunction—the tightening of conscience, a private vow, an expectation woven into one’s sense of self.

Seen from outside that same individual, it is observable behavior—compliance, deviation, and the measurable effects of keeping or breaking it. Seen from inside a group, it is shared meaning—a sacred obligation, a negotiated agreement, the felt sense of what “we” are protecting when we honor it.

Seen from outside the group, it is a system—laws, policies, protocols, and incentives, the architecture that lets large numbers of people coordinate without ever meeting.

Take something ordinary: a team’s rule that meetings start on time. From inside the individual, it is the small private pressure to stop what you’re doing and show up.

From outside, it is a data point—who arrives when, how often, and what the lateness costs. From inside the group, it is a statement about respect, so that arriving late reads as a slight rather than a scheduling accident. From outside the group, it is a coordination device—calendars, norms, maybe a door that closes at 9:01. One rule, four disclosures, every one of them accurate.

Laying these side by side is genuinely useful, and not only for tidiness. It guards against the most common move in any argument about rules, which is to grab one facet and call it the whole.

Rules are “really” just instruments of power. Rules are “really” just social agreements. Rules are “really” just internalized conditioning.

Each of these takes something true and inflates it into the entire picture, and the arguments that follow are often not about rules at all, but about which quadrant the speakers are standing in. A map of the perspectives dissolves a surprising amount of that noise.

But a map has a limit, and it is the one I keep running into in my own work. Knowing where a rule sits among the perspectives does not tell you the thing you most need to know when you are inside a collaboration: how to work with it.

Two rules can occupy nearly the same location—both written down, both formally legitimate, both endorsed by the group—while one is relied upon without a second thought and the other is routinely ignored.

One earns principled objection; the other, resentful going-along. Perspective alone cannot account for the difference, because it describes where a rule sits, not how it is met by the people living under it.

That gap is the reason my work exists. A map tells you what is where; it does not hand you a compass, or a gauge, or any of the instruments you reach for when you are moving through terrain rather than surveying it from above.

The people doing the collaborating—the ones who have to follow a rule, enforce it, author it, and revise it, often all at once—are not standing outside the map like cartographers. They are inside it, acting under pressure, and they need tools built for navigating rules, not just for describing them.

So integrating the perspectives is the right place to begin, and the quadrants are a good beginning: they make the plurality visible and keep any single view from swallowing the rest. But integration is the setup, not the payoff.

The harder and more useful question is what comes after the map—how a person inside a living system reads a rule, senses whether it actually binds, and decides what to do when it doesn’t. That turns out to be a very different project from mapping, and it’s the one I’m really after.

There’s a lot of talk right now about which decisions we should hand over to AI. Should it approve the loan? Grade the essay?

The conversation usually gets stuck on the big dramatic questions—is this thing really intelligent? What will AI-driven job displacement do to the economy? To our social institutions? I find those genuinely interesting.

My limited take is that a lot of these AI debates get more navigable the moment we stop indexing them against big unknowns and start indexing them against trust and legitimacy.

By legitimacy I just mean whether we accept the authority of something as trustworthy and valid. Not necessarily whether we like it.

It’s an old and ordinary idea—we lean on it constantly without ever naming it—but I think it’s the thing doing most of the real work in the AI debates.

The Issue of Legitimacy

Take medicine. There are already well over a thousand AI tools cleared for medical use, and most doctors use some of them. In a lot of specific tasks the technology is good—sometimes better than the humans.

And yet the regulators’ whole posture, so far, has been that these systems support a decision a person still makes. They inform; they don’t decide. So why would that be, if the software is often more accurate? It’s not really about accuracy.

In one survey, nearly half of radiologists said they thought patients simply wouldn’t accept a report that came from an AI alone.

That’s not a question about whether the machine can read the scan. It’s a question about whether we’ll accept its verdict—which is to say, a question about legitimacy.

There’s even a bill floating around Congress about whether an AI could count as a “practitioner” allowed to prescribe medication.

The point is, handing a decision to a machine sometimes makes us trust the outcome more, and sometimes makes us trust it less—with the very same technology.

So the deciding factor was never the chip. It was whether the output had earned the right to be believed and acted on. That’s an old question. AI just makes questions about truth and trust less philosophical and more immediate.

Example: Major League Baseball

Major League Baseball is running this experiment this season with an automated system—the same tracking tech behind all its stats—to help settle balls and strikes during games.

Read more

Creativity requires commitment. But not in the way we normally think about it.

I mean we can’t really feel free until we know where the limits are. So, openness and exploration require a clear way to signal when we need to start closing things down.

This happens in all kinds of deliberative decision-making, but it shares some important overlaps with how we operationalize rules.

For example, we might say something like:

• “I’m not saying we’re committing to this, but I’d like to just do some thought-experiments…”

• “This isn’t a final decision, we are just brainstorming…”

Normally, we don’t sense the need for this kind of explicit boundary-setting. We certainly jump back and forth between them all the time in our heads, and even when we’re in groups, the divergent and convergent rhythms can organically flow.

But—and here is the important piece—what happens when things go wrong? If you don’t have a way to make the conversational dynamics legible, you make them hard to correct.

Moreover, it’s one thing to consciously decide no explicit structure is needed, and another to leave it unspecified because you don’t have language for it.

Where this often becomes confusing is when we can’t reliably distinguish the rule itself (the explicit binding commitment, agreement, etc.) from communications or interpretations about the rule.

They may be the same thing like when a store owner posts a sign saying “We are no longer accepting ANY expired coupons.” The sign represents the unilateral encoding and a notification to staff and customers.

Yet, they are clearly different like when an employee handbook describes the company’s non-discrimination policy. You could erase that section of the handbook but the rule would still exist

These two elements can become confused, though, as when the OMB rescinded its 2025 funding-freeze memo: the administration announced the rescission but insisted it ended only the memo, not the freeze itself, leaving agencies and courts to sort out whether the official rule had actually changed.

Interestingly, the court ruled, essentially, that the memo did constitute a binding obligation.

This is why encoding’s critical function is clarifying the boundary between non-binding information and binding obligation.

The clearer and well understood that process is, the easier it is to communicate non-binding information.

This is because without that clarity, communication must be hedged and nuanced. The message carries too much of the interpretive burden. This predictably contributes to confusion, but also means less information is shared.

Conclusion

The point of all of this is, first, to recognize that while information and obligation are usually mixed together, that is only a justification for having a good way to distinguish them.

Second, and this is the part that often gets missed, having a clear boundary between the rule itself and anything else isn’t prioritizing formalization to the exclusion of the relationships.

In fact, when operationalized appropriately, rules not only provide clearer expectations and firmer accountability, but as a consequence they also free us up to play more knowing the structuring-space is always available.

For a more detailed explanation of how propositions differ from pointers, see this reference article.

In this video I ramble about the paradigm behind Holacracy and how it relates to integrating AI into self-management. And how people often say they believe in Holacracy, but actually mean the broader paradigm, which Making Rules Matter and Structure-Oriented Collaboration aim to articulate.

I also go on a bit challenging the idea that AI will au…

Read more

You can make a rule explicit. You can even make explicit the rule you used to make that rule, and the rule behind that one. But no matter how far down you go, you hit the same wall: eventually the dynamics in play are not going to be explicit. There is always context.

That context is what I call the wild—the unspecified, liminal, creative space that always surrounds us. No matter how clear or agreed-upon the structures, rules, or expectations are, everything exists within a chaotic and creative wilderness of unconscious and subtle dynamics, the stuff that makes “we,” to borrow from Ken Wilber, a kind of miracle.

In the MRM framework, norms and beliefs are already kept distinct from rules. The wild goes a step further out. It includes norms and beliefs, but also everything else that shapes how we think, what we experience, and how we act—norms and beliefs sit like a layer between the wild and the rules themselves.

Here is why it matters: just because you make a rule doesn’t mean anyone will follow it. Whatever your formal position in the hierarchy or the relationship, everyone always has wild power—the potential to reset and redetermine the rules of engagement. You never fully escape the wild, and you shouldn’t want to.

So rules are not the opposite of the wild. Much like frontier towns in the Old American West, created amidst a chaotic social landscape, they are a complement to human nature, not an attempt to eliminate it.

Civilizing anything is the same balancing act—checking the worst of our wild nature without smothering the best of it.

The “wild” is the word I keep coming back to. It conveys multiple things at once: natural beauty and chaotic uncertainty. Freedom and fear.

The point is, better rules come from working with the wild, not against it.

When you’re driving somewhere new, you don’t know all the turns ahead. And yet you trust you’ll get there. Why?

Well, there are two reasons. First, you have the ability to make adjustments as you go. Second, you have navigational aids that tell you where you are at any given moment.

You don’t need to predict the whole trip. You just need to keep sensing where you are.

That’s a useful frame for what a lot of people are going through at work right now.

With AI reshaping things at a pace nobody quite expected, more and more people are bumping up against the limits of their mind’s ability to predict and control reality.

The old habits of plan and predict aren’t paying off the way it used to. For most, that’s not a pleasant realization.

The old ways don’t work as well, but there aren’t clear alternatives. Well, I certainly don’t have all the answers, but I do know what the alternatives to the “predict and control” paradigm look like.

And the essence of that difference isn’t necessarily mystical—it’s just the capacity to get really good at sensing the present. But how?

How to Operationalize Presence

Read more