According to a 1984 PA statute, the portion of interstate 83 in PA is designated “Veterans of Foreign Wars of the United States Memorial Highway”. The problem is, this incredibly unwieldy name is used rarely to never in signposting along the corridor. On ramps use just the “I-83” shield designation. There is a sign with the official name when crossing from Maryland into PA, but otherwise the official name is not useful for routing or wayfinding. I am certainly not proposing removing the name entirely, but wanted to check if there was an alternative way of tagging it. I found the key official_name=* but validation in JOSM complains if there isn’t also a plain name tag. My searches in overpass for the official_name tag were also not very conclusive, so I would love to hear thoughts.
Ignore the JOSM warning. That’s a clearcut example of a road having an official name but no common name.
(You might be able to silence the warning by adding the noname=yes tag, but that’s likely to trigger validation warnings in other tools.)
Maybe add also note= explaining situation and linking this thread?
To reduce risk of someone else trusting QA tools too much?
(maybe this rule should be removed from JOSM validator? it is not so rare)
Can you use the noname tag in this case?
Setting official_name=* but not name=* asserts that the highway has an official name but no common name. Or, more precisely, no common name that isn’t a spelled-out form of the route number.
For completeness, another common approach is to set name=* to a systematic name, “Interstate 83”. This is controversial because, in principle, a data consumer should be able to reconstruct the name from the route relation’s network=* and ref=* tags or by looking up the label of the Wikidata ítem linked from wikidata=*, although nothing can do that currently.
At the very least, if name=* is set to this memorial name designation, then name:signed=no would tell data consumers not to mislead users into thinking they can find their way around using this information.
I’d call that name=* if no other name applies.
official_name=* without name=* and any name with noname=yes would both be errors.
I was able to wrangle a better query in overpass and located some existing examples of official_name on similar memorial highways, however, the results were mixed, especially on whether or not to use noname=yes when the name tag is not present on these ways:
| ref | noname | name | official_name |
|---|---|---|---|
| I 69 | yes | Iraq Afghanistan Veterans Memorial Hwy | |
| 401 | Highway 401 | Macdonald-Cartier Freeway (Canada) | |
| I 90 | New York State Thruway | Governor Thomas E. Dewey Thruway | |
| I 89 | Vietnam Veterans Memorial Highway | ||
| US 15 | Marine Corps League Memorial Highway | ||
| I 395 | yes | Governor John Davis Lodge Turnpike |
I am planning to move forward with setting official_name and noname on I-83. It seems to match up best with the wiki and the “on the ground” principle.
noname=yescan be used to mark the absence of a name i.e. where something really does not have a name in reality.
If an official name is more unwieldy or obscure than another name for the same feature and fewer people use it in practice, even if it is signposted, it is better to put the official name in
official_name=*oralt_name=*than to treat it as the primary name.
not really: something may have an assigned an official name without having an actual common name at all
Then that would be the common name, definitionally and by default.
Oh that name value would be straight up wrong. Refs are not names.
I’ve attempted to quantify this approach via QLever. This query finds all highway ways of tertiary and above with a ref and also a name like Highway 234, State Route 45, Strada Provinciale 12, Ruta Nacional 23, etc. Spot checking the results, there are some cases where the number in the name doesn’t match that of the ref (name=Old Route 17, ref=CR 179 for example), but most appear to match. The practice appears to be fairly widespread across North and South America with some across the rest of the world as well.
Here’s a slight variation on your query that finds highways whose name=* contains the ref=*:
This query finds highways whose name=* contains the ref=* of a parent road=route relation:
In the U.S., these statistics are skewed by dataloss from the haphazard removal of numbered name=* tags in some states many years ago. It seemed like a good idea at the time, but it turned out to be misguided. Unfortunately, it may no longer be feasible to bulk recover these names, which very often included important information like quadrant directionals, because TIGER tags have since been stripped from these ways as well. In some counties, we may be able to guess based on surrounding unnumbered streets with directionals, but elsewhere, we would need to run a fresh conflation with TIGER or other government datasets to correct these road names. In some counties, we might as well do that due to systematic renames driven by E911 needs.
At least we can all agree that names are not refs. Someone got the wrong idea about that when mapping the Arizona Parkways, Historic and Scenic Roads program, resulting in 10-mile-long route shields. 
Why? There’s no reason to include route number information in names on ways, and plenty of established precedent for not doing that. The US is not special.
Personally, I’m of two minds about systematic road names in general. From a data consumer standpoint, the explicit numbered name tags are either a godsend or a nuisance, depending on the particular situation, whereas memorial name designations are always a nuisance.
The U.S. kind of is special. In the real world, route numbers in highway names are a distinctly North American practice that predominates in only a few other parts of the world. Elsewhere, even freeways have ad hoc names much more frequently than here. Conversely, more of our city streets have ad hoc names than in some countries in Latin America or the Baltics.
Regardless, if a street is named “[Directional] [Route Number]” or “[Route Number] [Directional]”, deleting name=* loses the directional. A motorist may not need to distinguish between Northeast State Route 10 and Southeast State Route 10, because route marker assemblies are going to be more visible and include details such as route directions and destinations. By contrast, a pedestrian may find the quadrant directional more helpful, as the only signs posted for their convenience are going to be blade-style street name signs and address signs. This is of course a gross generalization, but the point is that we shouldn’t have gone around haphazardly deleting information before we had a handle on all the different cases where the information does matter. Now we’re left to recover what we can.
As to precedent, we were originally consistent about including the systematic name in all cases (to the extent that TIGER is consistent about anything). It took an active campaign by a lone mapper to move us toward relying more on structured tagging, but they got banned midway through their campaign and it never really resolved the issue.
The street would not have a name, and addr:street=* is a functional fiction based on the lack of name. Validators would hate it, but that would be the reality of the situation.
That doesn’t address the concern about dataloss, about losing the ability to record the reality of the situation.
The history is still in the database, there is no data loss.