In the importance proposal the suggested values are international, national, regional, urban, suburban and local which correspond to the values 1, 2, 4, 8, 9 and 10 respectively.
The station importance proposal includes the values continental, interregional, regional, urban, suburban and local which correspond to the same values as mentioned above.
It’s also used for thecapital=* key so it’s not something used for only one thing.
With the official road classifications you can also make an equivalence. Let’s give an example with Polish road classification: we have national, regional, county and communal roads which correspond to the values 2, 4, 6 and 7 (you could also add 1 for the E-network). The current tagging scheme is highway:category:pl=1/2/3/4 but maybe having it be admin_level values would be better. So I really believe that admin_level can be used to describe the scope of something but I don’t want importance=* to have numerical values since there aren’t enough highway values for that also unless we actually only allow the usage of importance=1/2/4/6/8/9 but that feels pretty weird in my opinion.
It’s not really obvious to me what “corresponds” means. Does it refer to being contained in a single administrative unit versus crossing the borders of administrative units? (I don’t think you are referring to which administrative body manages/designs/operates the roads, but I could be wrong).
E.g. in Spain, the A-92 is a 400km route across Andalucía, a region (autonomous community) comparable in area and population to Portugal or the Czech Republic. It doesn’t go anywhere near any other region (autonomous community) or even really head in the direction of one, as it runs parallel to the main land border of the region. So if admin levels matter I guess it would be a mere regional road.
But La Rioja, with about 5% of the population and area of Andalucía, is also an autonomous community with exactly the same status. Most roads other than really minor ones cross its border because, well, they run out of region pretty quickly.
How does admin level help to classify roads here? Does it mean that roads will generally have higher importance around La Rioja than within Andalucía?
(If the answer is “no, because there are no large population centres around La Rioja so we would not classify any of those roads as high importance”, that would just suggest that admin level is not relevant).
It’s also worth noting that the autonomous communities essentially date from the 1978 constitution. If OSM had been around before 1978, and importance relates to admin levels as suggested, would the insert of a whole new layer at the top of the admin hierarchy (just below the country itself) have led to a major reclassification in OSM highway data?
I think “importance” might not be the best word for the concept that @pavvv is focusing on. That’s what the importance and highway importance proposals called it, but it’s clearly causing some confusion. A road can be important or notable for any number of reasons, but what’s missing when we overfocus on route networks and road types is how well a hyperlocal stretch of roadway fits into a larger system. That’s what maps need at lower zoom levels.
There’s much more to functional classification than simply counting the number of connections or the length between intersections. (And functional classification affects cyclists too!) Otherwise, in theory, a renderer could integrate a routing graph to determine which roads provide the most efficient access between the largest cities, between parts of a city, and so on. There’s all sorts of reasons why this would yield unexpected results. I could’ve sworn that Google blogged about integrating a routing graph into Google Maps at one point, but at most all they did was to demote dead-end streets by pruning leaves in the routing graph.
In most countries, we accept that highway classification is holistic and subjective at some levels, because most official classifications don’t make nearly as many fine-grained distinctions as we have highway=* values, while some make even more. This doesn’t quite make highway classification unverifiable. The threshold between blue and green is subjective, it varies by culture, and some cultures don’t even distinguish at all, yet colour=* is still verifiable.
One way or another, I would caution against turning highway classification into an empirical science. By and large, we just don’t have the expertise or data to do the science right. Let’s have a key (call it highway=* or importance=* or whatever) that captures a classification that’s good enough, and leave the science to other keys that we can quantify without as much handwaving. For that matter, let’s put any official designation in a separate key explicitly. This will lower the stakes for getting the main classification just right.
Just by referring to the names. Since country = 2, importance=national corresponds to the value of 2.
importance=regional or =4 means of regional significance so probably contained but I just view it as connecting towns above a certain population threshold.
In France the next level below national highways are on the admin_level=6, the departments. Roads where the department is responsible can be anything from trunk to tertiary, but there is very little difference between highway classification in France and Germany at the French-German border. I looked at all cross-border roads (not tracks or pathes) between the tri-border area near Schengen and the Rhine. In just one case the classification differs: tertiary → unclassified between Ober- and Niedergailbach (Way: Rue de l'Europe (373991358) | OpenStreetMap and Way: Obergailbacher Straße (30741429) | OpenStreetMap)
You will find the official classification (in France and Germany) typically in the ref. There’s a really big difference between the forementioned D 34a and the D 620 just around the corner, with a part as highway=trunk with motorroad=yes …
I don’t see how your proposal will get better results as the current tagging.
That’s fine and doesn’t change anything. There just needs to be a name for the values that can be used anywhere and it could definitely be something like importance=county because we already have place=county.
Well, the network will definitely be consistent as per my proposal. Through the border and without random disconnected trunks.
It seems like importance:bicycle could really be used here.
Both roads are departemental roads (official classification), that means admin_level=6, currently “county” and you wrote:
I don’t understand the role of the admin_level just as @alan_gr wrote:
So importance=6 doesn’t really tell us anything. It’s a big difference between neighbouring countries. But highway=primary or highway=tertiary tells us a lot - and it translates better between countries too.
You took that out of context. I was asking if we should use numerical values instead of text in case it’s too difficult to come up with names like the tertiary-equivalent importance because as someone else pointed out, county isn’t a good idea because administrative division differs from country to country.
It’s meant to be the same thing. By making posts like the previous one you’re making more people hate on the proposition because they view it in a way it actually isn’t.
To reiterate, the importance=* tag is the same as highway=trunk through residential but with different values, e.g. importance=national, importance=regional and I’m open to suggestions about the value names for the equivalences of highway=trunk and other tags because the current contender is importance=county that is a problematic name for obvious reasons.
Thank you, again, for a well-argued post! Also, sorry for not repeating this here (as I have done here and here) that I’m of course fine with people inventing Any New Tag They Like for “importance” of cycleways. I do, however, vehemently oppose people using and overriding already established tag-namespace, like highway=*, cycleway=* or network=* for it, as is the case also here.
I reiterate that the fundamental problem I see in tagging the importance (or whatever) of cycleways is that cycleways don’t usually—and unfortunately from the point of view of the avid bicyclists I am—form a dense enough network that would warrant a complex gradation scheme. At least not in a way that does not simply coalesce with their physical features. This is true both in the long distance and hyperlocal cases, though for slightly different reasons.
Again, at least where I’m from, I decide my most efficient local bicycle routes by first excluding ways that have smoothess worse than intermediate and surface other than asphalt. I also severy penalize non-segregated cycleways. Here’s an Overpass estimation of the ways this leaves in my neck of the woods. During winters, I also penalize bicycle lanes[1], so that reduces the map further. Because this leaves a very blotchy map, I simply connect the ways with whatever else I have to use to get to where I’m going to. Choosing two random points A and B, how would you further prune that selection of cycleways based on—for lack of a better word—importance? To me the selection of importance would seem to depend mostly on the points A and B, and thus be subjective. There are some cycleways that run fairly close by to the same immediate destination (parallel to city blocks). So one of them could, perhaps, be deemed more important. The question is, which (if the choice is to be nonsubjective).
Because snow is ploughed onto them first before it is transported away, making then unusable for long time after a snowfall. ↩︎
If they are the same thing, why change the values at all? Introduce, say, importance=highest if you think the meaning of trunk has become confused. Leave the rest at Importance=primary and so on. That avoids introducing concepts like regions and counties that might falsely imply a relationship with admin levels. It would also make it very easy find and review places where mappers have decided to diverge from the current scheme, simply by looking for highways where the two tags have different values. If that turns out to be rare, it would suggest there is no need for dual tagging and we could stop doing it.
I agree with this perspective. For my part, I supported the development of the very colorful Cincinnati Bike Map, which relied on OSM to prioritize physical realities over the aspirational routes the city and state governments had designated and signposted (verifiable! deterministic! silly!). If a topologically coherent cycling network is uninteresting or useless, then the concerns about this proposal being car-centric must be a bit overstated.
We’re seeing these contradictory reactions because the proposal is breathtakingly ambitious in scope but raises more questions than it answers. I hope no one took me too seriously that one time we were talking about highway=path and I started digressing:
(So no one will think they’re alone in sometimes longing for things to have worked out differently.)
This is plain false: only 5 ways have both a highway= and an importance= tag.
I just checked with overpass because I never saw anything like this in France, and it is something that has never been discussed within the french community.
Can you list all of the values? Or actually just a replacement for =unclassified—would you suggest ‘quaternary’, as it’s interpreted in Poland, or something else?
It’s an interesting concept but 1. there would have to be very clear definitions about the scope of each value and 2. there are misinterpretations with the current values of highway because in the UK a primary road is a regional road while in Poland a primary road is a national road—this is why when it would come down to voting, many people would oppose changing their secondary to primary calling it confusing.
I was talking about railways, check the paragraph before that one.
The need for clear definitions applies regardless of the values used. I think a lot of the responses come back to this point. You haven’t really offered definitions for the key you are proposing, beyond the definitions that already exist. If those definitions are OK, we don’t need a new tag. If they are not OK, proposing better definitions would need to be a key part of the proposal.
Values such as “regional” and “national” are just as much subject to interpretation as the existing values. E.g. I suspect that many roads officially called Regional in Ireland, you would see as the level below regional in your scheme. Or when you say that UK primary roads are regional, what do you mean exactly? I am no expert on UK tagging but I believe that roads tagged as primary are usually part of the national A road network.
No, most if not all of them would be importance=regional.
I’m pretty sure the UK has it different from the rest of the world. Ireland, Germany, Poland have roads divided into national, regional, etc. while the UK has primary, secondary, etc. which is where the current classification comes from. It seems that it actually is that if we want to base importance off something, it’s probably going to be one of these countries which have their roads classified into national, regional, etc meaning switching from UK’s classification to another country’s but I’ll have to do more research.
A highspeed rail doesn’t automatically imply international importance and vice versa. Usage=* doesn’t have such a value either.
You’re not even pretending to having read my original post.
Interesting; I proposed a solution to remedy the former a few weeks ago (see highway:motorway on the wiki). Why do we have highway=motorway but also expressway=yes?
