Every system has those three parts: a toaster, a forest, a business, a town. Most trouble starts when one is taken for another, or when two separate things are counted as one. Sort yours here.
In a thing or a value→System a verb→Out the product
Getting the three right is the start. A system gives a clean, steady result when all five of these hold.
If one can change while the other stays the same, they are two things. A wage bill can fall while the people stay. The people can change while the wage bill stays.
A result can feed the next round, or another system. Trouble starts when the result is made the thing that directs the work.
A result that comes back and pushes against a change steadies the system. One that adds to the change makes it run away.
A correction based on how things were a while ago pushes too far, then pulls back too far.
An input the system cannot use produces nothing. One that looks right and changes how the system works is the harder case.
What goes in is a thing or a value. What comes out is the product. What sits between them is a verb: it does something. Put a thing where the verb belongs and watch the curve come apart.
Set what goes in, how hard the system works on it, and where the output goes. The chart shows what comes out, step by step.
A toy with made-up numbers. It shows the shape each arrangement makes, not a forecast of anything.
Picture a system as a landscape and its state as a ball. The ball rests in a hollow. A system holds steady because of a curve; a flat one holds nothing in place. The even curve is the steadiest of the three. The other two are that same curve leaned over, or turned over in the middle.
The shape of an arch or a bowl. The further it is pushed, the harder it is pulled back, and equally from either side. So small errors cancel instead of building up.
Good for anything meant to hold steady. Costs nothing but the upkeep.
The bowl with one wall lower than the other. It still rests, but off centre, and it gives more easily one way than the other.
Good for a valve or a ratchet, which are meant to favour one direction. In something meant to be balanced it is a distortion: one side moves the whole more easily than the other can move it back.
The bowl with its middle turned over. Two hollows, a hump between them, and nowhere to rest in the centre.
Good for a switch, which must be fully one thing or the other. In something meant to agree it depends how deep. A shallow split has a low hump that is easily crossed, and the two sides still work as one. A deep split has a high one, no common ground, and each side holds the other in place.
Hot iron is even: its parts point every way and it has no direction. Cooled, it is split: it must settle as north one way or the other. Hold a magnet near and it is lopsided, pulled to one side. The two poles that result are equal and opposite, so for all their pull they add up to nothing.
Most of what makes a system interesting happens when one of the three ends up where another belongs. Four of these are how systems work. The fifth is a mix-up.
The structure is the same wherever something builds up and feeds back. What changes is how well it can be predicted: exactly in a machine, and least of all where people can see the pattern and change what they do.
None of these is about any one kind of system. Each is something that tends to be missed.
Skill, trust, stock and heat all sit inside a system and drain slowly. That is why a change can look fine for a long time before its effect shows.
A loop that feeds itself always meets a limit sooner or later: room, fuel, customers, patience.
Work, weather and demand arrive unevenly. Spare room that looks idle on an average day is what absorbs a busy one.
Darkness does not spread; light is missing. When nothing comes out, look first at what is not going in.
Would another capable person in the same seat, with the same information, have done the same? If so, the system produced the result.