The Dark Matter of the Org Chart

09 Oct, 2026

Why your company works better than its structure explains, and why every reorg breaks the part nobody can see

The rotation problem

In the 1970s, Vera Rubin pointed a spectrograph at spiral galaxies and discovered something rude. The stars out at the rim were moving far too fast.

If you weigh a galaxy by counting the light coming out of it, the outer stars should behave like children on a carousel that is spinning too quickly - flung outward, gone. Instead they orbit serenely, held in place by something with no apparent interest in being visible.

There were only two ways out. Either gravity does not work the way we thought, or there is a great deal of mass we cannot see. Astronomers, conservative in the good sense, chose the second. It turns out that roughly five sixths of the mass of the universe is furniture we keep barking our shins on in the dark.

Now open your company's org chart.

Forty boxes. Eleven dotted lines drawn by someone who had clearly stopped believing in the exercise. Two roles marked "TBD" and one marked "TBC," which is a different kind of TBD. A matrix so ambitious it requires a legend. By the visible mass alone, this thing should not be able to ship a birthday card, let alone a product.

And yet the release goes out. The invoice gets paid. The furious customer is somehow retained. The velocity is real and the chart does not account for it.

You are looking at a rotation curve. Somewhere in your company there is unseen mass.

An inventory of the invisible

The dark matter of an organisation is not mysterious once you stop insisting it appear in a system of record.

• The person everyone asks. Her title says Senior Analyst. Her actual function is being the index of the company. Eleven years of institutional memory, no formal authority, and a Slack response time that shames your incident SLA.
• The channel where decisions actually happen. Not the steering committee. The steering committee ratifies. The channel decides, at high speed, in lower case, with reaction emoji standing in for governance.
• The translator. One human being fluent in both dialects of the same requirement, capable of explaining to Engineering what Sales meant and to Sales what Engineering heard, thereby preventing roughly one lawsuit per year.
• The archivist. Knows why the strange flag exists. Because he knows, nobody removes it, and because nobody removes it, billing still works.
• The unblocker in finance. Knows which field on the purchase order form triggers a two week delay. Will fix it for you in forty seconds if you ask nicely, and this is, in accounting terms, an asset with no line item.
• The trust ledger. Thousands of tiny favours, accumulated across years, spendable instantly and without paperwork. Nowhere on a balance sheet. Nowhere in a competency framework.

None of it is in the HR system. None of it has a cost centre. None of it is in a job description, and thank goodness, because the moment it were written down it would be measured, and once measured it would be targeted, and once targeted it would die of exposure.

Why the chart cannot see any of this

Institutions can only manage what they can see, so they quietly redefine reality until it becomes visible. This is the argument James C. Scott made about forests and cities and cadastral maps, and it applies with unpleasant precision to career ladders. The org chart is a cadastral map of authority. It records who is permitted to say no. It is silent on the far more important question of who says "actually, try it like this" at four on a Thursday and thereby saves the quarter.

Conway's law is the best evidence that somebody noticed this early and everyone else misfiled it. Conway said organisations produce designs that mirror their communication structures. Note the word. Not reporting structures. Not boxes. He was describing dark matter, and we have spent fifty years quoting him about hierarchy.

Three laws of organisational dark matter

1. It is load-bearing in inverse proportion to its documentation. The more essential the coordination, the less likely anyone has written it down, because people who are busy holding up the ceiling rarely stop to file a diagram of their arms.
2. It cannot be transferred by announcement. Authority moves by memo. Trust moves at the speed of shared failure.
3. It is invisible to the instrument that redraws it. The slide has room for boxes. It has no notation for a relationship.

The reorg as a gravitational event

The deck arrives. New structure. Clearer accountability. Closer to the customer. Someone has used the word "crisp" as though it were a design principle.

Here is what actually happens in the physics. A reorg is superb at moving boxes and completely incapable of moving edges. Titles migrate. Trust does not. Every new boundary you draw shears the informal connections that happened to run across it, and since those connections were never on the map, nobody is required to notice.

So the channel is archived "to reduce noise." The two teams who had, over eighteen patient months, developed a functioning shorthand are now in different pillars with different priorities and a shared calendar conflict. And the person everyone asks now reports to a leader who does not know what she is for, so she is loaded up with formal work matching her formal title, and all the invisible coordination she used to perform simply stops - not by anyone's decision, which is precisely why it is undetectable.

Eighteen months to accumulate. One calendar invite titled "Org update" to unweigh.

The attribution comedy

Then the velocity drops, and here comes the finest scene in the whole production.

At no point does anyone stand up and say "we appear to have deleted several thousand hours of undocumented coordination capacity, most of it in the form of relationships." Instead the diagnosis arrives, fully formed and beautifully blameless.

Change fatigue.

The prescription follows. A communications cadence. A resilience workshop, ideally with a facilitator who says "hold space." A listening tour, in which leadership listens. A pulse survey whose results are characterised as encouraging, given the context.

Change fatigue is a real phenomenon, and it is also the most convenient explanation available, because it locates the problem in the emotional stamina of employees rather than in the judgement of the people who drew the boxes. It is the management equivalent of proposing that gravity itself behaves differently in your galaxy. Rather than accept the existence of mass you failed to measure, you revise the laws of human nature. People are simply resistant to change. Perhaps. Or perhaps you shot the gravity and are now filing an incident report about the falling.

An honest caveat, because dark matter is not automatically benign

Not every invisible structure is load-bearing trust. Some of it is rent.

There is the gatekeeper who could document the process and declines to, because opacity is his pension. There is the single point of failure who is single by choice, because being indispensable feels remarkably like being valued. And informal networks are notoriously poor at inclusion, since they route along familiarity - which quietly favours whoever shares your accent, your alma mater, your commute, your smoking break. The invisible organisation is often warmer than the formal one and almost always narrower. Formal structure exists partly as a corrective. It is the reason someone with no golf handicap can still get promoted.

So the point is not to romanticise the shadow org. The point is to stop pretending it is not there while you take a chainsaw to it.

Gravitational lensing for the working manager

You cannot photograph dark matter. You can watch it bend light. The same trick works here, and none of it requires a consultant.

• Ask two questions of everybody. Whom do you go to when you are stuck, and who comes to you. Draw the overlap. That is your real chart, and it will not resemble the official one.
• Follow the traffic. Where do the fast threads live. Which channel gets the screenshot at eleven at night.
• Run the vanishing test. If this person took a month's leave with no phone, what would quietly stall. Not what would fail loudly - loud failures already have owners. You are hunting for the silent ones.
• Count the crossings. Anyone routinely operating across three or more functions is a broker. Brokers are your highest yield people, your least visible people, and statistically the ones most likely to be reorganised into a box that severs their reach.
• Interrogate every fast exception. Nothing moves through a large company quickly on merit alone. Every unusually fast outcome is a relationship wearing a process costume.

And if you are about to redraw the thing anyway, at least do it with the lights on. Name the brokers before the boxes. Leave the channels alone. Pair every new boundary with an explicit bridge - a named person, a ritual, or a standing forum with genuine decision rights, rather than a hopeful arrow. Then budget for reformation. Assume two or three quarters of degraded coordination while the invisible mass reassembles itself, put that in the plan, and have the decency not to call it fatigue.

The lens cap

An org chart is a photograph of a galaxy taken with the lens cap halfway on. Everything in the frame is real. Almost nothing holding it together is in the frame.

The velocity is the truth. The structure is only a hypothesis about the velocity.

So the next time somebody proposes a bold simplification of the structure, ask them one question and listen very carefully to the answer. What is currently holding this up. If they can only answer with the chart, they have just told you they cannot see five sixths of the company they are about to rearrange.