Talk:Named spots instead of street names

From OpenStreetMap Wiki
Jump to: navigation, search

Why not use the existing reference_point=yes

This proposal sounds very similar to the existing reference_point=* which was approved through voting less than a year ago. Since the tag junction=* already exists to indicate the type of junction (junction=roundabout etc.) it sounds counter-intuitive to me that junction=yes is supposed to indicate that its name is of importance. I would prefer adding a reference_point=yes to junctions whose name is used for address/orientation purposes. EAman (talk) 13:30, 24 August 2014 (UTC)

Hm, I think there is indeed a difference: In Japan, you have official names for traffic signals. The description of reference_point is in the end that you can add it to virtually everything that you consider arbitrary worth for orientation. (But buildings or trees are yet rendered, even without this tag. And for a map of e. g. Korea, you would expect that the crossroad names show up in a special style – and no other reference points. So you would need both junction=yes and reference_point=yes, and the renderer will probably interpretat junction=yes and ignore reference_point=yes) However, on this page I’ve only documented what is yet in widespread usage: We have 19 269 elements with the combination highway=traffic_signals and name=* in the database (mostly in Japan). We have 3920 nodes with the combination junction=yes and name=* in the database (mostly in Korea). But we have only 95 elements with the key reference_point in the database. I agree with you that junction=yes is not too intuitive, but is that enough to change a tagging that is widely accepted in the local community in Korea – towards something that doesn’t give you any real advantage? sommerluk 14:27, 24 August 2014 (UTC)
I actually believe that the best long-term strategy for OpenStreetMap would be to try to find the "best possible" tagging scheme and then either manually or automatically change the tags that have been entered before the tagging scheme had been properly debated. I know that many in the community may disagree with that opinion and let's not start a philosophical discussion here.
I'm still afraid that the tag as proposed will be misunderstood to the point where it will be difficult to know what the mapper intended. Where I live (and I guess also in most of Europe and North America) junctions (particularly large ones) may have an official name which may even appear on a map. The name is however mostly relevant to road traffic as such. A mapper who sees the possibility to add junction=yes to a node will most likely do that to mean "yes, this is a junction" just like bridge=yes means "yes, this is a bridge".
Obviously we already have the ability to add a name=* tag to a junction or traffic light and a renderer can be configured to display all names for junctions and traffic lights. Your proposal indicates that this isn't enough, but there is a need for another tag to indicate that something isn't just a junction with a name, but the name of the junction actually matters to people even more than street names in finding their way through a city. How much does this differ from the Nicaraguan or Costa Rican addressing systems where certain points (not just junctions) are important for navigating and may even be part of the postal address? Even if the description of reference_point=* is that it can be added to almost anything, the intention is obviously that it should be used only for points already in use by the public for navigation. EAman (talk) 16:01, 24 August 2014 (UTC)
Some thoughts about adding (additionally) reference_point=yes to road junctions: Would we also have to add reference_point=yes to various millions of highway=* roads? Probably most people will answer “no” – me too. But with which reason? Because highway=* is expected to be meaningful for orientation and everyone expects that all streets get rendered with their name (at least on a normal street map). The same way, the names of places or of highway=motorway_juncion are rendered also without reference_point=yes. We could say that the tag highway=* implies that this name should also rendered because it is important for orientation. We could say highway=* implies reference_point=yes. The same is IMHO true for junction names. In regions where junction names are not important, they will not get names. In regions where junction names are important, they get names. So I would say that the same way like for highway=*, also for junction=yes it is not necessary to add reference_point=yes because this would be redundant.
Some thoughts about the usage of reference_point=yes: This is different. In Japan you have official names for traffic signal systems. In Managua (Nicaragua), following the description at the pages that you have linked, you do not have any really operative address system – neither official nor inofficial. In the abscence of an address system, people have created a workaround, using instructions for getting somewhere, starting from well-known existing (or previouly existing) landmarks or reference points. What sort of rendering would people expect? Probably, a building has still to be rendered as a building, a monument has still to be rendered as a monument, and a hospital still as a hospital, and a shopping mall still as a shopping mall, and a water tower still as a water tower. But probably the rendering of the name should be bigger – more prominent – than normal buildings, monuments… to recognize this on the map. The reason for the lack of support for this is maybe that OSM ist still quite europe-centric, and in other parts of the world, the things may be different. Here in Ivory Coast we have some junction names, but in smaller towns we deal often also with this type of “instructions for getting somewhere” instead of an address system in the european sense. (Unfourtunally – opposed to Manuaga – in Ivory Coast, everybody has a different opinious of what is a well-known reference point and what not.)
Some thougts about the relationship between junction and traffic signal names and reference_point: Maybe we could describe the relationship like this: There are regions in the world where people orient themselves with names spots. There are three possibilities what these spots are: They can be systematically (and officially) named traffic signals (like in Japan) or junctions (like in Korea) or they can be (unofficial, but de facto widly used) reference points/landmarks. --sommerluk 13:40, 25 August 2014 (UTC)
It's hard to argue with some of that. Of course we don't want reference_point on everything either. Let's try to see what we actually agree on:
* Almost anything in the OSM database can have a name=*.
* The tag junction=yes might exist.
Where we seem to disagree is what junction=yes means. You seem to want to use it where the junction names are more important than street names. I'd say it can be used for a junction in general. Therefore I think the wording currently on the page is misleading. Wouldn't it be enough to add a statement on the general page about the tag junction=* that it may be combined with name=*. I could also imagine that is's a good idea to have a page for Korean/Japanese mappers describing how they should map the names of junctions and traffic signals, but we don't actually need to change the meaning of any existing tag to do that. It's of course then a very good idea to configure a rendering engine to display those attributes prominently on maps of Korea, Japan, etc. You have documented a lot of highly relevant things. What I didn't like is attempting to link junction=yes with the fact of it having a name or that the name is important. EAman (talk) 18:24, 25 August 2014 (UTC)
Okay. Considering that that wiki documentation of junction=yes exists (at Key:junction) since 2010 and that it has always been described as for usage together with name, I don’t want to go behind this. But I’ll try to find a wording that makes clear that junction=yes does not make any assumtion on if the junction name is more important than street names or not. I’ll try also to mention reference_point=yes. Give me some hours of time… --sommerluk 19:40, 25 August 2014 (UTC)

What's new?

The combination junction=yes + name=* has been documented on the junction=* page since 2010. The key name=* is listed as useful combination on the highway=traffic_signals since 2008. No need to make a proposal for tag combinations that have been in use and documented for years!

Your proposal is essentially about a minor change in documentation. If you start such a proposal, you should not copy the proposed text to the respective features pages immediately, as you did! First let there be a voting, then do the change, not the other way round.

At the first spot your changes look like an extension, but what you did is actually restricting the usage to certain countries for no reason. Junction names are not only used in countries like Korea, but all over the world. E.g. there are many junction names in my country (Austria). So please undo your edits on the junction=* and highway=traffic_signals pages. --Fkv (talk) 18:47, 24 August 2014 (UTC)

Well, overpass does not find more than 4 times a junction=yes in Austria ;-)
My intention was to do two things that are independent:
The first was to document the current tagging practice. This is what I’ve tried with junction=* and highway=traffic_signals. This does not pretend to be something new at all. It’s not a proposal, but just a documentation of what is already going on in the database. I don’t have copied the proposal text just to the wiki – at the opposite I’ve first tried to document the current tagging practice and after that made a proposal to extend the current tagging practice. Concerning the documentation of the current tagging practice: It was not at all my intention to forbid the usage of junction=yes in other countries like Korea. Indeen, I’m living in Ivory Coast and here we have also junction names and of course I think that it is okay to use it. I’ve put Korea as prominent because – following overpass API – about 80% of all occurences of the combination junction=yes and name=* is in Korea, and only 20% in all the rest of the world. Obviously my text creates the impression that a usage on other countries is discouraged. This was not intented and I will fix this. Same thing for traffic signals/Japan.
The second was independent of the documentation of existing practice: It was to create a proposal that extends the current tagging practice to have a clear solution not just for simple junctions, but also for complex junctions. This is what I’ve done at Proposed features/Tagging for complex junctions or traffic signals that are named. My intention is to have a concensous which is also widely accepted in the two most affected countries (Japan and Korea). While I have also a personal opinion how to tag complex junctions at best, I don’t have made a simple proposal of what I think, but I’ve collected different proposals (5 in total) also of other people to discuss this with the community to choose this one which is most widely accepted. All of them have advantages and disadvantages. I agree that the tagging for complex junctions will not (and should not) be a very big change, but nevertheless it is worth to be discussed by the community. --sommerluk 20:47, 24 August 2014 (UTC)
I checked some facts using the Overpass API. It's true that most nodes with a combination of junction=yes and name=* are in South Korea (2581 out of 3921 = 66 % in my search). However, when you start looking closely at those results, there is an interesting pattern. Most of them were created by user sanha two years ago. They are obviously imported automatically, because they are generally not connected to any ways and in many cases the OSM data does not indicate an intersection or even a road at that location. This Overpass API query shows the 1781 nodes (out of a total of 2581 in Korea) last edited by user sanha. There are a few more that have since been touched by other users, but I don't know how to search for those. Note how the junction nodes are suspiciously located in one square of South Korea. There is one changeset with 1841 nodes. Thus, this tagging scheme doesn't seem to be particularly well established even in the Korean community. EAman (talk) 22:37, 24 August 2014 (UTC)
The fact that there has been an import does not automatically mean that the tagging scheme is not accepted by the local community. --sommerluk 13:40, 25 August 2014 (UTC)
True, but this scheme still doesn't seem to be well accepted by the average Korean mapper, but rather be an effect of a fluke in some import script: I noticed that out of those 2581 named junctions i Korea all but 13 nodes have the tag source=mp2osm. They are thus automatically imported using some script. If the tagging scheme were well established, shouldn't we see a few more manually entered junction names spread throughout Korea according to this scheme? Out of all the nodes in the world with junction=yes and a name the ones located in Korea constitute just 1 % if you count the ones not created using mp2osm. EAman (talk) 17:54, 25 August 2014 (UTC)
I'm new to OSM (I joined wiki today and saw this on talk-ko mailing list), but if I can make some opinion, it is common scheme to use Junction name, instead of Street name - even some local government issues a separate name for each junctions and other maps (google maps, local internet map services, paper maps) uses this informations. — revi^ 16:56, 12 September 2014 (UTC)