“I wanted something you could quickly sketch on a steamy shower window — and it would still make sense and play.”

Machinations draws the system as a diagram you can Play. You see every resource, rate and loop, change any of them, and find out what happens before a decision costs anything real. It rests on three things.

System dynamics studies how the parts of a system act on one another over time: stocks that fill and drain, the flows between them, and the feedback loops that carry results back to their causes. Some fifty years ago it was chalk on a blackboard and a simulation left to run overnight. Today a sentence drafts the model, an AI agent can Play a thousand futures in minutes, and the whole team can still see why each one happened. Machinations is built for that second half.

Donella Meadows

Donella Meadows was one of the field's forerunners, and its clearest voice. An environmental scientist at Dartmouth College, she co-wrote The Limits to Growth (1972) and wrote about systems in words anyone could follow. Her essay Leverage Points: Places to Intervene in a System (1999) lists twelve places to push on a system, from a single number to the mindset the system comes out of.

In this 1977 talk she argues that a model is a set of assumptions. Machinations is built on that idea: put the assumptions where everyone can see them.

The three-part Machinations series Limits to Growth explained answers the book's population questions with models you can Play.

Where the field lives today

John Sterman directs the System Dynamics Group at MIT Sloan and wrote Business Dynamics (2000). Peter Senge took systems thinking to managers in The Fifth Discipline (1990).

• MIT Sloan System Dynamics GroupWhere the field began. Its researchers still build simulation models for business and public policy.
• System Dynamics SocietyThe professional society for the field. It runs the International System Dynamics Conference.

Teaching, researching or studying systems? The Academia Plan makes Machinations free for eligible teachers, researchers and students.

Describe the system in a sentence and the AI Agent drafts the diagram; every node, rate and loop stays visible and editable. Play it step by step to watch resources move, forecast many runs to see the range of outcomes, and balance it to find the inputs that move the result.

Your AI assistant comes along: connect Claude, ChatGPT or Cursor through Machinations MCP and it builds and Plays models for you. So do your own numbers: Google Sheets, the API and your files feed the model real values (connect your data and AI).

Joris Dormans studied how the rules of a game give rise to play nobody scripted. His 2012 doctoral research, “Engineering emergence: applied theory for game design”, set out the theory, and in 2013 he showed the visual language at GDC Europe.

In 2015 a small team joined him to move it from Flash into a collaborative platform in the browser. Machinations.io opened to everyone in 2019, and HIRO Capital led a Series A in 2022. Games remain its core; the Machinations Manifesto sets out how to build sustainable game economies. The same models now answer questions far from games.

Economies are not drawn once. They run every day and produce data, and AI systems now have economies of their own: agents spend budgets and queue for shared compute.

So Machinations is building the loop around the model, step by step. It starts from your data, builds the model, tests the options, puts the decision live and watches what it changes, with AI agents doing work at every step. Each turn starts from what the last one learned.

A system is easier to change well when people can see it. That is why Machinations is a visual language before it is anything else: every resource, rule and loop has a shape, and a diagram that reads clearly is usually a model that thinks clearly. We care that it is beautiful too, because people only question what they can read.

Donella Meadows hoped that seeing systems whole would help people live within them more wisely. We share that hope, in plainer words: more decisions made with their consequences in view, by the people who will live with them. In economies, in companies and, as AI agents take on more of both, in the systems those agents work inside.

It is also why Machinations is free for universities: the habit of thinking in systems starts in a classroom. You can help fund that free use.

The world can feel frightening, and the future more frightening still. Yet economic behaviour is one of the oldest archetypes in people: more than 100,000 years ago, shell beads were already travelling hand to hand, far inland from the shores where they were picked up. From those shells to coins, markets and now AI agents that trade on our behalf, every economy has been a system of people and rules. As Edward Castronova put it: a piece that might seem confusing at first is actually part of an interactive system that has logic to it. Understanding that logic, and improving it for the good of everyone who lives inside it, will always be needed. It can start today, with one diagram.