116 Hamilton Place Vernon Hills, IL 60061-1041 January 14, 2000
	Sierra Club Woods & Wetlands Group

Computers and Tollways

Figures don’t lie, but liars figure. The computer programs used by LCTIP to compare the congestion of their 8 competing transportation improvement scenarios are just the fancy way they figure. In spite of requests that they publish their models on the internet so that we can check how they drew their most recent set of projections, they insist that we be satisfied with just their answers.

They claim their computer programs meet the challenge of predicting human behavior. All the programs rely on census data that revealed where people lived, worked, and shopped - in 1990. Each model is built on assumptions and sensitivities which the programmers adjust to get the answers that they think make sense. But we knew what the Tollway thought would make sense before they created LCTIP:   Rt. 53.

The best way to check whether one of these computer programs can predict the future is to ask it to predict the past, and see whether it gets the answer right. The most overlooked and remarkable thing about the Crossroads Study is that it’s answers matched our known data better than the models used by IDOT and the Tollway. This was especially true when they looked at actual peak congestion levels that are the big problem. Now, amid claims that their model can be trusted and the Crossroads model is wrong, LCTIP wants us to believe that their comparative scenarios are suddenly, magically, accurate. They claim that, because the answers came from a computer model, they must be unbiased. Should we believe them this time, or is this another case of "garbage in, garbage out"?
 

Evan L. Craig Chair