Paving is not the answer

To tackle America’s congestion problem by building new roads would be hideously expensive, not workable at all in many metro areas (where’s little-to-no space to add roads), and almost certainly futile over the long run. But what this natural experiment suggests [a small decrease in drivers yields a large decrease in congestion] is that to tackle congestion with “demand management” policies such as congestion pricing and market-rate parking would be relatively easy. In other words, you would only need to decrease the volume of peak-hour traffic by a pretty modest amount in order to produce a dramatic gain in the ease with which traffic flows. Meanwhile, the modest charge would generate some revenues that could be used for public purposes.

[From Matthew Yglesias » Driving Down Slightly, Congestion Down Dramatically ]