Skip to content

All Models Are Wrong…

July 18, 2012

By Matt Artz

The world around us is a complex place, and one way we manage that complexity is through a process of abstraction. In its purest sense, abstraction is a reduction of detail down to the bare essentials we still need in order to understand.

Maps are a fascinating example of abstraction.  Maps are abstractions of landscapes and geography, and have proven to be a particularly useful aspect of human technology throughout our history.  Until relatively recently, maps were predominantly two-dimensional: paper maps with complex geography abstracted onto a flat surface.  New methods of presentation were created in an attempt to relay geographic information that moved beyond the two dimensions, but these methods, while useful, often fell short of conveying the true nature of complex and dynamic nature of geography.

Enter computers.  The move from paper-based abstractions towards computer-based abstractions of geographic space has given us a powerful new context for understanding—and not just for two-dimensional landscapes, but for geography spanning the third and fourth dimensions as well.

It’s that fourth dimension—trying to understand what happened in the past or what might happen in the future— where things can get really complicated.

Geospatial professionals are often faced with tasks that involve modeling—the attempt to simulate what might (or did) happen in a particular system through a process of abstraction and simplification.  But the process of reducing a complex system down to its essence without losing important details is fraught with uncertainty and peril.

It’s important to grasp that more than just individual models are at play; complex systems such as earth’s climate require multiple models of component systems.  We’ve reached a high level of sophistication with many individual models, and progress is being made at integration between models.  But we need a more comprehensive and open method of consolidating and relating inputs and outputs from all models.

While much progress has been made in recent years to develop models to help us to better understand our world in the context of these domains, there is still much more to be done at the macro scale—especially in the area of integration.  As we gain more detailed understanding of different granular systems and their components, the challenge in addressing complex issues such as global climate change is coupling these models together to gain a more complete picture.  The combination of powerful hardware, sophisticated software, and increased human knowledge have all contributed to better models and more accurate simulations, but a geographic information system (GIS)-based framework for integrating these disparate representations of past, present, and future states is key to understanding the whole earth.

GIS itself is an incredibly valuable tool for spatial analysis and modeling, but there are a many standalone models available designed for highly specialized, domain-specific modeling, analysis, and problem solving.   Most domain-specific models are not yet and probably never will be fully implemented in a GIS framework; however, the spatial display, analysis, and data management capabilities of GIS can still be utilized to greatly streamline almost any modeling workflow.

GIS enables a comprehensive modeling framework where the software is used for workflow management and post-modeling support for multiple domain-specific models; in addition, outputs from multiple models can be compared, analyzed, and modeled within the GIS system itself.  Such a GIS-based framework offers a comprehensive environment for modeling across complex earth systems.

Creating a framework that successfully brings together and manages a plethora of data sources and modeling systems to tackle the most pressing environmental issues of our time is surely a monumental challenge, but it is a challenge for which GIS is well suited.  Once the data and technology framework is in place and a clear workflow is established, the challenge then becomes organizing a large group of people to do the work of modeling multiple complex scenarios in order to identify the best of possible design futures for the planet.

“Essentially, all models are wrong, but some are useful.”
George E. P. Box, Statistician

Prediction is a tricky business.  It’s a mixture of science an art.

“All models are wrong…”

From the perspective of science, we’re looking for absolutes, and the world of absolutes is where modeling often falls short: models rarely predict things with 100% accuracy.  If you evaluate the success of a model in such absolute terms, you will almost certainly be unhappy with the results.

“…but some models are useful.”

This is where the “art” comes in to play.  Art is about creation.  And this is where modeling has the greatest promise: by helping us to understand what may happen in the future, we can make better decisions today; we can plan for, design, and actually create a better future.

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: