Simulation and Similarity: Using Models to Understand the World

by Michael Weisberg

In the 1950s, John Reber convinced many Californians that the best way to solve the state's water shortage problem was to dam up the San Francisco Bay. Against massive political pressure, Reber's opponents persuaded lawmakers that doing so would lead to disaster. They did this not by empirical measurement alone, but also through the construction of a model. Simulation and Similarity explains why this was a good strategy while simultaneously providing an account of modeling and idealization in modern scientific practice. Michael Weisberg focuses on concrete, mathematical, and computational models in his consideration of the nature of models, the practice of modeling, and nature of the relationship between models and real-world phenomena.
In addition to a careful analysis of physical, computational, and mathematical models, Simulation and Similarity offers a novel account of the model/world relationship. Breaking with the dominant tradition, which favors the analysis of this relation through logical notions such as isomorphism, Weisberg instead presents a similarity-based account called weighted feature matching. This account is developed with an eye to understanding how modeling is actually practiced. Consequently, it takes into account the ways in which scientists' theoretical goals shape both the applications and the analyses of their models.

Genres: Philosophy, Nonfiction, Science

212 pages, Kindle Edition

First published December 21, 2012

About the author

Librarian Note: There is more than one author in the Goodreads database with this name.

Reviews

[Alex · Rating 4 out of 5]

Review to come when I write up notes on a second read. Pleasantly short. Satisfying regarding the big picture but many details don't make sense for formal models, especially probability models.

[Katharina]

Why do scientists build models? What is a model anyway? How many different categories of models are there? Weisberg argues that there are only three different categories of models: Concrete ones, mathematical and computational models. A model is, in his view, more than just the description of its structure, it always needs to be combined with a "construal", the modeler's interpretation of his or her model.
For me as a scientist who tries to understand the interaction between social actors and software artifacts, this book was crucial for being a better modeler.

[Heather Browning · Rating 4 out of 5]

It seems like work on modelling is a really important direction in philosophy of science currently, and this book gives a really good overview on the different types of models, their uses and features, as well as making a strong argument for understanding the model-world relation as one of similarity, relative to the scientific context and goals of the model and modellers.

[Franck Chauvel · Rating 2 out of 5]

M. Weisberg explains the why and how of simulation as means to gain knowledge: Under what condition can we safely draw conclusions from few experiments carried out on "fake" world, be it smaller physical replicates, virtual worlds within a computer's memory or abstract mathematical formulae.

I found the book too theoretical, especially the beginning which I endured as a too long essay on Philosophy of Science. Yet, I liked the last three chapters: I understand better how I can discuss the "similarity" between my models and the real phenomenon they mimic. I liked as well the discussion about robustness analysis, which I also something I did learn about. These are applicable knowledge for anyone doing research using simulation.

[James Townsend · Rating 4 out of 5]

If you are a scientist of virtually any sort, you should at least consider reading the last 3 chapters of the book. Just saying.

That said, overall a very good account of the practice and pitfalls of modeling. A very interesting read.