3 ms·

Bullshit, bullshit everywhere. An abstract model can predict nothing, especially when the system is not fully observable, discrete, fully deterministic. Abstr

• betwixthewires 5y ago
  As said above, no models are correct, some models are useful. I see the statement as akin (in some ways) to lies, damn lies and statistics, or the spherical cow. The models cannot predict the future. Too many charlatans have attempted to tell the public that they can, but they can't, and as a result people look for models that can, criteria by which they can trust models based on the credentials of their creators or the granularity of the source data, even reasonable implementation of the source data, but of course, no model can. But a model can teach you something about the mechanisms behind the interplay between different systems. It can tell you something general about the effects of certain situations. A good example of this can be seen in Lorenz's weather prediction models leading to the formal discipline of chaos theory. Often, algorithmic interaction between different values in models can lead to a discovery of a previously unknown mechanism by which something about the world can be gleaned, even if the model itself is useless at accurately predicting the future. But you're right, a model can predict nothing, and people need to bear that in mind, because pop science has people convinced that they should live their lives as if existing models can and do predict the future when in fact they do not.