Abstract: Many legal scholars and urban planners extol the virtues of the great American downtown grid: the uniform blocks and parallel streets that dominate cities from New York to San Francisco. Against this precision, the serpentine roads of many early American towns are viewed derisively, as an undesirable consequence of disorganized colonization. The history of America’s first planned city offers a natural experiment for examining the legal and economic consequences of both types of layouts — and evidence about when the conventional wisdom on grids is wrong. This Article tells the story of the failure of America’s first city plan: the Nine Squares grid in New Haven, Connecticut. The Squares were problematic from their inception because they were too large and improvidently located. To adapt to land conditions and a commercial future far from what the town’s founders anticipated, eighteenth-century civic leaders resorted to a variety of processes to revise the layout, including a major subdivision that required use of the eminent domain power without payment of compensation in the 1780s. Town planning within the grid contrasted sharply with planning in areas surrounding the grid during the same time frame. In other parts of New Haven, incremental street decisions, legal mechanisms for resident involvement, and laws permitting in-kind compensation for new roadways allowed the town responsively to plan streets suited to changing land and settlement conditions. This Article advances a new theory of street planning drawn from the New Haven case study, aiming to surface the virtues piecemeal planning can bring during some points in a city’s development. Streets can be thought of in market terms, and comprehensive grid plans may act as market distortions, preventing settlement forces from organically producing more effective street layouts. Particularly where information about land is dispersed among members of a small population, bottom-up street plans may be desirable because they reflect residents’ preferences and harness dispersed knowledge about land conditions and uses.