I live in Denver and the parts of it laid out up to what I guess is the 1950s are mostly a coherent and easily navigable grid pattern, with some exceptions. But everything after that time is just a meandering non-sensical jumble of roads. But its not just Denver its every city I've ever lived in. What happened?
In Denver, and most other cities of the American West, the difference is primarily on the residential "side streets." The grid of section-line roads at one-mile intervals (92nd Ave., 104th Ave., 120th Ave., 136th Ave.; Chambers, Buckley, Tower, etc.) generally either was there prior to development or was strengthened in conjunction with development. Perhaps for topographic reasons, very few section-line roads were ever built in the areas south of Denver, so even arterial patterns have been left to modern planners and traffic engineers.
The sections marked by these roads (the Public Land Survey System ) were the basic module for sale of farm- and ranchland in the 19th century. East of the Mississippi, farms were typically 40 or 80 acres; west of there, settlers often were allowed to homestead entire quarter-sections of 160 acres. Out West, in ranch areas, entire square-mile sections could be claimed. As later urbanization engulfed farmland, subdividing a 40-acre farm parcel with two east-west and four north-south streets was a quick and easy way to plat an addition to a town and sell new houselots.
The undifferentiated gridiron layout for “additions” to cities was criticized in the early 20th century as insensitive to topography, wasteful, and ill-suited for growing auto traffic. When the Federal Housing Administration was established in 1934 to guarantee mortgages, homebuilders saw it as devoted to ensuring the success of their projects rather than as governmental interference, so they were much more receptive to FHA’s underwriting criteria than to local subdivision regulations with similar aims. FHA’s 1935 technical standards suggested that streets should fit the local topography and have widths suited to their requirements (not be overbuilt on the assumption that every lot might someday contain a factory or apartment house). FHA standards also promoted the “functional classification” familiar to us (arterials, collectors, local streets) that had been made necessary by the growth of auto traffic. “Best principles” publicized by homebuilder organizations Urban Land Institute and National Association of Home Builders and by planning officials of American Society of Planning Officials and the American Institute of Planners during the 1940s followed similar concepts, recommending loop and cul-de-sac layouts that followed local topography and efficiently subdivided often-irregular parcels. Cul-de-sacs branching off collector streets were promoted as the most attractive layout for single-family dwellings.
Postwar homebuilders initially turned to modified grids to relieve the “monotony” of long, straight local streets, but research in the 1950s by traffic engineers seemed to show dramatically lower accident rates for T intersections and bending streets. By the 1970s, homebuilders increasingly favored cul-de-sacs even in places where the terrain was completely flat.
From the late 1980s, planners began to note the drawbacks for pedestrians and cyclists of cul-de-sacs and the functional classification system. The movement initially known as neotraditional development (later the New Urbanism) promoted permeable networks that didn’t push all traffic onto collectors and arterials, and that allowed pedestrians and cyclists to take reasonably direct paths through neighborhoods.
Probably the best source on this subject is Michael Southworth and Eran Ben-Joseph, Streets and the Shaping of Towns and Cities, though it can be instructive to look at ASPO/AIP/APA publications for city planners, and at the ULI Community Builder’s Handbook and Residential Streets.