I just saw a tweet saying: “A “digital twin” is just a pretentious name for a model.”

I don’t think that is entirely correct. A “digital twin” is a model, but it is a specific kind of model. Imagine a model of a photovoltaic power station (a pretentious name for a solar park). You can build a model of a solar park and fit the parameters to some plausible solar park that you think could exist somewhere. You can also fit the parameters of the model to the average solar park if you have access to that kind of data. Neither of those models is a digital twin. A digital twin is a model that is fit to a specific entity that exists somewhere. Furthermore, the digital twin usually comes with some kind of data pipeline and simulation-based inference method that keeps the model parameters in line with the existing entity.

Keeping the digital twin well fit to the actual solar park makes sure that it works well as a generative model or for forecasting output of that specific park . Therefore, pretentious or not, I think “digital twin” describes a specific type of model.