Interesting analysis, thanks for sharing. I wonder if another reason for low mapping levels is that mappers think there are “obvious” defaults, e.g. tagging a residential street with both sidewalks feels redundant if 95% of residential streets in a city have two sidewalks, and tagging that a rural road has no sidewalks feels equally redundant in areas where nobody would expect them to have one.

I don’t know if you are aware of OpenStreetBrowser.org, it shows a lot of detail about sidewalks in the Transportation -> Walking -> Footways layer.

Hi Chris

Thanks for the comments.

I have had some interaction with the StreetComplete account on Mastodon on this topic. In fact that’s who suggested loading the GeoJSON into SCEE - I was only familiar with vanilla SC and didn’t know that was possible until that conversation.

In generaI I’m not sure if this approach is really scalable. One thing I mentioned in the repository documentation, but not here, is that my initial experiments with generic global elevation models weren’t very successful at discriminating elevation changes over short distances. It was only when I switched to the high resolution Spanish national data that I started getting interesting results. I’m not sure how that would translate to a global application.

Alan

In fairness to mappers, it’s not always easy to derive a name from what is signed in the real world.

Many waymarked routes near me don’t have nice well-defined names like “King Charles II England Coast Path”. I have a photo of an information board labelled “GR249.4 Variante Villanueva del Rosario - El Chorro - Etapa 2: Antequera - Valle de Abdalajís”. The guideposts along that route route are labelled “GR249” (not GR249.4 like info board) and sometimes “Gran Senda de Málaga” (mentioned only in the small print in the info board). Disentangling a ref and a name that is not a description from all that is not trivial!

Indeed, I understand the aim of the ATP work and that it is not intended to replace manual mapping. I just thought it interesting that in an almost-central zone of European city, even if we had perfect information about all brands, it would only cover a small fraction of POIs.

Yes the location within malls does seem to be a problem. Probably not helped by the fact that (near me at least) there is no consistent addressing system for malls (even the name of the mall is often missing from published addresses, so there is just the name of a long avenue with no number).

Tiene sentido, muchas cosas han cambiado desde 2011 y el PMUS parece conformar más a la realidad actual.

Miarando el PMUS, me parece que hay un único eje de primer orden oeste-este por el centro (Paseo de las Curas, Avenida Manuel Agustín Heredia). Entonces unas calles en el eje Paseo del Parque - Avenida Andalucía, actualmente “highway=primary” en OSM, bajarían a “secondary” o incluso tertiary”- lo que no me parece mal.

Pero mirando el PGOU no lo tengo tan claro, me parece que hay dos ejes de primero orden pero no estoy seguro.

From what I’ve seen, in Spain “amenity=fast_food” is used mainly to tag the international burger chains plus some pizza and kebab places. In the thinly populated “doughnut” (surrounding Madrid but excluding the coast), I would guess that a lot of small towns have one “nightclub” but nowhere that specifically matches that definition of fast food. The stereotypical “morning after” food is “chocolate y churros”. Some places that specialise in this are tagged as “cuisine=churro”, but in general it would be hard to tell these apart from standard bars and cafes in OSM tagging.

Hola Daniel, Gracias, ha sido muy útil, y me ha dado una pista para solucionar algo que me había molestaba un poco: ¿cómo recordar en el terreno cuáles contenedores he comprobado antes, y cuáles no? He añadido otro paso más: hago la consulta exactamente cómo describes. Desde el botón Exportar, elijo la opción GPS, y paso el archivo .gpx a mi telefóno. Luego abro el archivo en la aplicación OsmAnd (ménu My Places, Tracks). OsmAnd interpreta cada contenedor en el archivo cómo un “waypoint” y lo destaca en el mapa. Así es facil dirigirse a los contenedores que aún hacen falta comprobar. Imagino que se podría hacer algo similar en otras aplicaciónes que acepten archivos gpx.

I haven’t mapped a lot of this myself so my comments probably have limited value, but based on my observations of how this tag is used in Ireland and Spain this would be my reaction to your examples:

1 - I don’t think the size of this area is a problem in itself - I would be OK with a large polygon if the whole area is genuinely residential. But I feel that is not the case here, as it contains a lot of non-residential landuse (hospitals, the Jesuit centre, sports grounds that are used by people beyond the immediate area). This feels like a first step at mapping, aimed at conveying the general idea “this is a built-up area, not countryside”, that could have been refined over time but wasn’t.

2 - I have never seen this before and I don’t think it is consistent with normal use of this tag. First impression is that it has been created by someone unfamiliar with the use of multipolygons, resulting in meaningless strips of land tagged as residential. Changing to a multipolygon would be technically better, but I don’t think I would be happy with it even then, as there seems to be no reason to connect this specific set of houses. The information “there are scattered individual houses here” could better be conveyed, I think, by mapping building=house (presumably if the available imagery allows mapping of residential plots at this level of detail, it would also allow mapping of individual buildings). Perhaps adding landuse=residential to the surrounding private gardens if someone really wants to be precise.

3 - I’d view this as OK but perhaps unnecessarily complicated. For me it is fine to include highway=residential and small pieces of grass/park within residential landuse, so if I was mapping a new development myself I would map a simpler outline. But I wouldn’t change it if I found it already mapped with this style, provided the excluded parts have an identified landuse as they do here (implied in the case of osm.wiki/Tag:highway=, explicit in the case of landuse=grass).

Having grown up in Ireland, this all seems quite normal to me. The houses on our road were numbered 1, 2, 3, no number, no number, no number, 4, 5, 6. The ones with numbers had been built at the same time, after the unnumbered ones. I think the numbers were really just plot numbers used by the builder, but my parents and a few neighbours stuck up plaques with the numbers, so that became the address. Generally the postman would know the names of the residents. In those days it was quite rare to receive a delivery other than through the official postal system. This was in a suburb of Dublin, so it’s not just a rural thing.

So, I wouldn’t expect any particular “logic” except where houses have all been built as part of the same development. Your Collins Park / Mill Street example sounds like it might be a similar situation to my parent’s house - the numbers might originally come from some kind of identifier used by a developer/builder, and the unusually high numbers may have had some logic in that context.

One of the justifications given for introducing Eircode was that 35% of addresses in Ireland are not unique, so don’t expect to be able to map unique addresses using name/number/street/location. We have jumped from probably one of the least precise addressing systems in Europe to one of the most precise in one go, skipping the intermediate step of “official” street addresses.

It don’t find it that difficult to move stops above ways, at least in JOSM. Generally I prefer to add the ways to the relation first, as adding the stops feels easier when the ways are already outlined on the map. But it shouldn’t really matter which way you do it. Perhaps look at using JOSM if your current tool is difficult to work with?

I agree with the previous comment about concentrating on platforms rather than stop positions.

Maybe you already know this, but a lot of these routes are covered by the Transport for Ireland website and app https://www.transportforireland.ie/plan-a-journey/ - certainly route 882 is included. Incidentally they use OpenStreetMap as the background for their map display. I don’t know how reliable their stop locations are for the kind of rural route that doesn’t have signposted stops.

Sometimes there is also useful information on the Local Link websites for the various regions, e.g. https://www.ringalink.ie/bus-services/kilkenny/

I am not suggesting you use these sites directly for mapping - I don’t think the information is available under a licence suitable for OpenStreetMap. But they may help in your own explorations.

(I think you might be referring to the bus stops on Castle Road in Kilkenny? I have taken a few buses in the past there to get me to the starting point of hikes along the River Barrow and the River Nore, and it certainly was confusing at that time due to the lack of signage of the stops and even on the buses themselves. Sounds like it has not changed in the meantime).

Good luck exploring! ~Alan

There are only two racecourses in Northern Ireland so you overall impression was probably fair enough. One of the two is Downpatrick, in the area you were looking at.

I think Jude_2437 is right, it looks like a circular track (probably dirt surface) for exercising or training horses. Here is an example from the Republic of Ireland: way/108239686

The building:part increase in Spain may be due to the ongoing import of data from the national Cadastre. The data has quite a lot of detail about the height of different sections of buildings and thus generates a lot of building:part elements. I don’t know for sure if the provinces you mention were especially active in this process.

Hola Pablo,

A veces maps.me tarda bastante en actualizar sus mapas, así que es normal que haya cambios en la base de datos de OSM que todavía no estén reflejados en maps.me. En alguna pagina de ayuda leí que generalmente lo hacen cada dos o cuatro semanas. Puedes ver la fecha de los datos en la pantalla “About” del app, en el mío dice 18042018, el 18 de abril. Maps.me tiene sus ventajas pero para ver tus cambios en seguida no es lo indicado.

Should Valencia be included? Metrovalencia is one of the larger networks in Spain; https://www.metrovalencia.es/page.php?idioma=_en

@RobJN - it is interesting that you asked for +1s on GitHub, which as far as I can see would require creating a GitHub account. Do you think that might be an example of the kind of accidental forms of exclusion we are (at least partly) talking about here? I don’t mean to pick on your question, I know it is common among active OSM participants to refer people to GitHub pages. GitHub describes itself as a site for software developers, and its pages are full of developer-specific vocabulary like “pull request”. Does reliance on a tool for developers (as part of general communication) send the right message to potential contributors? Or might it be one (admittedly minor) example of the kind of thing that tends to keep OSM participation confined within a relatively non-diverse group?

@-karlos- I believe that until now they used OpenStreetMap for the display map in only a few countries such as South Korea - everywhere else was based on Google Maps. They have probably been using OSM worldwide for some underlying game mechanics (e.g. to identify parks), but until now that was not directly visible in the in-game map.

@SomeoneElse - that is interesting about iD (I don’t use it often). In JOSM if you click the bus icon on the toolbar you only see “Stop Position”, “Platform”, and “Stop Area” options, so right from the start you have to figure out the differences between these three things and wonder on which of them you should put information such as network or operator. In the menus there is a separate “Public Transport (Legacy)” submenu for highway=bus_stop, which again seems to discourage actually using this option.

Does iD provide an option to add route information in some way? I think it may be at the stage of wanting to tie stops to routes that things get tricky for some people. I now think that highway=bus_stop nodes can be added to route relations, but at one stage I had the impression that I would have to use platforms and stop positions for this.

I agree that the wiki should be treated with a degree of caution, but it takes time to develop the confidence and knowledge to do this, and I suspect it is also more difficult to judge how far a given page may be prescriptive v descriptive if you are reading it in something other than your first language.