Functionally, putting aside being pedantic about tagging-for-the-renderer, the purpose of the “highway=” key is indicate how dominantly to show a given road on a map. It is a measure of the importance of a road in the network - an idea put to, and approved by, a vote (albiet an old one). Physical characteristic, if that is what one wants to render, can be described effectively by other tags (motorroad=, lanes=, oneway=). Especially in the U.S., the necessity of an “importance” tag makes sense, because physical road characteristics do not necessarily indicate any kind of network importance. There are countless high-speed, limited-access, divided, multi-lane, non-motorway roads that service small, far-flung suburbs. There are also many examples of plain old two-lane roads that are as crucial to the transportation network as our interstates.

The “expressway” definition works in many European countries for the same reason the U.S. has very little debate about the “motorway” definition. Expressways are signposted and clear-cut, in the same way that there is zero confusion about whether or not you are on a freeway in the U.S. (barring fringe cases like super-twos & rural western Texas at-grades). Here’s a list of expressway signage used in different European countries. This can also be seen on our international highway equivalence guide here. Roads signposted with expressway signage in these countries are tagged trunk. Makes sense! The U.S., outside of cases like the Santa Clara expressway system mentioned above (which are obvious candidates for trunk status), does not have such clear cut definitions. There exists MUTCD standards for “expressway” signage (similar to our “begin/end freeway” signage), but it is rarely used outside of cut-and-dry expressway systems, again like the Santa Clara system. Even on OSM, if someone proposes a definition with a criteria more objective than “it’s kinda like a freeway but not really”, you can almost always find a road that meets the criteria proposed, but is not so important that it needs to be shown at low zoom.

Instead of thinking of trunk as a demotion from a motorway, I think the U.S. would benefit from treating trunks as the most important roads, with motorway as a qualifier for trunk indicating that the road is a freeway. Every U.S. road atlas ever put to paper works like this, for good reason. Take a look at Canada. Maybe their definition is easier for them to work with since they don’t have an interstate system like the U.S, but the road map there works beautifully. It is simple to understand how to get around between major cities. This is mostly not the case in the U.S., because we treat primary as most important, and then upgrade sections to trunk when they exhibit freeway-like features. One of the more egregious examples I have seen of this is at the center of the map here. You can almost see that there are two important, cross-country routes through Kansas, Colorado, and Oklahoma. But you can’t really, because we’ve insisted that only the grade-separated portions are trunks. If you zoom in, you find a sea of primary roads that provide little illumination as to which routes you would use for cross-country travel. Examples similar to this are peppered throughout the U.S., which I don’t understand at all. What map consumers, aside from highway engineers, are benefited by seeing in very obvious terms, when a two-lane road has a grade-separated interchange or shortly turns divided as it circumscribes a town?

A “most important road in the country that’s not a freeway” definition for trunk de facto includes nearly all bone-fide expressway systems, so it doesn’t affect how most U.S. mappers are already using this tag. This definition would allow us to do away with having randomly peppered trunk segments scattered among a sea of primary roads, and instead build a more useful and cohesive road map. Maybe “most important non-freeway road” is just as vague as “it’s like a freeway but only kind of”, but we have been using this definition with the primary to unclassified spectrum for years without too much trouble. The local & consensus-based approach to mapping is excellent for its flexibility (important when trying to topologize the entire Earth!), but I think it falls a bit short when it no consensus can be reached on something as fundamental as the “highway” tag.