Re: [talk-au] Tagging towns by relative importance, not just population size
Good conversation, thanks, everyone! On Thu, 28 Sept 2023 at 20:04, Andrew Davidson wrote: > > Err downtown Rathdowney has a population of 161.I might be OK with > village, but it's a bit of a stretch to call it a town. > I was looking at https://profile.id.com.au/scenic-rim/population?WebID=160, but https://www.abs.gov.au/census/find-census-data/quickstats/2021/SAL32394 says 320 ? I guess that's "town" vs area? Just reading the wiki on it, & it mentioned showgrounds & Post Office. What do the presence of them do to the "relative importance" scale? Thanks Graeme ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 28/09/2023 11:18, Andrew Davidson wrote: On 28/9/23 08:21, cleary wrote: Windorah Qld and Ivanhoe NSW are both currently shown as "town" in OSM but neither has more than rudimentary health service (if any), a hotel, small primary school and service station. I couldn't buy a coffee in either place last time I visited. That is the general problem, most people want to inflate the importance of a place so that it renders. Windorah has a population of 76 and Ivanhoe 202. If it's lucky Ivanhoe might rate a village but Windorah is most firmly in the hamlet class. Perhaps this apocryphal Ireland solution should be used? :-) A house - building A house and a church - hamlet A house, a church and a pub - village A house, a church and two pubs - town ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 28/09/2023 10:19, Warin wrote: On 28/9/23 17:04, Michael Collinson wrote: TL;DR: We need to get a systematic measure of population density into OSM to act as a guideline for mapping software to vary what goes at what zoom level. Off topic: On a global scale that does not work due to the population densities changing over the world. When adjusted for Europe to have a 'good map' then using the same software rules the map goes blank in various places like central Australia. Yes, agreed, it has to be regional/local. That is thrust of the essay, so I'd reword the summary as: TL;DR: We need to get systematic measures of regional/local population density into OSM to act as a guideline for mapping software to vary what goes at what zoom level. That can be done in the existing db by attaching a tag to admin boundary relations. The drawback is that it needs to be done at at least a sub-state level to accommodate, say, Western Australia minus Perth economic envelope. I personally feel the long-term solution is to be able to define more arbitrary polygons as they can be used for many other metadata use cases. My thinking is the map generating software should fill the map at a zoom with data until the map density reaches a certain level and then stop. This way the map would not be blank nor over crowed, but what is displayed adjusts to suit the data available. There could be limits on what detail could be displayed in both directions - minimum data and maximum data but what it uses is simply between the two limits and adjusted for data/map density ... Of course there is a lot more to this .. like the tiles being sized to suit the data density rather than an arbitrary lat/long size. Yes, another good idea. Potentially practical at "big iron" level, such as commercial server solutions like Mapbox or the OSM server itself where you have processing power and memory. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 28/9/23 17:04, Michael Collinson wrote: So, I think some sort agreed national level hierarchy of populated place is important in order to jive with cultural, legal, cultural and broad population density criteria. But to vary it locally or regionally is dangerous and I agree with cleary (if I am reading the quote levels right). I'm in agreement. The current tagging guidelines are already too vague. I don't want to add on the idea that you can vary it across Australia, this will just encourage more place inflation. I'd rather tighten up the definitions, so they are more verifiable. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 28/9/23 09:08, Graeme Fitzpatrick wrote: Against what you said, Rathdowney in SEQ, with ~1800 people, Err downtown Rathdowney has a population of 161.I might be OK with village, but it's a bit of a stretch to call it a town. But Maroon, 20k the other way, with only a primary school & a RFS station, would only be a village. Looking at the aerial imagery I'm not sure this would even count a a settlement. There is no clustering of dwellings, it's just farms strung out along a road with a school. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 28/9/23 08:21, cleary wrote: Windorah Qld and Ivanhoe NSW are both currently shown as "town" in OSM but neither has more than rudimentary health service (if any), a hotel, small primary school and service station. I couldn't buy a coffee in either place last time I visited. That is the general problem, most people want to inflate the importance of a place so that it renders. Windorah has a population of 76 and Ivanhoe 202. If it's lucky Ivanhoe might rate a village but Windorah is most firmly in the hamlet class. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 27/9/23 16:29, Ian Sergeant wrote: Aren't most places classified by the government authority as cities/villages/towns/localities/suburbs? Not in a way that is useful for using in OSM.They tend to be classified under the state's local government act, which is an administrative arrangement not an indication of where they would fall in the OSM tagging system. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
On 28/9/23 17:04, Michael Collinson wrote: TL;DR: We need to get a systematic measure of population density into OSM to act as a guideline for mapping software to vary what goes at what zoom level. Off topic: On a global scale that does not work due to the population densities changing over the world. When adjusted for Europe to have a 'good map' then using the same software rules the map goes blank in various places like central Australia. My thinking is the map generating software should fill the map at a zoom with data until the map density reaches a certain level and then stop. This way the map would not be blank nor over crowed, but what is displayed adjusts to suit the data available. There could be limits on what detail could be displayed in both directions - minimum data and maximum data but what it uses is simply between the two limits and adjusted for data/map density ... Of course there is a lot more to this .. like the tiles being sized to suit the data density rather than an arbitrary lat/long size. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] Tagging towns by relative importance, not just population size
TL;DR: We need to get a systematic measure of population density into OSM to act as a guideline for mapping software to vary what goes at what zoom level. This can be done either by adding the appropriately calculated/derived density measure to admin boundary relations or, more radically, as part of a separate Metabase where more arbitrary polygons are allowed. And now a bit of an essay: For me, population size is the only meaningful indicator of relative importance as it is quantitative albeit fuzzy (to me it doesn't really matter whether it is for an admin area, urban envelope, metropolitan area, whatever - if anything more rigorous is desired, use specialist tagging). But the rub is the word "relative". Relative to what? When I mapped Dalby, Queensland in 2006/2007 the ABS population was below 10,000 which in the then Brit-centric guidelines made it a village, which is ridiculous given the importance of the town within the area. So, I think some sort agreed national level hierarchy of populated place is important in order to jive with cultural, legal, cultural and broad population density criteria. But to vary it locally or regionally is dangerous and I agree with cleary (if I am reading the quote levels right). Graeme then says: > ... but it would be good to do something that fixes the vast empty when you cross the Great Dividing Range. Yes. I think this is a map presentation issue not a map data issue. I have a series of Android hobby apps published for specific areas and the way I resolved it was to simply have a "low zoom" flag in some of them which tells the map style sheet to show farms, hamlets and villages at much lower zoom levels and with greater prominence at higher levels. place=locality can also be a good one to pick out as can landuse or even buildings. For Australia, specifically, my tip would be to systematically tag main farm building(s) as place=farm or derive it from named landuse=farmyard. But that raises another question. Is there a generic way to generate a "low zoom" flag? There are at least two possible solutions. The first is to use the existing OSM data structure. Calculate or derive (ABS??) population density for administrative areas and put it on boundary relations, national, state and "local". It is then up to the mapping software to see what is available and make zoom-level detailing decision based on it. This is doable but makes things hard for the small mapmaker like me to implement. More democratic is to use a notion proposed by Sarah Houseman for geocoding and I believe has much wider implication and is an important step forward for OSM. I floated this at 2020 or 2021 SOTM. This is to have a separate "metadata" database of polygons with, following OSM practice, whatever you like attached to them. The point of the polygons is that you can attach rules and hints to them. They can follow legal jurisdiction boundaries or can be more general. As an example "All of Western Australia except the Perth metropolitan envelope". Or, outside this discussion, "the area where is the main spoken language". Here are the three main areas that I propose. (3) is relevant to this discussion. (1) Rules. In the NSW polygon, bicycle=no where footpath=sidewalk except for children under 16. In the Australia polygon, driving is to the left. (2) Default hints. In the YYY polygon, surface=unpaved/paved where highway=primary and surface tag not defined. (3) Hints. Population density. Main spoken language(s). How addresses are structured. ... and anything else that could be useful for mapping, searching or routing in this area. [Having such an open data, systematically structured database removes a danger that map making moves back into the realm of companies with deep pockets because only they have the resources to 1) collate the data, 2) be able utilise it on the fly when presenting maps, routing, searching based on OSM data.] Mike On 28/09/2023 04:04, Graeme Fitzpatrick wrote: On Thu, 28 Sept 2023 at 11:25, cleary wrote: All valid arguments, thanks. If everything is exaggerated so that villages are described as towns and towns as cities etc., then I think it just devalues the whole database on which the map is based. I certainly see where you're coming from, but it would be good to do something that fixes the vast empty when you cross the Great Dividing Range: https://www.openstreetmap.org/#map=7/-27.163/145.569 I'll be interested to read comments from other mappers. So would I, but so far there's apparently not too many interested in it? Thanks Graeme On Thu, 28 Sep 2023, at 9:08 AM, Graeme Fitzpatrick wrote: > Thanks! > > Yes, it probably shouldn't be a one size fits all equation. > > Against what you said, Rathdowney in SEQ, with ~1800 people, only has a > cafe / takeaway / store with a few grocery items, pub, currently closed