his wikipedia summary
Fogel's first major study involving cliometrics was Railroads and American Economic Growth: Essays in Econometric History (1964). This tract sought to quantify the railroads' contribution to U.S. economic growth in the 19th century. Its argument and method were each rebuttals to a long line of non-numeric historical arguments that had ascribed much to expansionary effect to railroads without rigorous reference to economic data. Fogel argued against these previous historical arguments to show that onset of the railroad was not indispensable to the American economy. Examining the transportation of agricultural goods, Fogel compared the 1890 economy to a hypothetical 1890 economy in which transportation infrastructure was limited to wagons, canals, and natural waterways. Fogel pointed out that the absence of railroads would have substantially increased transportation costs from farms to primary markets, particularly in the Midwest, and changed the geographic location of agricultural production. Despite this consideration, the overall increase in transportation costs, i.e., the "social savings" attributable to railroads, was small – about 2.7% of 1890 GNP. The potential for substitute technologies, such as a more extensive canal system or improved roads, would have further lowered the importance of railroads. The conclusion that railroads were not indispensable to economic development made a controversial name for cliometrics.
An immediate problem with this argument is the supposition that canals, like railroads, can go anywhere. It is very hard to run a canal through a mountain range, without constructing many expensive locks ( as an example, consider the series of locks at Lockport NY just to get the Erie Canal up the Niagara escarpment...) It's hard to run a canal across a desert, because it's hard to have feeder streams and rivers keeping it watered if there are none. It can also be hard to run a canal in the depths of winter in higher latitudes, because it can be frozen solid.
Importantly, also, since a railroad can run through all these different landforms and climates, all of them can be a part of a network. Even though a canal is very good for some purposes ( slow transport of even very heavy objects requires very little energy) because it can't go everywhere, it is impossible to load three tons of gravel in New York and off load it in Seattle like you could with a railroad.
There's also speed: although slow transport on water takes very little energy, rapid transport can consume quite a lot: as velocity builds, so will a mass of water in front of the hull, that the hull essentially attempts to climb over. Unless someone made a quantum leap to planing hulls and even hovercraft, mass transit on canals would have to have been limited to a small fraction of railroad speed. 30 mph is easy for a train, hard for a canal boat.
Then there's repairs. The railroad from Washington DC to Cumberland MD was built at about the same time as the C&O Canal. The Canal was able to support itself hauling bulk commodities like coal and stone, but damage from floods was eventually just too expensive. After a big one in 1884, it was bought by the B&O Railroad and run as a subsidiary ( mostly to keep control of the right of way) until a major flood in 1924 caused so much damage that the canal was essentially abandoned.
It is hard to deny that some canals, like the Erie Canal, were enormously successful, especially the ones that connected big bodies of water: the Erie's success came from its connection of the Great Lakes to the Eastern seaboard. And it is possible to imagine further profitable building of them in places like the lowland South. But Fogel was stretching things, in trying to think of them as railroad substitutes.