The highway= tag was codified in March 2006. It has remained largely unchanged since then - a few extra values, some more successful than others, a few refinement tags. But it’s essentially the same.
OSM is used by millions of people around the world, and the highway= tag is the most commonly consumed tag. For 95% of these people it works absolutely fine.
Trying to replace it with motorway=yes or service= or importance= or whatever else is suggested in the latest bout of threads is not, ever, going to be practical. It doesn’t matter if it’s objectively a better idea than the current tagging (spoiler: you are not uniquely clever and it almost certainly isn’t). Rewriting every single data consumer to use your new tagging scheme can no more happen than changing the national language of the USA from English to Esperanto. So discussing it is a waste of time.
I used to joke that the tagging@ list was OSM’s equivalent of the Golgafrinchan B Ark - a containment zone for people who didn’t really think (the thinkers were in the A Ark) or do anything useful (the doers were in the C Ark). Now that the mailing lists have largely died, that discussion has migrated over here. Unfortunately, Discourse’s design, and the seductive allure of the “Latest” tab, means that OSM’s A Ark and C Ark folks are getting sucked into these unproductive, frustrating discussions. This doesn’t make the discussions any better, it just means that they now waste even more people’s time.
OSM works remarkably well. Please let’s stop trying to reinvent something that works. Please let’s stop responding to these topics - just popcorn them and move on. There are lots of exciting things we can do with OSM, and these energy-sapping discussions stop us doing them.
Nope, it is vastly different. Even if we ignore the path-related values, notice how the initial values are all importance-based… with the exception of motorway. But per this approved proposal, all values were meant to be dependent on importance!
The fact there even had to be a proposal means people were using the values to mean other things than importance meaning the tag wasn’t unambiguous as you might have initially thought.
What’s even funnier is that the proposal didn’t even change anything and today each country has its own interpretation that’s not necessarily related to importance.
Now back to explaining why it’s vastly different. Each of the next introduced values (with the exception of tertiary), did not depend on importance which makes the tag really diverge in interpretation. There’s now so many values that over half of them doesn’t depend on importance.
This is the first thing that’s really awkward for the tag and — this issue hasn’t happened with any other tag.
Maybe it does work for them, maybe it doesn’t. Really, maybe it does work fine in your country, alright. But when you go and see what happens in different countries, you can see it’s so much different. OSM is not a project where each country does its own thing but instead, we want a coherent definition to use across the globe but right now that’s not happening. Data consumers do not know what a highway means and can’t interpret it in a way it’s the same in all countries.
Stuff like this is just unacceptable.
Yeah, I guess it does but something to agree on is that the highway=* tagging scheme is the worst scheme in all of OSM and doesn’t work great. An amazing example is the 1.1k post long discussion on the Polish forum about the usage of highway=trunk. Additionally, every couple of months there comes up a suggestion to raise the highway=* value of some road.
All in all, I’m flattered to see an alter-ego evil version of my thread but what you’re doing here is not helping to fix the issue which clearly does exist, even through your own way.
For what it’s worth, I proposed extending expressway=* which would alleviate much of the concern being discussed while allowing the existing highway=* to remain as-is. I would appreciate any feedback on the discussion page.
Just say you’re out of arguments to support importance=* and hush.
Or even better: just apply the Any Tags You Like principle and add it on roads yourself. (all OSM tags are freeform)
BUT DON’T EXPECT ANY DATA CONSUMER TO MAKE USE OF IT ANYTIME SOON. Unless the developer is enough of a masochist to comb thru the initial 200-message thread that contains almost nothing but yapping.
The existing highway tagging works most of the time for most uses in most areas. As evidence for this, people are doing useful stuff with OSM road data around the world.
Any.change proposed on the basis that the status quo is unusable will be rejected because that is clearly not true. It isn’t perfect but it generally works.
THAT is a problem with the Polish community being unable to work together to form consensus. You all spend more time voting than actually attempting to have constructive conversations. I’ve pointed out several times that you aren’t even able to agree on what problem you’re trying to solve. Just to be clear, that’s a Poland community problem, and not an OpenStreetMap problem.
Please read the room. This looks like “I couldn’t solve our mapping problems at home, so I’m coming to the global community to override this.”
We often see differences of opinion about “what OSM is”.
Some people see “a map” (such as on osm.org), and they sometimes complain that “OSM doesn’t show (something) properly” when (something) is already very well mapped in OSM.
Some people see a data project that they can contribute their expertise to, either locally or remotely, improving the data from which maps and more can be made.
Some people see it as a great source of data to make maps and other things from, even if they don’t contribute themselves.
Some people see it as a descriptive project, where how we describe things matters - and how this can be used to address historic biases (e.g. road descriptions that are car centric, or fairer information recorded about the global south).
Expanding from (4) some people see it as some sort of ontology, where how we describe things is important, and also see it as a people-management project, where because it’s “important” to use the right words you necessarily need to tell other people to use them too.
The problem is that (5) - as a world-building exercise, can be really fun, but it has genuine downsides. Changing how humans operate is difficult (and OSM has a lot of humans). Changing how software operates is easier, but still a lot of work that requires lots of planning.
As long as you can express an idea, which tags you use is irrelevant:
is entirely correct.
Sometimes also people interested in (5) might have less experience of running the large real-world projects (outside and inside OSM) than those who’ve been with OSM for longer, and may also have done less of (2) and (3) above. Chesterton’s Fence is often mentioned here - sometimes people proposing (5) just need to get a bit more experience by doing stuff like (2) and (3) before advocating for significant change.
I’m not forcing anybody to the discussion. You may not engage in a discussion if you want. Besides, there aren’t that many people specialising in highway classifications.
That’s unrelated. The key thing is that both sides have shown arguments for and against and the conclusion was that we couldn’t agree on a singular version. Getting into the root of the problem one can see that highway=trunk is ambiguous which caused the whole debate.
What’s this disagreement caused by? Most other situations like this exist because of equating national classification with OSM highway classification.
actually, not really. The proposal has to be seen in the context, it is not perfectly written to be exhaustive with exceptions (it doesn’t name any iirc). It assumed people were in the topic (lots of discussions at the time) and knew what was voted upon. At the boundaries, “importance” is not the decisive criterion for highway, and this wasn’t in question, legal designation is decisive for several highway classes (no byway, busway, links, ladder, via_ferrata here, for “clarity”):
footway
cycleway (ok, at the time, and maybe still in some areas, some people insisted a cycleway is a way where you can cycle, without designation or sign implications).
bridleway
motorway
pedestrian
living_street
There are also other criteria applied for tracks (for trunk things depend on the country)
path (is a mix, either legal criterion if signposted, or physical, i.e. it is too narrow for a car)
track (functional criterion vs. roads and physical criterion vs. path (width))
service (functional)
steps (physical)
so basically the importance hierarchy is
(service)
residential
unclassified
tertiary
secondary
primary
(trunk)
if you wanted, you could put motorway at the bottom and track at the top, because that’s their assumed “road grid importance for motorized traffic”, but it is not how we decide the tagging for them when we are not sure. Estimating the “importance” is not an algorithm that will lead to a perfect 1:1 alignment of all roads internationally or within an area. Rather, it serves as a guidance for local communities to negotiate the details of individual streets.
Hmm. I don’t doubt it was clear to you as the author but it wasn’t clear to me as the reader. It seems that by wrapping the values in parentheses you are suggesting there is something a little bit different about trunk and service in the importance hierarchy, but it isn’t clear to me what exactly that difference is meant to be.
And it can occur as well after redoing highway= schema using your idea or idea coming from someone else. Root cause here is differing interpretation of tagging schema in sub-communities, and it can occur with any tagging schema.
And while undesirable it is hardly a critical issue anyway.
not really, as it occurs over a national border between Belgium and Luxembourg.
Hard to blame silly attempt to replace systematic highway=trunk mistagging in Poland by another systematic highway=trunk mistagging in Poland for this one.
(at the risk of dragging this thread along even more) I’ve seen exactly that situation in the real world (in the 1980s or 1990s between Belgium and, I think, France) where one side had built a road to a standard that was “important” based on any metric that you could imagine but the other didn’t consider it so and hadn’t).
Some people want to see an “elegant” structure of highways in OSM but the real world, however you choose to measure it, sometimes disagrees. Perhaps those people need to experience more of the real world.
To me it’s easily a trunk as an extension of a motorway which leads from one of the biggest cities in Canada to biggest cities of some states like Portland and Boston. I don’t know why would Canadians think the road isn’t important enough.
Don’t you think it’s just silly to have an importance hierarchy values and functional values in one key? A better option is to have a functional hierarchy at the highway=* tag and if you think that importance is not an algorithm, then it might not be a required tag. Maybe we should abolish importance altogether. Somehow Google manages, doesn’t it?
No, because this one is caused by OSM classification == national classification and my concept separates the two.
Well importance depends on the de facto situation while the standard would also be tagged as it is, unrelated to importance.
