That’d be my take too.
I’m curious about this - given that US 50 is explicitly a through route, but may not be great at local connections (because there’s almost nothing local), it feels like it actually does have use to convey that there’s a through route here. But it sounds like you’re thinking differently and I’m misunderstanding. I’d be genuinely interested in more of your thoughts on this.
Sorry, I got in the weeds in my own thoughts on that a bit. What i mean to say is that the Bend-Boise route is barely used more than 395 since it connects Boise to basically nowhere, whereas Twin Falls to Vegas basically soaks up all the the interstate traffic in that direction connecting two major endpoints.
By explicitly, do you mean due to it being a round-number US route? Certainly US 50 was originally designed as a long-haul route across the country. But with the advent of the interstates, I’m not sure if still constitutes the best connection between any two population centers, which is how I assess a road qualifying for trunk per the wiki guidelines. Basically, I agree with this:
For US 50 across Nevada, the best I can come up with as to which important population centers it connects is maybe Carson City, NV and Grand Junction, CO? But those two cities are 12 hours away from each other, so saying that’s an especially relevant routing seems iffy to me. Then, for what it’s worth, all three OSM routers suggest taking I-80 (though those depend on the road classifications themselves), and Google Maps says going via US 50 or I-80 takes about the same amount of time, in which case I’d guess most people would favor the interstate.
So, while being a primary US route is an argument in favor, I just don’t think that stretch of road really serves a major connective role in the current road network (as its “Loneliest Road” moniker suggests). But definitely interested in the discussion here, there’s definitely arguments both ways!
Yeah, that, and just that it goes basically coast to coast, from Sacramento, California to Ocean City, Maryland. Even if the segments that connect it all together don’t share a common design, it’s a route that’s designed to get you across the country on a single numbered route. In my mind, that makes it a through route, even if it’s not the best option for getting between population centers. I do understand what you’re saying though. I also recognize that my argument here is much ado about nothing - those that care about the through-route aspect can parse the relations to get the info.
It’s honestly probably fine as a primary, but also, when I think of Nevada, I think of 50 crossing right through the middle, so for a mostly-cartographic tag like highway=trunk, I lean toward it being appropriate.
Last thing is, I went looking at the wiki definitions, and I think they’re pretty good (though I imagine we’ll all read them a bit differently), in saying for the US that trunk can be assigned “in areas of medium to low population density, to the most important non-motorway roads that provide principal, long-haul connections between population centers of regional importance.”
I’d personally call Ely a population center of regional importance. I know it’s a small town, but being the only town for 100 miles I think helps it clear that bar pretty easily. I bring this up to say that, in my reading, we don’t even need to reconsider what it means to be trunk to consider US 50 through Nevada as trunk.
This also describes U.S. 6, which an author once compared unfavorably to U.S. 40:
By comparison, Route 6 runs uncertainly from nowhere to nowhere, scarcely to be followed from one end to the other, except by some devoted eccentric.
Multiples of 10 are more important in the U.S. Route numbering system, but system overall is a time capsule. Very little of U.S. 50 is currently classified as trunk in Ohio, because State Route 32 has thoroughly bypassed it. It would be redundant clutter on the map at low zoom levels. If we were having this discussion 60 years ago, it would be a different story.
Just noting that I think we broadly agree. In areas where there’s a better route, highlighting an old US Route doesn’t make sense (e.g. US 33 north south through California, running near I-5), but I was mostly just making the point that I do think it’s a through route, given both its local prominence and the fact that you can get to the other coast on it.
That might not be enough to mean it’s a trunk - I can see it either way
I gave it as an example similar to Dalton Highway or Route 50, in that it’s the only only major road connecting communities that function as the main population centers in vastly empty regions, i.e. a trunk road relative to its region.
The issue is “major population center” is subject to how dense or sparse the region is, and my point is that we should look at it from the perspective that in a sparsely-populated region, the biggest, most prominent population center you’re going to get is Ely, and the most bustling highway you’ll see is Route 50. In a massive, empty area like where Ely is located, it is the main hub in regards to population, commerce, and transportation. If it were a suburb of Vegas or Reno, it would without a question be a town, or even a village, linked only by primary or secondary roads—but this instead the primary community in an area that’s completely isolated from the rest of the state, only accessible via remote two-lane highways and small regional airports.
For example, if US 93 were removed, traffic between the major population centers of Las Vegas and Boise would require a multiple-hours detour through US 95. Therefore, despite how remote and relatively quiet this road is, it is a ‘trunk’ because it serves an important inter-regional connection.
I 15 is an hour-longer detour to the east, and it’s arguably a safer route given that it’s a divided, free-flowing highway. Boise and Vegas would get by just fine without that route. Ely (from the perspective that it is a major population center), on the other hand, would need to turn its 3+ hour commute to Vegas into over 5 hours. That’s a huge impact for a region with really only one major north-south corridor.
I understand your point, and I think everybody agrees that smaller population centers are going to be more “important” in low-population-density areas. But we don’t have a clear enough criteria yet as to where the cutoff should be.
I bring up Burns because it is in a very similar boat - a remote town of only a couple thousand, well over 100 miles from any other population center, that mostly owes its existence to being the only pit stop for hours along the main route connecting SE/central Oregon to the Snake River Plain and beyond. There seems to be agreement here that to call Burns a ‘trunk’ destination would be silly, otherwise we would probably bump 395 up to ‘trunk’. Why Ely, but not Burns?
(As it happens, Burns has its own urban area too, so my proposal would also classify it as place=town, just like Ely.)
I think what we’re dancing around is that the midpoints are insignificant, it’s the endpoints that matter. Twin Falls is large, Las Vegas is huge. Ely and Burns are so insignificant that if they weren’t on a highway junction, they wouldn’t continue to exist.
I have implied that I would consider Alturas and Burns trunk destinations as well, by establishing a trunk route between them. They’re the largest cities in extremely remote areas, and have the significance in their respective areas that any city would in their own regions, they’re just smaller than what most people would be inclined to call a “city”. But to the surrounding areas, they provide the same valuable services and amenities a “real” city would and are disproportionately more populous than other communities in those corners of the states.
The OSM Wiki itself states that the trunk network, in a low population density scenario, should consist of the “most important non-motorway roads that provide principal, long-haul connections between population centers of regional importance”. Well, the most important road of this massive, low-density area is Route 395—there is no other way to get between these two points without taking an outrageously longer detour. The most important population center in southeastern Oregon, for example—other than Ontario, which is all the way over at the Idaho border—would be Burns. Unless I’m somehow misinterpreting “In areas of medium to low population density, to the most important non-motorway roads that provide principal, long-haul connections between population centers of regional importance”—I don’t know how else to provide a justification without rephrasing my “area is empty, except this one town” argument again. I don’t think there needs to be a specific “cutoff”, because both highway and place classification almost always relies on a case-by-case if you don’t want to aimlessly classify everything based on data that is trivial to navigation like a road’s number or a place’s incorporation status.
We’re talking about places being so remote that it makes them more important to their respective areas than suburbs like Sparks or North Las Vegas are to their own. Considering there’s only so much space in the US to have areas this remote, and that these areas completely vary in how remote they are, there isn’t really any fair way to have a narrowed criteria for what makes them worthy of being “trunk destinations”.
This is overstated. If US 93 / NV 318 were removed, traffic from Boise to Vegas would take Interstates 84 and 15. The additional time is one hour on a 9-10 hour drive. In other words, only 10% longer, but with the advantage that you’re going through Salt Lake City - an actual “major population center” by any definition. I wonder how many people driving that way take the route through civilization vice the shortcut through the desert just to avoid getting stuck in the middle of nowhere.
If they grew up in a small town/rural area they definitely take the shortcut, I personally highly dislike driving through cities if I don’t have to.
Fundamentally, both highway and place classification establish an information hierarchy. These are editorial decisions, as subjective as this country is diverse. Even the choice to base a classification system on a quantifiable metric would show a certain bias.
Our primary classification tags should be designed for a general audience, for the majority of maps made out of OSM data in their default state. These data consumers will need to clearly communicate the region’s remoteness to end users. A map can still label Ely, based on place=town, to indicate that the region isn’t pure wilderness and invite the user to zoom in to learn more. We’re currently making these decisions by the taste testing method, but more scientific approaches are possible, like Apple’s, seemingly based on AADT counts.
Using other keys, a more specialized map that focuses on numbered routes, truck routes, or scenic drives is free to make different editorial decisions, and sophisticated applications could freely switch between them based on user preferences such as “Avoid Highways”. We should try to facilitate these use cases too. But if not for a moderate bias toward city slickers, there would be little point to classifying rural roads at all.
How to address situations with long routes in wilderness with few cellular communication access, few services and repair stations, and few lodging where desertic or winter conditions make it even more dangerous to travel by car ? Reclassification to highway=trunk will make routing tools to redirect more people using these roads. Using highway=trunk seems more to play with the renderer. But this can create some dangerous situations. Should we create a highway=trunk_w for wilderness to make a distinction with other major roads and better inform on the precarity of travelling on these roads ? Or should we add more tags to describe these ?
In my opinion, that’s completely the router’s responsibility whether it chooses to route traffic between, for example, Grand Junction and Sacramento along either I 80 or Route 50. OSM simply supplies the basemap and the object data that third-party services can interpret however they choose to. We have tags such as speed limits, lane counts, access control/expressway, HGV access/max weight, and so on that routers can contextualize to determine the best possible route between two places, and amenity/business tags to show what services are available along a route. Not to mention “wilderness” is subjective to different people and regions and creating an entirely new highway tag out of something like that would make trunk classification more convoluted and contentious than it really is at the moment.
I’d prioritize routing over rendering in mapping, but I’m not going to blame OpenStreetMap if a router wants to send me down a rural two-lane highway over a freeway or an alleyway over a street.
An unusual, avoidable risk should disqualify a road from highway=trunk. We can further explain and emphasize that risk using objective tags, such as hazard=* based on warning signs and by not mapping the amenities that don’t exist. But that leaves a lot unsaid, and data consumers aren’t going to perform statistical analysis just to determine that a road is actually only a good idea on paper.
Subjectivity is not fatal to a tagging scheme. If highway classification is subjective, that means it’s the most effective tool for communicating something holistically.
I’ve recently put together a (not very complex) Ruby script to promote small places in remote areas, so that rendering looks less sparse. Here’s osm-carto in north-west France:
And here’s cycle.travel for the same area:
It’s mostly population-based and only takes a few seconds to run on a typical osm2pgsql PostGIS database. Once I’ve tidied up the code I’ll post it somewhere.
Well, Ely isn’t served by any Interstates whatsoever. The nearest Interstates—I-15, I-70, and I-80—are all too far away. It’s not even being considered for Future I-11, so using Interstates alone is not enough.
Actually, on an unrelated note, I’m thinking of adding classification tags for motorways. I’m considering adding a proposal to the wiki, with tags used similarly to highway=proposed and highway=construction.
For example, in my new proposal, the entirety of I-215 in Las Vegas, according to Joseph RP, should be classified as highway=motorway and motorway=trunk. Meanwhile, a local freeway (like Summerlin Parkway, AKA SR 613) would be tagged as highway=motorway and motorway=primary. This can be useful to users and data consumers as they can see those tags and not have to upgrade it to highway=trunk, because having a highway=trunk located between a highway=primary; expressway=yes and motorway=primary; highway=motorway would be awkward.
