(German version of) flowchart for way classification in Brazil

Hallo! Wie geht es Ihnen?

(Sorry but I’m not brave enough to continue with my rusty Deutsch. :P)

Following a recommendation by a German user (username “nesol”) and because the German-speaking community is the most active in OpenStreetMap, I would like to share with you a flowchart that the Brazilian community has recently come up with to simplify assigning the “highway” tag. It kind of synthesizes what is written in many articles in the wiki, and also adapts these definitions to the Brazilian context. We’ve tried to use as few ambiguous characteristics as possible in order to minimize edit wars. Such a simplification is good for beginners and (we hope) it may be an aid in ambiguous cases because it clearly states priorities established by consensus. I think it is reasonable to say that we aimed at never producing incorrect results, just maybe slightly inaccurate in a few cases (hopefully much less than 5% of the time).

Nesol helped me improve the German translations. He also pointed out a major difference with the classification style in Germany, Austria and Switzerland (DACH): the usage of “tertiary” roads outside urban areas, which are paved in DACH but (now) unpaved in Brazil. The flowchart also contains an official national classification in the boxes with a “[br]” prefix. Note that the nodes in green (outside urban areas) and yellow (inside them) are very specific to Brazil, while the others are somewhat more “universal”; perhaps a notable exception is the application of “living street” since this kind of way does not exist in Brazil officially.

You can see the German version here: http://i.imgur.com/CuZcyRC.png
There is also an English version here: http://i.imgur.com/YH8azIA.png
Source yEd graph files: https://www.dropbox.com/sh/8ifgduyovz9880u/sOE2QvDLci

We’ve gone through hundreds of e-mails making improvements (adding or loosening requirements, and also changing the flow). It hasn’t suffered any major change in the last week, so it can be considered stable. However, opinions and suggestions are more than welcome. You can also freely modify and use the graph files.

We are largely based on the text in the English wiki, which is the most accessible to the Brazilian community, but we also considered approximations to the classification style used by our Latin American neighbours.

I know there is a user currently applying this method to major roads in the state of Rio Grande do Norte (RN), but I don’t know his progress yet. I have applied it to all urban ways in the city of Porto Alegre, where the results of “rising by priority” can be clearly observed.


my comments:
I consider your proposal not as a tool for simplification but for clarification (which imho is more important). It could e.g. act as a tool to decide whether to tag as footway or as path, which is used quite confusingly at present).

The layout of your graph was confusing me first: You have several entry points (red at top, orange scattered in between). Perhaps it would be better to arrange them at the left border in order not to introduce very long arrows.

I don’t understand the brazilian classification:
pavimentada → surface=paved?
leito natural → natural bed / surface=ground?
implantada → introduced???
duplicada → duplicated???
multi-faixa → lanes >=2?

Thanks for your effort!

There is a massive fault in your chart. Please don’t use surface and width as a criteria for highway-class. highway-class is just based on importance of the road. Surface is an additional feature, which should be taged in surface-tag. Please take in mind, that wiki-pages are mostly written by european or north-american guys and most of them doesn’t think about road-conditions in other parts of the world. So there might be some explanations, which fits only to european roads.

highway=track should only be taged, if the way is used (mainly) for agricultural- or forestry-use.

Thank you for your comments seichter. I’ll try to improve the layout, which was done automatically by yEd (it tries to minimize the number of arrow but not to clearly present the entry points). Initially there was an “entry” node that was removed to make the graph look simpler, perhaps it’s best to put it back.

The terminology used in the Brazilian classification is actually very vague and have little use beyond labelling. Official documentation actually describes these classes of roads in some detail. Such documentation is linked to in the wiki (http://wiki.openstreetmap.org/wiki/Pt-br:How_to_map_a) but since the text is in Portuguese I’ll translate.

  • Duplicada (literally “duplicated” [carriageway]): a divided highway with at least 2 lanes per direction; division means there’s central reservation with a rigid separator (e.g. guard rails) or two independent carriageways with distinct contours around obstacles
  • Multi-faixas (literally “multiple lanes”): at least 2 lanes per direction but not divided
  • Pavimentada (literally “paved”): superior coating such as asphalt, concrete, cement or paving stones [the former two are actually very rare in Brazilian highways]
  • Implantada (literally “implanted”): built according to geometric design standards, with unpaved surface with primary coating [compacted soil], supports traffic all year round
  • Leito natural (literally “natural bed”): very first opening on the natural ground (may or may not have primary coating), not following geometric design standards

These official definitions still leave space for some ambiguity, but an advantage is that it can be easily copied over from an official source, which makes the task of updating the map much easier. Maybe only those of type “pavimentada” need a little extra work because they require checking the number of lanes. After a lot of discussion we ended up agreeing that checking for the presence of other characteristics such as hard shoulders or central reservation is important but not fundamental for an acceptable initial classification.

You are right in Germany where official classification is based on responsability (maintenance) (which is slightly different from importance) and not on surface. You may find highway=tertiary still unpaved and very often tracks with surface=asphalt.

Classification in Brazil (and other countries) might conform to other criteria e.g. it could be stronger related to road condition / hard shoulders than in Europe, but it should not contradict tagging practise here. Width and surface can’t be used for classification in Germany (and most of Europe)

As a preliminary conclusion: It might be a demanding and even impossible task to classify all countries by the same decision flow.

You are absolutely right, but we ended up with heated discussions about what “importance” really is (or, better, how not to fight over it). Each person had a different view: some thought that importance is linked to the size of the cities that the road connects (raising questions about what to do when the road goes through cities of very different sizes), some thought that the administrative classification should be used, some thought that the structural characteristics are the only aspect to be considered, and then some simply preferred to use their subjective sense of importance without any justification. It was also clear in the discussions that importance was evaluated by different people considering different contexts: some thought that importance was to be judged in the local context of a state, others of a country, and then others of the world. The winning argument (least objections but still contentious) was that importance is correlated with traffic volume and that traffic volume usually affects structural characteristics which reflect government’s priorities in improving or expanding particular roads. Therefore these easily measurable characteristics are “likely” to indicate an approximate level of importance. More than that, classification would then acquire some useful meaning to the final user of the map.

This argument gained support when we discovered that the official classification by the government is based on several structural characteristics.

It is not forbidden to diverge from the classification indicated by the flowchart, but then it is important to justify it in the “note” tag (sorry, I should have said that before, it’s written on the wiki in Portuguese). Otherwise, anyone could disagree and change the classification again, possibly beginning an edit war. If newcomers get involved in arguments over advanced topics, they are less likely to wish to participate. Also, if classification is too complex, it’s likely to be done only by a few specialized collaborators (I even labelled this “inaction by ignorance”). Both are bad for Brazil, as there are still too few contributors.

And also, classification neither determines the value of other tags nor eliminates the need to use other tags such as surface, lanes, width, etc. (also written in the wiki in Portuguese). Parts of the flowchart allow for ascribing an “acceptable” classification when not all characteristics are known, thus classification itself only offers some basic guarantees which may be even overruled (though rarely, we hope).

One thing we discussed about is how well our classification matches that of the rest of the world (and well, you know, it’s very different in every country). There were no objections to approximating the expectations of European or American OSM users - quite the contrary, we think that tourists using the map should apply similar judgement here as they do in their home countries. Several roads in Brazil are European-class and many are not, that’s why we found useful to diverge from most countries by using “tertiary” for unpaved roads in good conditions and use “unclassified” for those in poor conditions.

“track” was perhaps the most difficult one to accommodate (and we heard that even in DACH there are divergences). We tried to ascribe structural characteristics to it since many agricultural roads are very similar to certain urban roads under specific conditions (such as those in unpopulated city outskirts, or even inside green areas in large parks). There was a need to mark its difference to “unclassified” (the unpaved roads in poor conditions), and so we decided to use “width” as a strong indicator. As a principle, it can be wrong, but it will be right in the vast majority of the cases.

Sorry for the long answer. I’ll bring your comments to the Brazilian community later on.

In fact highways are classified in Brazil also by administrative level (usually in only two groups: national highways and state highways). The problem we found with this approach is that road quality varies a lot within the same class. Several state highways are large and safe motorways, while many national highways are small unpaved roads in poor conditions. There are even some city-managed highways in better condition than some national highways. So using the administrative level as a classifier would be somewhat useless and even misleading here. The level is indicated on the road name anyway using a prefix: national highways start with “BR” and state highways start with the 2 letter code of the state.

These are probably the most controversial parts of our decision flow. Because of that, in order to reach consensus, many were left optional. I know there is some controversy over footway vs path in DACH, I think we may have “oversimplified” the issue by defining footways as paved. What it means is that one probably expects footways to be paved since most are. A similar statement could be made for footway vs pedestrian based on width.

I agree. I don’t think it is impossible, but it would require a huge amount of democratic work and there would be almost endless arguing.

This seems to differ from European practise: Classification is done almost only on administrative level, state is covered by tags. Especially tracks can be everything from smooth asphalt to almost untransversable mud holes. That means: For usability the additional tags are crucial. Another extreme example: There are paths where you can walk on high heels, for others you need climbing equipment. Distinction has to be made e.g. by SAC-scale.

As I understand you have kept more things optional as can bee seen from the decision chart. But if you keep too much open, it’s practical use gets questionable.

Indeed, my very first proposition to the Brazilian community was to use administrative level for classification, but many objected. I will bring your comments to them, but I find it hard to change it since I couldn’t get a single person to agree with my point of view.

As for paths it seems that a distinction with footways is still unclear. I’ve heard of suggestions that footways are used in cities while paths are used outside them, but to me it doesn’t make much sense to create different categories for similar things that only differ by their location. The Brazilian community did not have many comments on this topic, I believe most were indecisive and just waiting to hear what’s the right thing to do.

Agreed. We attempted to limit optional choices to classifications that we found quite similar, with almost no usability impact. They were mostly added to allow a minimum-quality classification based on the official classification with minimal extra work.

Ideally one should gather as much information about a way, add as many tags as possible (surface, lanes, etc.) (a lot of work, possibly involving some research for each tag) and decide about the classification using the flowchart as a guide and a keen subjective sense of how appropriate the flowchart really is for each case. I think a better moment to aim at this ideal is when the Brazilian community grows, when there will be more contributors in every state that care about details. Currently the map in Brazil has significant deficiencies when compared to its commercial alternatives and few will be attracted to the project before these deficiencies are addressed. Classification is probably one of them since current differences in style are confusing to (non-specialist) users and are likely to just be ignored if they don’t display a consistent, recognizable, useful pattern.

I’ve followed this suggestion and this is how it looks: http://i.imgur.com/pFCApOT.png

Hi, thanks for your efford.

I agree with aighes that the definition of tracks does not fit to the german practise.
A highway which make buildings accessible should not tagged as track.

An other point i disagree with are the living streets. If you come from the residential area/ Wohngebiet trunk it should be residential.
Even from the other trunk residential fits much more, but there i am open for discussion.

living street should be reserved for streets, where pedestrians have legal priority over cars.

I see. I tried translating the German wiki article to see if I could understand the German criteria for tracks, and it seems to focus on usage: only for forestry and agricultural ways (as said before by others). Some questions could be risen when realising that some of the lowest-class ways maintained by the Brazilian federal administration can be found in more difficult transit conditions than a typical track in a farm, and also that several paths in parks and in suburbs (but not in farms or forests) look a lot like the photos of track examples in the wiki.

We could surely switch back to a “usage” criterion, so I’ll bring that up in the Brazilian discussion.

By the definition of “living streets”, no such ways exist in Brazil as there is no legal traffic sign indicating priority for pedestrians (there’s only a sign indicating exclusivity for pedestrians). That would be a good argument for not applying “living streets” here at all, but I think (and no Brazilians objected so far) that this would be a wasted opportunity. There are places where it is presumed (merely by practice) that pedestrians have some priority, such that the ways would function like living streets. One such place is in tight streets inside shanty towns, where traffic is slowed down by heavy pedestrian traffic. Similarly, for example, in my hometown’s downtown there’s an area where it’s common practice to just walk on the street without caring about passing cars, simply assuming they will all stop, though that’s not legally mandated. It is a small area comprising about 6 blocks and it doesn’t occur anywhere else in the city. People (pedestrians and drivers) have practised that assumption for decades without major incidents. There is so much pedestrian traffic (due to high concentration of street commerce) that going through that region by car (or even by bike) is undesirable almost every hour of every working day. So it arguably fits better the typical expectation of a living street (a street with many pedestrians that have priority in practice) than that of a residential way (a street with many residences, eventually with a few pedestrians here and then). Knowing about such culture is important not only for residents but also for tourists, which would then take extra care when passing through on a car, or (more reasonably) completely avoid them.