Every org chart I have ever drawn was a lie I told to feel in control.
I drew them in slide decks at two in the morning, boxes nested under boxes, dotted lines that meant kind of reports to, solid lines that meant definitely blames. I drew them for board meetings and for myself, and the act of drawing was soothing in the way that arranging the deck chairs is soothing. Here is the marketing function. Here is engineering. Here is the new VP we just hired to make the chart look the way charts are supposed to look at our stage. The chart said the company was a tree. The company was not a tree. The company was a tangle of people texting each other, ignoring the boxes, routing around the boxes, getting work done in spite of the boxes.
I kept the lie because the truth was harder to draw. The truth is that an org chart is not a plan. It is a fossil.
What carved the channel
Think of a river. No one designs a river. A river is the record of where water found it cheapest to go. Soft ground erodes, hard ground stays, and over enough seasons the water carves a channel that looks, to anyone arriving late, like a deliberate shape. It is deliberate the way a callus is deliberate. It is the accumulated memory of friction.
A company is a riverbed, and the water is information. The org chart is the channel the information cut. We pretend we drew the chart and then the work flowed through it. We have it backwards. The work flowed, and the chart is what was left where it flowed.
This is not a new observation, though we keep rediscovering it and acting surprised. In 1967 a programmer named Melvin Conway noticed that the software a group ships will mirror the communication structure of the group that shipped it. Four teams writing a compiler produce a four-pass compiler. The seams in the product fall exactly where the conversations were expensive. We named it Conway's Law and then mostly ignored it, because it implies something inconvenient: that you cannot draw a better product without first redrawing who talks to whom, and you cannot redraw who talks to whom by issuing a memo.
The deeper version of Conway's point is the one nobody quotes. The org is shaped like its communication costs. Where talking is cheap, the org runs thin and fast. Where talking is expensive — across a time zone, across a function, across the gap between someone who knows and someone who needs to know — the org grows a structure to manage the expense. Managers exist, in large part, to be the place information stops so it can be repackaged for the next layer. A layer of management is a toll booth on a road that information was always going to travel anyway.
I am not being cynical about managers. I have been one. The good ones earn the toll. But it is worth saying plainly what the structure is for, because the moment the cost it was built to manage changes, the structure has no further reason to exist, and it will not notice this on its own. Riverbeds do not relocate themselves out of politeness when the rain moves.
The hidden tax
Most of what a large company does is not the work. It is the coordination of the work.
Consider what actually happens when a decision crosses a boundary. A salesperson learns something true about a customer. To make that truth useful, she has to encode it — a note, a ticket, a message — strip it of the context she carries in her head, and hand it to someone who will decode it imperfectly, lose half the nuance, and ask three follow-up questions across two days. Multiply that by every handoff in the building. The product manager translating engineering into roadmap. The roadmap translating into a deck. The deck translating into a quarter. Each translation is lossy. Each loss is paid for with a meeting.
We have a word for this tax and we say it with a straight face: alignment. Whole weeks dissolve into alignment. We are not aligning because alignment is good. We are aligning because the context that one person holds cannot be cheaply moved into another person's head, so we convene the heads in a room and hope proximity does what bandwidth cannot. The all-hands, the standup, the sync, the offsite — these are not culture. They are the visible cost of moving context through human beings who can only hold so much of it at once.
For a hundred years that cost was fixed, or near enough. You could shave it with better process, flatter hierarchy, a tool that made the handoff slightly less lossy. But the underlying constraint — that a human can hold only a few things in working memory, and that moving the contents of one mind into another is slow and expensive — that was bedrock. Every org structure we have ever admired was an arrangement of that bedrock. The matrix, the squad, the pod, the holacracy: variations on the question given that context is expensive to move, who should sit near whom?
What happens when the rain changes
Now hold a different fact next to that one.
This month a model arrived that can read a quarter's worth of a team's documents, the threads and the tickets and the half-finished spec, and hold all of it at once, and answer a question about it as if it had been in every meeting. It is not the first such model and it will not be the last. The capability is no longer a demo; it is becoming ambient, the way search became ambient. What that capability does, underneath the chatter about productivity, is attack the exact constraint the org chart was built around.
If the context a team used to pass from head to head can instead sit in a place that anyone — or anything — can query in full, then the toll booth is on a road that no longer needs a toll. The reason the marketing box and the engineering box stayed separate was partly that the cost of keeping them in sync was high. Lower that cost toward zero and the boundary stops being a feature of physics and becomes a choice. A choice you can revisit. A choice you probably should revisit, since you never chose it in the first place — the riverbed chose it for you, back when the rain fell differently.
I want to be careful here, because the easy version of this thought is wrong and a little embarrassing. The easy version says the org chart dies, hierarchy evaporates, everyone becomes a free agent with a robot. That is the kind of prediction that ages like milk. Coordination does not vanish; coordination is not only about moving context. There is judgment, accountability, the question of who gets fired when the thing breaks. Those don't dissolve because the bandwidth went up.
But the shape — the shape was never sacred. The shape was a cost structure wearing a costume. And when the cost underneath an old shape collapses, the shape becomes, for the first time in roughly a century, genuinely up for grabs. Not abolished. Negotiable.
That word is the whole essay. Negotiable.
A smaller observation that I trust more than the large one
Let me drop the altitude, because grand claims about the future of work are cheap and I have made my share.
The thing I actually notice, running a company, is smaller and stranger. When a piece of context becomes free to retrieve, the people around it change what they spend their time on, and they change it before anyone gives them permission. The person who used to be valuable because she was the only one who knew how the billing system really worked is suddenly valuable for a different reason — because she can ask better questions of a system that now also knows. Her moat moved. It moved from holding the context to interrogating it. She is doing more interesting work and she is, quietly, harder to slot into a box, because the box was drawn around the old moat.
Repeat that across a hundred people and you do not get a reorg. You get something subtler: a chart that is increasingly fictional, increasingly a polite story we tell newcomers, while the real work routes itself along channels the chart does not show. The lie I told at two in the morning to feel in control becomes, over a few years, simply false — not because anyone redrew it, but because the water moved and the old channel silted up.
The companies that will look prescient in a decade are not the ones that predict the new shape. Nobody can predict the new shape; rivers are not predictable in their particulars. The prescient ones will be the ones that stop mistaking the riverbed for the river. That hold their structure loosely. That treat the org chart as a hypothesis about communication cost, and recheck the hypothesis when the cost changes — which, right now, it is changing fast.
What this asks of you
So here is what I would do, and what I am doing, which is most of why I believe it.
Stop drawing the chart to look like the company you are supposed to be at your stage. Draw it, if you must draw it, to match where information is actually cheap to move today, and accept that the drawing will be wrong again by the time it prints. Find the toll booths — the places where context stops to be repackaged for the next layer — and ask, honestly, whether the toll is still buying anything now that the road has changed. Some will be. Many will be habit wearing the costume of structure.
And notice the people whose value just quietly moved, the ones who got better at asking and lighter at holding. Build around them. They are standing where the new channel is cutting through, ankle-deep in water nobody has named yet.
The org chart was always a communication diagram. We just got to pretend it was a map of authority because, for a hundred years, the two were carved by the same river. They are coming apart now. The boxes will stay on the slide a while longer, because boxes are comforting and the deck is due Thursday. But the water has already moved. It moved before we noticed, the way it always does.
The only real question is whether we redraw the chart on purpose — or wait, again, for the river to do it for us.