Run it before anybody lives in it
A diagram of a process is a claim about structure, and it is read exactly once. You look at it, you find it broadly right, and what you learn is roughly what you suspected before you looked. Boxes and arrows are very good at saying what the parts are. They are silent about the only thing anybody actually experiences, which is a Tuesday.
Draw the same work as commitments and something becomes possible that a diagram cannot do: you can run it. Not as a metaphor — the acts happen in an order, one at a time, and at every moment somebody is holding something they have not yet answered.
An order the drawing only implies
On the page, a network is all there at once. Every loop is drawn, every line joins two of them, and the eye takes it in as a shape. But a network is not a shape. It is a sequence with a great deal of waiting in it, and the sequence is implied by the drawing rather than shown by it.
Running it makes the implication explicit. The request goes out; the answer comes back, and it is not always agreement; the work happens somewhere below and takes as long as it takes; the report arrives; somebody accepts it, or does not. Each of those is one act by one named person, and watching them arrive in order is the first time the drawing has a tempo.
The tempo is where the surprises are. A network that looks balanced on the page routinely turns out to ask nothing of somebody for a fortnight and then three things in an afternoon.
Whose day it is
The thing this produces that other tools do not is not a statistic. Process simulation, in the ordinary sense, answers questions of quantity: throughput, utilisation, how many of these can we push through a week. Useful questions, and none of them is the question a person has about their own work.
Open a work inbox during a run and you get the other thing. Here is what is on Sofia's plate, right now, in this design. Here is what she cannot start, and the name of the person she is waiting for. Here is what she will be asked for, and roughly when it lands. A network is a diagram of an organisation; an inbox is a description of somebody's morning, and it is the same information read from inside.
That is a strange and useful vantage point, because a process is almost always designed from above and lived from below. The person drawing it sees a flow. The person in it sees a queue with names on it, and the two of them will not have the same complaints.
What a run finds that the drawing does not
Four things turn up so often they are worth naming.
A moment nobody stands behind. A step performed by a rule rather than a person — a threshold clears, a job fires. Drawn, it looks like any other. Run, the act arrives and no name is attached to it, and somebody in the room says wait, who decided that?
A loop nobody closes. The run reaches its end and something is still owed. The work was done, it was reported, and nobody ever said yes, that was it. On paper this is invisible, because the loop is drawn complete. In motion it is a thing left over.
Somebody who cannot say no. A performer whose agreement was never in question, because the design never gave them anywhere to put a refusal. It reads as efficiency on the page. In a run, watching an act go past that nobody could have stopped, it reads as what it is.
A promise waiting on the wrong act. Two loops joined by a line, and the line leaves from a report rather than from an acceptance. So the next piece of work starts on somebody's claim that they finished, rather than on the customer's agreement that they did. Both are legitimate designs and they behave very differently, and almost nobody notices which one they have drawn until the run walks the line.
Being in it rather than watching it
There is a second way to run one, and it is the one worth trying: say who you are, and let it stop when something is yours to do.
The run pauses and waits. Nothing advances until you take the act, and the act is a real choice — agree, counter, decline. What that does is convert an argument into an experience. Being told that a design leaves you no room to decline is a debating point. Reaching a moment where declining is not among the things you can say is a different kind of knowledge, and it arrives in about four seconds.
It is also the cheapest way to find out whether a process is asking something unreasonable of one person, which is to sit in their seat for a minute before asking them to sit in it for a year.
Before it is built, or after it breaks
Two moments make this worth the drawing.
The first is design. Somebody is putting a way of working together — a handover between two teams, an approval path, an onboarding — and it exists so far as a document and a set of assumptions. Running it is a rehearsal, and rehearsal is cheap. This is sharpest where the design has software in it that acts. A model handed a tool and a goal is a party in a conversation for action, and the question that matters about it is who accepts its work. A run answers that one the direct way: it shows you the acceptance moment, with nobody's name on it.
The second is repair, and it is the likelier one, because hardly anybody draws a process before building it and everybody has one that hurts. Take the handover people complain about, draw it as it actually goes rather than as the procedure says, and run it. The complaint is usually that something is slow. What the run tends to show is that something is never closed, or that two people both believe the other is waiting.
An instrument for designing, not for running the work
Everything above is about design, and the difference is worth being exact about.
A run is for working something out. You draw how coordination is meant to go, watch what that asks of the people in it, change the drawing, and watch again — and then you go and have the conversation the drawing turned up. What comes out of it is a better design and a sharper question, and both of those end up in people's heads and in the way the work is arranged afterwards. This is not where the work is then done, and not where it is tracked.
That is the whole of the boundary, and the reason it is worth holding is not the one usually given. It is not that the logs are dangerous; it is that they cannot say what any of it meant. What it does to software has that argument at length.
What it does not tell you
It is a rehearsal and not a forecast, and the difference matters enough to be plain about.
Everything here models work, not waiting. In the run, an answer comes the moment it is asked for. In life a request sent on Friday is answered on Monday, and the fortnight a network takes on the page is rarely the fortnight it takes in the building. The figures are the earliest the design allows if nobody sits on anything, and nobody has ever not sat on anything.
Nor does it tell you what people will do. It tells you what the design asks of them — whose turn, whose plate, who may refuse, what is left owed at the end. People will route around a bad design in ways no notation predicts, and the fact that they have to is itself the finding.
And it is only as honest as the drawing. A network drawn the way the process is supposed to work will run beautifully and teach you nothing. The value is entirely in drawing it the way it actually goes, including the moments where the honest answer is that nobody knows who is meant to accept.
Where to start
Whatever you draw first, run it twice. Once from outside, watching the acts arrive in order, which is where the missing acceptance and the moment with no name on it turn up. Then again as the person in it who complains the most. Pay attention to the stretch where the run is moving and they have nothing to do: that is the waiting they have been trying to describe to you.
The cheapest version of all of this is to watch somebody else's: there is a network already drawn, and pressing play walks it act by act with the work inbox open beside it. It takes about a minute, and the two things worth noticing are the moment nobody's name is attached to, and the loop that is still open when the run stops.
What it costs and what it buys has the advice on choosing the first thing to draw, and the honest account of what happens to a team that starts doing this. The vocabulary has the terms with their exact meanings.