Hence “considered” – by the local OSM community. If I had my druthers, I would probably do it differently.[1] After all, the Highway Code and the Traffic Signs Manual do refer to these numbers as “route numbers”. So an unnumbered primary route can traverse multiple route numbers that individually refer to A roads and B roads. Not confusing at all.
(Find a meaningful signposted route concurrency on Great Britain and then the party can start.)
But I don’t even know what druthers are or if they’re any good to have. ↩︎
It’s great to have a broad discussion on the topic, and I truly appreciate the input from everyone. However, since I do need to achieve a bit of what I was aiming for, I’d like to explain a little further why I’m focusing on the physical characteristics of “unclassified roads.”
As I’ve mentioned before, road systems differ from country to country, and each has its own unique structure, which can cause some confusion when trying to apply the road system defined by OSM (OpenStreetMap) to different countries or regions.
To align the OSM road system with real-world systems in different countries, a certain level of understanding and organization is necessary. To do this, it’s crucial to understand the concept of “unclassified roads” and how they differ from other road categories. This will allow regions to either choose not to use them or adapt them to fit their national road systems. (Honestly, if the concept is unclear and leads to confusion, I think we should consider not using it at all.)
As for the characteristics of “unclassified roads,” I understand them to be part of the administrative road system but are considered somewhat “insignificant” and don’t fall into other categories of roads.
From the discussions I’ve had, it seems like the characteristics and condition of “unclassified roads” vary across different countries and regions. However, I still think it’s important to try to identify some common traits and physical features that we can agree upon.
Ah, that’s a good approach. Since we are trying to identify common physical characteristics, it might be helpful to draw some inferences from actual cases.
I’d like to clarify that we’re focusing on physical features here, and this discussion is not related to the administrative attributes or classifications of the road itself.
In the case of images 1 and 2, while the photos might seem a bit ambiguous, I believe that in some areas, roads that pass a few independent buildings or facilities (instead of a village center) could definitely be considered as an ‘unclassified road.’
As for images 4 and 5, I believe they clearly match the characteristics of an ‘unclassified road.’ Specifically, image 4 aligns perfectly with the features I’ve identified as characteristic of an ‘unclassified road.’
In the case of image 3, it might initially seem hard to distinguish it from a ‘track,’ but regardless of whether the road is paved or not, if we examine the actual condition of the road on-site or on a map, there’s enough reason to classify it as an ‘unclassified road’ rather than a ‘track.’
Overall, considering the surrounding environment and related context, I believe these examples faithfully reflect the characteristics of an ‘unclassified road’.
Additionally, in some areas, the condition of being ‘wide enough for one or two cars to pass,’ meaning a single lane or a two-way lane (in which case there would be a lane marking in the center of the road), could be a relevant criterion. However, in other areas, this is not a valid criterion at all, which is why I believe it should not be considered as a factor.
The main problem is that there aren’t exactly any visual characteristics for highway=unclassified. The above images could very much be primary, secondary, tertiary or even trunk instead of unclassified since the classifications have to be looked at a grander scale (not to mention that built-up areas often retain the old street width).
The only thing the physical properties can affect the classification is if they’d be classified higher or lower than they officially do e.g. a well built tertiary road (e.g. two lanes throughout, few stops) might be better tagged as highway=secondary than highway=tertiary while a narrow tertiary road (one where two cars can’t physically fit even in the open field) might be better of as highway=unclassified.
Can we please not get sidetracked into this discussion for the 500th time. Or at least you can sort out your county roads in the US (current relation status: approx. 0.1% done) before telling us how to do it in the UK. Thank you.
I think you’ll need to face the fact that the ‘unclassified’ literally is an administrative term.
However, being pragmatic: if you find a small rural road, not very wide, no road markings, and isn’t listed on road signs and (ahem) ‘other maps’ as having an official number, then the best tag to use for it on OSM will be ‘unclassified’.
That should work well for all countries where there is a some kind of statutory body which classifies and numbers its road network.
The main reason for a ‘false positive’ when using the ‘unclassified’ tag in this way would be where that country either (a) isn’t very good or consistent about signing its classified roads, or (b) it has a class of roads which is one level above ‘unclassified’, but whose number doesn’t get listed on maps and signs.
An excellent example of that is the UK, where UK ‘tertiary’ (C-class) roads are often visually indistinguishable from UK unclassified roads. However, this has not proved a problem on OSM, because in the UK, the data on which roads are ‘C’ and which are ‘U’ is readily available from statutory sources (to the point where it arguably becomes ‘local knowledge’ for OSM editors…)
The A205 is the ‘South Circular’ road, an ‘inner ring road’ through south London. So it is assigned that number by the highway authority (‘Transport for London’ in this case). But all the individual segments of it it will have street names (assigned by the borough councils).
The ‘A205’ designation is used on road signs to help motorists follow the route.
The street names are used to form the address for properties, for postal deliveries, legal things, etc.
Pretty much all roads in the UK follow this same logic.
I can recommend the Making sure you're not a bot! site as a treasure trove of geeky information about UK roads, including how they are classified.
A road is best defined by who uses it. A road is for example not residential because of adjacent houses but because it is used by people who want to reach a house. A tertiary road is used by people who want to move from one town to another. And so on. This should be similar for all countries.
yeah, pretty much all UK roads have a name and a number.
And these will typically be implemented as numerous individual ways in the database (potentially hundreds, for a long numbered route - but even a short named road can have numerous ways).
But given that pretty the entire UK road network in OSM is currently done using just ‘name=’ and ‘ref=’ tags for these on individual ways, and it works fine, I’m not sure what problem we’re trying to solve here?
To further confuse you, sometimes scenic tourist driving routes are signposted with brown tourist signs, explicitly named on the sign as a ‘route’. And then there is a whole concept of Making sure you're not a bot! ‘Primary_Destinations’ which can be expressed as destination route relations.
Well, in the case of the A205, it does have a relation, though the name is incorrect (refs are not names, so the name of the relation should be South Circular Road. Fix that and it’s a great example of a well mapped route). I already pointed out that it does, indeed, cross multiple roads, even if you call routes “roads” themselves, and gives A205 its own entity. We should consider phasing out ref=* on ways for consistency since the reason for their existence died 15 years ago and it’s way more difficult to infer routes that way than just using the relation.
Nederland here! The situation is not very clear. We have clear consensus mapping rules for most road levels, but unclassified is not that clear. Some mappers still use unclassified to express that the highway type is unknown; same as highway=road.
In the past, a major import of roads and streets was done, where all residential roads and streets got the tag highway=unclassified. This was not corrected at the time, and now, because of all the intermediate edits, it can not be corrected by mass edit. So many residentials remain tagged as unclassified.
Discussions about which roads to tag as unclassified never lead to a consensus. Here, too, some mappers say we don’t need the level; others use it for roads above residential and below secondary, sporting a heavy overlap with service, track, and tertiary.
Surface and markings do not indicate anything for unclassifieds. They may help, but are not defining criteria.
So, in the end, highway unclassified is still used for unknown highway type, in Nederland. We don’t have an official class “unclassified”, so that doesn’t help, and every single new mapper has to be explained that unclassified is a classification, in the UK.
‘South Circular’ is probably the name of the engineering or transport project that conceived the road (although much of it was never built as intended - YouTube has some good videos about ‘unfinished London’…), and is more a colloquial usage, rather than something designated by the relevant highway authority, so its probably incommensurate to add it to that relation.
To be honest I feel your point is somewhat academic - personally I have about 300 bus route relations to manually create before I fiddle around removing ‘ref=’ tags from Britain’s roads. And relations are always bloody breaking! Do we really want a tangle of them all over our roads?
So your conclusion is that a visual characteristic of some unclassified roads is that they pass a few buildings?
And what about the village centres? What do you suppose those roads should be tagged? In the photo below, those are shops and offices lining the street. The street doesn’t have much if any through traffic and would be a clear candidate for unclassified for me.
How do those characteristics square with the characteristics of images 1 and 2?
The road in image 5 is in a city, BTW: Way: Keefer Road (55774910) | OpenStreetMap. (Please ignore that it’s tagged residential - when the import was done, it didn’t make a distinction between unclassified and residential and no one has been around to correct it yet.)
The tag highway=track is used for minor land-access roads that are not considered part of the general-purpose road network. Track roads are mostly used for agriculture, forestry, outdoor recreation, and similar activities on open land. Roads used for access to permanent human settlements or facilities should generally not use this tag.
Fifteen Road is considered part of the general-purpose road network, it’s used for access to permanent facilities, and it is not mostly used for agriculture, forestry, nor outdoor recreation. (You can find it in OSM here.)
This is backwards. These roads aren’t tagged unclassified in OSM because the images I pasted faithfully reflect the characteristics of an ‘unclassified road’. The roads are tagged unclassified because they’re neither tertiary nor service nor residential nor track.
You wanted to discuss visual characteristics of unclassified roads, but when given photos, you don’t say what characteristics the unclassified roads have in common.
I think that’s because there aren’t any visual characteristics that would encompass all or even most unclassified roads.
Indeed it is - if you want to know the history of the B679, then that is the site for you. Unfortunately, some denizens of SABRE do have a habit of making up road names - for example, based on natural features that predated the road and were mentioned as things that the road might go past in pre-planning documentation.
