Talk:Proposed features/Highway administrative and physical descriptions

From OpenStreetMap Wiki
Jump to: navigation, search

First discussion

There are some things here that don't fit how the system works in Norway:

  1. Motorways are not a designation, but a standard for the physical buildup of the road
  2. E-roads are cross-european, although formally also national highways with ref=E<num>
  3. The physical properties have many distinctions on painted center lines, here roads always have painted center lines if two cars can pass each other.

Bobkare 00:49, 26 January 2008 (UTC)

Motorways are both a physical characteristic of a road, and an official classification, depending on a jurisdiction. Physically, a motorway is a road with two lanes in each direction, generally divided by a barrier or a wide median, and with controlled access (i.e. all access must be done using on and off ramps. Sometimes, they are substandard (e.g. short acceleration and deceleration lanes, sharp turns, narrow or no shoulders) but they are still motorways.

Legally, many countries (or regions of countries) have special designations used exclusively for "motorways". For example, the UK has M and A(M) roads, the US has the Interstate system, and Ontario, Canada has certain highways that start with "4". Not all roads built to a motorway standard are classified as motorways - e.g. the Westway is now an A road, and the highway to Peterborough, Ontario, Canada ( does not have the prefix "4" like other Ontario motorways - it is classified as a regular provincial highway. Often, but not always, these are substandard motorways. In addition, some roads are classified as a motorway, but have traffic lights in sections, e.g. Autoroute 20 in Quebec, Canada west of Montreal (

These official classifications vary a lot by country and region. Perhaps we should use country/region-specific tags for these?

As for E-roads, E-road simply indicates that the road is a major international route. Not all E-roads are motorway standard, especially in Eastern Europe. However, some countries use only the E-number on signs.Andrewpmk 07:39, 26 January 2008 (UTC)

Great summary, Andrew. It sounds like the "motorway" issue is fairly simple (it needs to be both a physical and administrative designation). I'm still not sure how the "substandard" motorways should be handled on the physical side (those that almost but not quite meet the requirement). Is the "motorway_almost" good enough for that?

On the classification side, I think we can probably get away with the national/regional/(county)/local system most places. I'm a little unclear on how the A/B system of the UK (and germany?) works though. Does A vs. B indicate who maintains the roads (that is, could it be mapped like A=national, B=regional?) I wouldn't have a problem with "A" and "B" values for highway:admin, if it doesn't correspond. Are the UK A/B roads equivalent to the German A/B roads?

As to E roads, do you think an "E" administrative classification would be necessary or helpful? Hawke 17:21, 28 January 2008 (UTC)

The administrative tag should be a country specific street type, because there is no way to really match all administrative types for all countries. e.g. "Bundesstrasse", "Staatsstrasse", "Kreisstrasse"... for Germany or "Class A Road", ""Class B Road" for the UK (if this is the right denomination). --D.S.E 09:00, 12 February 2008 (UTC)

Parking and road width

In my area, there are many roads which have parking for all but designated hours, in which the road gains a travel lane. How would these fit into the scheme? --Cohort 01:52, 12 February 2008 (UTC)


-1 from me for naming categories.


In Poland there are no counties and perishes (according to dictionaries) are part of a church not the state administration.

  • In the US, individual states are welcome to set their own terms for counties. Most states use 'county', but several use other terms like 'parish' (Louisiana, for one), 'borough' (Alaska), etc. --Cohort 23:27, 14 February 2008 (UTC)
  • Also, the existence of the highway:admin=county doesn't mean that it needs to be used everywhere. If there are no counties, and therefore no county roads in Poland, there will simply be no use of that tag in Poland. Hawke 20:31, 18 February 2008 (UTC)
    • OK. But there are four administration levels that need to be put into tags. What bugs me most is that naming always refers to a particular jurisdiction/tradition/language while numbers like NUTS are kind of an abstraction which may fit several different systems at once. Steelman 00:01, 19 February 2008 (UTC)
      • Well, it is pretty well-defined that OSM uses English for its tag names. The problem with a numeric system is that either they don't correspond with reality (NUTS), or they don't leave space for additional levels in between. --Hawke 21:26, 11 March 2008 (UTC)

There are two lists of road categories and classes in Polish wikipedia used in Poland. To sum up the article. There are eight classes (highway:physical):

  • freeways, denoted with A symbol,
  • express ways (trunk), denoted with S symbol
  • fast main roads (arterial), denoted with GP symbol
  • main roads, denoted with G symbol
  • collective roads, denoted with Z symbol
  • local roads, denoted with L symbol
  • access roads, denoted with D symbol

This is my translation, it's not the official one (if there is such). There is another article in english. The tables are in Polish but I can try to translate them if you think it would help

Each of these classes has got precisely defined:

  • width of the road (incl. sidewalks and shoulders)
  • width of each lane
  • minimal distance between crossings
  • width of shoulders
  • distance between sidewalk and the lane

There seem to be no problem with the first two categories which fit what we have now quite good. There is no reason to change motorway and trunk as it seems to be quite clear what they are.

What I really dislike:

  • motorway_almost, if a road does not meet motorway standards, it meets others (eg. GP class in Poland).
  • motorway_twolane,it can be drawn as a single way and tagged as oneway=false and motorway.
  • super_two, what is the difference between this one and motorway_twolane? Access control? It is not a physical feature.
  • boulevard = 2 * (multi_lane + oneway=yes)
  • two_lane + oneway=true = ?
  • significant (I'd call it major by contrast to minor), it's very highly subjective as street may become significant (havy traffic) because some traffic reorganization.
  • minor what is the physical difference between this one and two_lane?
  • primitive, we've just abandoned unsurfaced in favor of surface=paved/unpaved
  • 4wd, I doubt we can call such a thing road.

Hmm, I take your point and agree to some extent. However:

  • While each of the Polish designations may have precise definitions, those definitions do not apply globally. Not all countries have a designation for all(?) roads as Poland does. Even for those that do, they are not going to match the Polish definitions.
  • You say that it is clear what trunk is; I disagree. Further, I am not proposing replacing the existing highway=* tags.

Counterpoint to your dislikes:

  • motorway_almost, Not meeting motorway standards does not mean that a road will meet some other standard. As I said, not all countries have such precise standards for roads.
  • motorway_twolane, a motorway is assumed to be two lanes in each direction. This provides for a motorway which is one lane in each direction.
  • super_two, The difference is that a super_two is not divided nor access controlled.
  • two_lane + oneway=true = ?, I don't see your objection here. Are you saying that a two-lane road cannot be one way?
  • significant, I only hesitate to call it major because "major road" has other connotations around here (generally indicating some kind of highway).
  • minor these may be one or one-and-a-half lane wide (not in each direction). They are the "normal" residential roads (generally highway=residential or highway=unclassified
  • primitive, the surface of the road is not the same as its physical classification. That is, a road can be paved with gravel and therefore not be "primitive". The designation significant+surface=unpaved is a valid combination.
  • 4wd, You don't have to map these if you don't want to. That doesn't mean that they can't/shouldn't be mapped.

--Hawke 20:31, 18 February 2008 (UTC)

  • motorway_almost I am afraid there is no country where a highway that is almost a motorway is an unclassified road.
  • motorway_twolane two ways each one with oneway=true and lanes=1
  • super_two if it is not divided then we draw it as a single way, access control is not a physical feature.

There is no good one dimensional description of multidimensional space. Allow me a digression. I teach computer graphics and one of the projects is writing a programme that implements any of the visibility problem solutions. What most students choose is the painters algorithm with polygons sorted by the distance of their centers from the observer which obviously fails under certain conditions. It fails because they try to map 3D space into 1D. We should describe highways' physical features one by one.

  • Number of lanes,
  • width of a lane,
  • width of a shoulders,
  • what else?

Steelman 00:01, 19 February 2008 (UTC)

The US is a country where there are specifications for a motorway (Interstate highway) and if those specifications are not met the road is unlikely to be designated as a motorway. I believe the only exceptions to this are historical. Below that level, there are no standard official designations for the physical design of a road.

Access control is a physical feature, but unrelated to super twos.

Describing all the physical features one by one fails because it is unwieldy to tag every single highway with every single scrap of information, even if that information is available. I for one will not be getting out my tape measure and trying to see how wide the lanes are on a motorway. (It's unsafe, even putting aside the legalities).

The highway:physical tag is intended to be a shorthand for describing at a high level the basic physical structure of a road. --Hawke 21:26, 11 March 2008 (UTC)


There are also categories (highway:admin) which describe roads' function in the communication network and depend on who governs it. There are four categories:

  • state, classes A, S, GP and sometimes G
  • voivodship, classes G, Z, sometimes GP
  • powiat (county), classes G, Z, sometimes L
  • gmina (commune), classes L, D, sometimes Z

I suggest using NUTS and LAUs numbers for highway:admin.

Sseems to me that those map just fine to the documented classifications:

...while the NUTS classifications do not map well to the road classes you describe:

  • NUTS-1: Groups of Voivodeships -> ??
  • NUTS-2: Voivodeships of Poland -> ??
  • NUTS-3: Podregiony (Groups of Powiats) ->??

..and I don't see any documentation of the LAU values.

--Hawke 20:40, 18 February 2008 (UTC)

The LAU values are here. LAU1 is Powiaty, LAU2 is Gminy. Ivansanchez 01:09, 19 February 2008 (UTC)

Eurostat presents appropriate value/name pairs here. It is one of article's external links. As you may see in Poland NUTS-1 level units are named Regiony (Pol.sing. Region) which makes naming voivodship (NUTS-2) highways regional quite misleading.

It is a fact that not all NUTS numbers map to real Polish administration levels hence not all would be used for tagging. There are no administration units for NUTS-1 and NUTS-3. Nevertheless NUTS/LAU make the values comparable across different countries, that's good. You might argue that in a different country there will be no NUTS-2 roads (voivodship roads). I say: so waht? Should highway:admin influence rendering? IMHO not.

Steelman 22:45, 18 February 2008 (UTC)

In countries like Spain, the administration level paying for the road does influence rendering. At least, in the official government-issued road atlases. That's red for nationwide, orange for NUTS2-equivalent (autonomical), green for NUTS3 (provincial), yellow for LAU2 (municipal).
Live in a country that doesn't have any government body at NUTS1 level (like spain)? Don't worry - just don't tag any roads as NUTS1-maintained. Ivansanchez 01:17, 19 February 2008 (UTC)
Steelman, you pointed out that the "Regiony" of NUTS-1 are not an actual administrative level used in poland, so the risk of confusion is low. "voivodship" seems to translate to "province" in English. Unfortunately, "province" has existing meaning as an administrative level in several countries. I used "region" in hopes of preventing conflict with the existing name. Since it seems to be unavoidable, would you be more comfortable with "province" instead? --Hawke 21:52, 11 March 2008 (UTC) these NUTS levels correspond to an entity that maintains the road, or in any way to the road's administrative designation? If they don't, and if that's the case for many countries, I would say that the NUTS levels are totally inappropriate for this classification. I guess I don't understand what you're suggesting. Are you saying that the road classifications should be by the NUTS-1/2/3 levels in Europe? How can that be done if the NUTS levels don't in any way correspond with the road classifications?

--Hawke 23:19, 18 February 2008 (UTC)

Can we get an example of a road classification not bound by NUTS levels? Ivansanchez 01:17, 19 February 2008 (UTC)
From the above: Poland's highway administration levels are: state, voivodship (province), powiat (county), gmina (municipality). Its nuts classifications are: NUTS1=Groups of Voivodeships, NUTS2=Voivodeships, NUTS3=Podregiony (groups of powiats). There is no correspondence whatsoever. The only correspondance is at the Voivodeship level. --Hawke 21:02, 11 March 2008 (UTC)


I propose this one as a positive or negative numeric value giving a hint for autorouters.

Steelman 13:28, 12 February 2008 (UTC)

This is not relevant for this proposal. --Hawke 19:55, 18 February 2008 (UTC)

Dual carriageway

What is the intended purpose of the dual_carriageway tagging? I know of several dual carriageway roads that change in spacing along their distance. In my opinion, dual carriageways would best be represented as two separate ways with a relation. --Cohort 23:16, 14 February 2008 (UTC)

Dual carriageways would indeed be represented as two separate ways, perhaps with a relation. But each of those ways still need a classification. --Hawke 20:41, 18 February 2008 (UTC)
However, not all dual-carriageway roads are alike, so IMHO should not be tagged alike. (ex. Route A is a minor roadway with one motor vehicle, one cycle and one parking lane in each direction; Route B is a major route (state highway) with three motor vehicle lanes in each direction.) --Cohort 23:43, 18 February 2008 (UTC)
OK -- I guess to answer your question, dual-carriageway is a catch-all for DC roads that don't fit with the other categories, in the same way that unclassified is for highway, and that minor is for other highway:physical. What suggestions have you for additional designations, with the physical characteristics to differentiate from other roads?
Just use highway=trunk, lanes=3 for each carriageway on the 3 motor lane one, and highway=tertiary (or unclassified), lanes=1 for each carriageway on the minor DC. This whole proposal overcomplicates things which can already be done very simply with the current tagging scheme. Richard B 08:53, 19 February 2008 (UTC)
That's how I have it now. --Cohort 09:43, 19 February 2008 (UTC)
Except for the part where it kind of mashes the physical and administrative designations into one thing that is both and neither. Remember, this proposal is intended to be additional information beyond what highway describes. It is *not* intended to replace the existing highway tag. --Hawke 16:13, 19 February 2008 (UTC)

Dual carriageway has been removed.


Some interesting edge-cases listed here (e.g. motorways with roundabouts on). Ojw 17:41, 24 March 2008 (UTC)

more coices

Why not make a longer list of choices for the physical type of road? Say, 25 values from super-highway to unpaved-unmarked. Then create a big table that maps the different kinds of roads that are in the different countries to a subset of road types. e.g.

highway:physical type US GB Germany Lala-Land
A_motorway directions seperated, multiple lanes, hard shoulder interstate motorway autobahn not used
B_motorway directions seperated, multiple lanes  ?  ? autostraße not used
C_motorway multiple lanes, hard shoulder  ?  ? autobahn-like bundesstraße not used
A-primary paved, wide, high-traffic, hard shoulder  ?  ? not used Gaga-road
B_primary paved, wide, high-traffic  ? not used Bundesstraße Umum-road
A_secondary paved, full-sized lanes  ?  ? Landesstraße, Staatsstraße president's home street
B_secondary paved, middle-sized lanes  ? not used not used not used
C_secondary paved, narrow lanes  ? not used not used president's garden ways
A_tertiary paved, no lanes  ? not used Kreisstraße not used
B_tertiary paved, no lanes, middle-width  ? not used not used not used
C_tertiary paved, no lanes, narrow  ? not used not used not used
H_tertiary unpaved, no lanes, extremly narrow not used not used not used most streets
A_residential paved, lanes  ? residential streets with line wohnstraße mit mittelstrich not used
D_residential paved, narrow  ? narrow residential streets wohnstraße unter 3m Breite not used

Henry Loenwind 00:22, 2 May 2008 (UTC)

Two reasons: First, because graded levels of the highway=* tags are not what I'm going for. This is not intended to supplant the highway tag, but to complement it. Second, because physical attributes can not and should not have international equivalence. "Full-width hard shoulders on a divided, limited-access road" is going to be the same no matter which country you're in, or what such a road is called in that country. --Hawke

Simplify + relations

This should be simplified. The existing tagging with highway=* is used to tag the physical nature of the road while the ref=* tag shows who have the administrative role in Brazil. The administrative part can also be put into the relation, as the entire stretch of the road will have the same administrative role, while different parts of the road can hold different quality (part motorway, part trunk, part primary). I feel this is a much better way to do it, as the relation also can hold instructions to how to render the identifications of the road (Ref numbers), and so on. --Skippern 19:07, 23 February 2009 (UTC)

highway=* doesn't describe the physical nature of the road; it just claims to.
ref=* is close to highway:admin=*, except that it doesn't indicate international equivalency, nor any sort of hierarchy. Nothing shows that ref=I39 is an administrative equivalent to ref=M25. Nothing would indicate where ref=US:WI 29 falls in comparison to ref=B123.
Obviously, dealing with relations is unrelated to this proposal. As with other tags, you're free to use it in relations, or not, as you see fit.
Thanks for the input. --Hawke 23:39, 23 February 2009 (UTC)

I agree that a relatively simple tagging scheme is best. The one redeeming feature of the current highway tag is that it is simple. The way I see it, roads have three groups of characteristics to be mapped:

Physical - What the road is like to travel on. Mainly the size and road type.
Importance - How far the road goes, what it connects, how low of a zoom level it should be rendered at.
Administrative - Who maintains the road, and what type of signs are found along it.

Currently the highway tag tries to encompass all of these and, while it works, I think it could work much better. The highway tag takes care of physical information well enough for a basic road map, in my opinion. A more detailed, optional physical tag, as proposed here, could be useful for future renderers.

The move toward tagging routes with relations is great. Because a route name/number may stay the same while the physical nature of a road changes, I think relations are the way to go for the administrative information.

I feel that the real missing piece in OSM is the Importance. As you zoom in on the slippy map, first you see motorways, then trunk, then primary, etc... The problem with this is that sometimes a small two lane road is the most important road in an area, and it should be rendered at a very low zoom level because it is the only road that connects two large towns. This road should be tagged as secondary for is physical attributes, but we need a tag to indicated that is should be rendered at a low zoom level. Alternatively some local areas have many short motorway spurs and loops that don't matter to someone looking at an entire region. At low zoom levels these roads are clutter. We need a tag to indicate that these roads should not be rendered until a high zoom level.

Hawke, I'm trying to come up with a proposal for road importance. It seems to overlap a bit with your proposal, hence my comments here. What do you think?

--Ezekielf 15:41, 24 March 2009 (UTC)

Physical characteristics

-1 from me on the whole idea of highway:physical as currently proposed. highway=* and all its values already describe physical characteristics, please do not duplicate this existing tagging scheme. lanes=* already describes the number of lanes. Links can already be represented with highway=motorway_link, highway=trunk_link etc. "Significant", "minor", and "narrow" are highly subjective. "primitive" and "4wd" are types of the existing highway=track, with different surface=* values. Probably there's a need for tagging the presence (or absence!) of hard shoulders, and the existence of centre turning lanes, but those should be raised as separate proposals. Too much duplication in this scheme, and yet another rather arbitrary sliding scale. --achadwick 22:50, 23 February 2009 (UTC)

It seems clear that highway=* does not currently describe physical characteristics. Key:highway claims that it does, but then goes on to say things like "Important roads that aren't motorways" and "Administrative classification in the UK, generally xyz" These are not physical descriptors. Perhaps highway:physical=HQDC[1] is acceptable?
"significant" is admittedly subjective, but I can't think of any way to describe the same thing which is non-subjective, yet can be applied globally. This is similar to how highway=tertiary is currently used, I think.
"narrow" is clearly defined as "not wide enough for two cars to fit (pass) side-by-side". This is not subjective.
surface=* is not related to this "primitive" or "4wd": either can be made of gravel, dirt, and pretty much any surface material you can think of.
Thanks for your input. --Hawke 23:16, 23 February 2009 (UTC)
If you read the country specific definitions on highway=* you will see that some countries (UK) describes the road class, while other countries (Brazil) describe the physical character of the road. You claim that highway just claims to describe the physical attribute of the road. Read the country specific definitions (all of them) before coming with this. You also mentioned that this proposal have nothing to do with relations, I feel that most of what you have put in administrative have more to do in the relation than the actual tagging. I suggest you split the suggestion if you want to improve the highway=* definitions, one for better descriptions in the existing scheme, and one for additional tags to describe what is missed by the existing scheme. --Skippern 11:38, 24 February 2009 (UTC)
OK. Australia, Austria, Bulgaria, Denmark, France, Germany, Hungary, Iceland, India, Ireland, Italy, Lithuania, Netherlands, Norway, Poland, Switzerland, Spain, Sweden, and the UK all use administrative designations. Brazil, Russia, and the USA use a mixture of designation and physical characteristics. Korea uses physical characteristics. I was unable to determine for Slovenia. So that's nineteen administrative, three mixed, one physical, and one unknown. Plus the official descriptions are administrative. So I'd say it indeed only claims to describe the physical attributes of the road. Regarding relations: If people want to put these tags on relations, they're more than welcome to do so, but thatis not part of this proposal. Regarding "improving the highway=* descriptions": Given the overwhelming existing use of highway=* as describing administrative classification of a road, I would prefer not to try to change that. --Hawke 17:14, 24 February 2009 (UTC)
So you think it is better to discard highway=* for a completely new tagging scheme? Would think it was less work, and probably a better result, to improve the existing scheme. Confused.... --Skippern 17:25, 24 February 2009 (UTC)
Absolutely not. I do not want to discard highway=*. As I see it we have two options:
  • Redefine highway=* so that it actually describes physical characteristics. This would go against the huge amount of effort which has gone into the highway tag. In addition, the information captured in highway=* is definitely useful and relevant, so it would need to be captured somewhere. Doing this would require a huge effort of moving this data over to some new tag, educating users to use the new tag, etc. That's all assuming that by some miracle everyone involved agreed that changing it like this is a good idea.
  • Use an entirely new tag which actually does describe physical characteristics. That's what this proposal attempts to do. All that's needed, if/when Template:Highway:physical is accepted, is to change the text in Key:highway which claims that it describes physical characteristics. I expect this will also be a hard change to get people to accept, because it's long been a tenet of that tag that it describes physical characteristics (even if the reality is different). But the effort involved in that change is relatively small compared to the other option. And even if the change to Key:highway would not be accepted, at least we gain a way to actually accomplish what highway=* claims it does.
Of course there is always the third option of leaving things as they are, continuing to pretend that highway=* describes physical characteristics, and having no simple method for describing the physical characteristics of a road. But I hope you'll agree that that third option is not a good one. --Hawke 18:08, 24 February 2009 (UTC)
Don't get me wrong, I am not trying to prevent the system from becoming better, my reason to question changes is that I don't want us to head down the wrong road, so to say. Any improvements are always welcome. --Skippern 18:13, 24 February 2009 (UTC)
Glad to hear it. I do appreciate the input. --Hawke 18:20, 24 February 2009 (UTC)

Administrative characteristics

highway:admin seems reasonable as a concept to me, though I won't be using it myself for my tagging. Have you considered use of the widespread operator=* for this concept, and if not why not? Maybe use state for the top level to avoid reuse of the word "motorway" which currently describes physical characteristics. Maybe not. Please give examples for regional. In the UK, a parish is a rather small division which is not responsible for road upkeep (any more). Do these divisions have an external standard definition? --achadwick 22:50, 23 February 2009 (UTC)

I have not considered "operator" because operator doesn't make sense in this context. operator=Foo County, operator=Bar Parish, operator=Baz Borough don't show that they're all at the same administrative level.
Why would I use "state" for motorway, when motorway is much clearer? There is some potential for international confusion, but it's easier to pick one term from "motorway, interstate, autobahn" and explain that they're different countries' terms for the same thing, than to pick a completely irrelevant term which will certainly confuse people in countries that have states.
Examples are already provided for "regional", please read the main page.
In at least one state of the US, "parish" is a term used in place of "county". "borough" is another such term. They're all equivalent.
Thanks for your input. --Hawke 23:30, 23 February 2009 (UTC)
Why not use admin_level=* for roads maintained by governments of various levels? For me operator=* would indicate that maintenance is turned over to a private company (which have happened with several federal and state roads in Brazil), but the admin_level combined with operator can show that it is a federal road with private maintenance..... --Skippern 11:16, 24 February 2009 (UTC)
Hmm, good idea. admin level=* seems pretty tightly bound to boundary=* though. Any idea how to describe that tag so it can apply to both? --Hawke 17:17, 24 February 2009 (UTC)
I have asked to have admin_level=* separated from boundary=* for two reasons:
  1. boundary=* are at the moment only describing boundary=administrative
  2. This kind of misinterpretation the admin_level=* only applies to boundary=*
--Skippern 17:23, 24 February 2009 (UTC)

Missing trunk (or similar)

At least in Norway, Sweden and Finland, and probably other countries as well, there is a national road network serving pretty much the same function as the autobahn in Germany or interstate in the US. These are, however, far from motorway quality. In Norway these are called stamveg (stamväg in Sweden, kantatie in Finland). Today, trunk is used for these roads, which have a separate planning and budgeting process from other nationally maintained roads.

I would suggest adding Trunk Part of a nationally maintained nethwork of through roads. -- Gustavf 11:56, 24 February 2009 (UTC)

Are you suggesting this as an administrative level, or physical? --Hawke 17:15, 24 February 2009 (UTC)
Administrative. I am not sure if trunk is the best word, however, since it is already used for the highway tag (but so is motorway). -- Gustavf 17:58, 24 February 2009 (UTC)
On the other hand... Is "motorway" really an administrative level? They are, I suppose, nationally maintained. -- Gustavf 10:02, 25 February 2009 (UTC)
Not necessary. I know of a few that are maintained by the city (i.e., "motorway" cannot be used as an administrative level) --Skippern 17:23, 25 February 2009 (UTC)

Superflous descriptions of tree structure

The tree structure described in the proposal page, is that it really can make a lot of confusion about the road structure in a country where road networks is not as uniform as in western Europe and Northern America. For instance, an elevated road where traffic flow in different directions are divided by level, all intersections are ramps leading to different levels, and each level have two or more lanes. This would indicate motorway. Speed limit are 60kph, this would indicate something lower. Same road as above with speed limit 100kph, this would indicate motorway, but speed bumps are introduced before it merges with another road of same level/type, this is no longer a motorway. Besides a large part of the road structure in Brazil is unpaved, though you do not need a high clearance vehicle. Almost all roads between the major highways in eastern part of the state Espirito Santo are gravel, dirt, earth, or sand tracks, some have as much room as to fit 3 lanes. Is this really a "primitive"? Should more than half of the roads in the state be tagged as such? --Skippern 17:43, 24 February 2009 (UTC)


I'm not convinced this complicated scheme is entirely necessary. There is a dichotomy at the moment between some areas using an administrative classification (e.g. UK, where the existing scheme originated), and others using a physical one (e.g. USA). When you look at the reasons for this difference, it becomes less different than it first appears - in those territories where the administrative classification is used, one can deduce some information about the road from this classification. In Britain, the first system of classification gave A and B, which translate to "primary" and "secondary". Then the government introduced (in two stages) "trunk" roads, which were to be maintained by central government. Then came the "motorway", so named because they are restricted to certain types of motor vehicles. "motorway" and "trunk" can (usually) be observed from this classification, and these are maintained centrally (Highways Agency, Transport Scotland, Welsh Assembly). These roads are maintained centrally because they are considered strategically important, while other roads are maintained locally. Signing is generally of good quality, and one can deduce the classification from the signs, because this produces a logical hierarchy. In the USA, there are the Interstate Highways and the US Highways, which are complementary rather than hierarchical, and then you have state and (where applicable) county roads. However, these don't always make a clear hierarchy. For instance, the main ring road around Las Vegas is a dual carriageway that is mostly grade-separated, but yet it is still only a county-level road. This status comes from the fact that the road was funded and built by the county - in general, the origin overrules any other useful information, so high-speed toll routes operated by the state are generally not assigned to the US or Interstate system. A US Highway may be a two-lane road through small towns, or it may be a dual-carriageway. In particular, the US has a propensity for building and routing major highways through rather than around towns, so the more desirable by-pass may carry a lower classification than the through route. Throw into this that signing is frequently misleading due to some states operating a policy of not signing out-of-state destinations on their highways. Some states number their routes by importance, others by geographic orientation - a smaller number does not necessarily mean a more important highway. Also, to add more to the confusion, county roads are maintained by the county, state roads by the state, but the US and Interstates are maintained not by the government but by the states with funding from the federal government. As a result, one can't necessarily observe a clear hierarchy of values from the administrative classification. So, in both cases, the value of the "highway" tag amounts to characteristics that can be observed from the highways in the relevant country. Chriscf 17:30, 26 February 2009 (UTC)

Was there a purpose for which you explained how the American road system works and the history of the British road system? I agree with the facts you've listed, but to me that points more strongly toward a need for a way to distinguish the physical characteristics from the administrative classification, but to still be able to tag it simply rather than having to enter ten tags for every road.
Perhaps your comment was only directed at highway:admin=*, and there I'm increasingly inclined to agree. admin_level=* can accomplish the stated purpose of highway:admin=*. And I think highway=* is much closer to what I was trying to do than I realized. All that's needed is to remove the pretence that it is about physical characteristics, and I think we're golden. What do you think? --Hawke 19:16, 26 February 2009 (UTC)