Signal
The temptation with simulation is to treat it as a crystal ball: build a good enough model and it will tell you what to do. That gets the value backwards. A simulation's real work happens earlier — in the room, before the decision, when it puts something concrete on the table.
The more abstract and many-sided a problem is, the more any piece of concreteness helps. A simulation frames the conversation. It anchors an abstract argument to something everyone can point at, push on, and disagree about in the same terms. That framing — not the forecast — is where the value starts.
Analysis
This is why a simulation belongs at the centre of a discussion, not at the end of it. Researchers who study collaboration call this kind of artifact a boundary object: something concrete enough that groups who see the world differently can still work around it without first having to agree. A simulation is a boundary object you can run.
A digital twin doesn't have to be a copy of a product. It can be a copy of a whole system — dozens or hundreds of interacting processes — and once it is, you can run scenarios against it endlessly. That is what turns it from a model into a decision instrument: not that it is right, but that it lets a group ask "what if" cheaply, many times, together. This is the idea behind what MIT's John Sterman calls management flight simulators — compress space and time so a team can feel the long-term consequences of a decision before committing to it.
Two cautions, both earned in practice. First, a simulation never captures the whole. It never reproduces reality's full complexity, and it never includes the human factors that decide how things actually go. All models are wrong, but some are useful — the discipline is to use it for the useful part and keep human judgment in the loop for the rest. Second, simulating one or two variables is worse than useless: it looks rigorous while missing the point. A system's behaviour emerges from forces pulling against each other — the value is in the interaction, the feedback, the thresholds. So the real preparation isn't building the model. It is listing, honestly, everything that bears on the whole — and accepting that anything you leave out, you have quietly decided doesn't matter.
References
- Star, S. L. & Griesemer, J. R. (1989), Institutional Ecology, "Translations" and Boundary Objects — concrete artifacts that let different groups collaborate without first reaching consensus. Boundary object (overview)
- John Sterman, MIT — management flight simulators and system dynamics for decision-making in complex systems. Business Dynamics · Management Flight Simulators (paper)
- George Box — "All models are wrong, but some are useful." All models are wrong