On Wed, Apr 24, 2019 at 12:45 PM Christoph Hormann <o...@imagico.de> wrote: > The problem is that these two visions of OSM are mutually exclusive. If > you want OSM to focus on cooperatively assembling local knowledge you > have to accept that the only way you are going to fill blank spots in > the map is by motivating people to acquire local knowledge of these > areas and sharing it in OSM and that imports can only work in a support > role for craft mapping and not to fill in blank spots.
They're mutually exclusive only if you draw an artificial hard line between them. I challenged Frederik with drawing the line between 'local knowledge - good!' and 'import - bad!' on the following spectrum. 1. I go out, walk the trails of a preserve, make field notes, collate my results, upload to OSM. Conventional. 2. My GPS happens to go wonky, but Jon also happened to be along on the trip, and had collected tracks redundantly, so I do (1), only with Jon's tracks. 3. Jon likes the idea, but doesn't want yet to bother learning the OSM tools, so he goes out on a different weekend and collects tracks and notes. I collate and upload them.(I try to convince him to learn to do this for himself, but he's not ready yet.) 4. Jean-Claude has no Internet access because he lives way out on the sticks in an area that doesn't even have cell coverage, and his copper phone line won't support fast DSL. He does own a Garmin, and has sitting on it the tracks from many trips. He also has extensive local knowledge, can point to his tracks on the maps and give me trail names and characteristics. I copy his tracks, make notes from his descriptions, and do the work that he can't because he's hampered by a lack of technological infrastructure, and of knowledge - he's grown up almost entirely computer-illiterate, and has little incentive to become any sort of technology geek because in his area there are no geek jobs. 5. Ranger Judith grabs a bunch of GPS tracks of trails during an inspection tour. I ask if I can borrow her tracks to fill in gaps, and she says, 'sure' and e-mails me a bunch of GPX's. I follow the same procedures as above. 6. Judith and her colleagues have decided to pool the information that they've gained in their tours, and someone has put together a shapefile with trail alignments. They volunteer to share it with me, I do checks to make sure that topology is consistent, adjust tagging, conflate with OSM, and import. 7. Judith's agency decides to compile the alignments of all the trails they manage, using the data captured by groups of rangers on their inspection tours. They publish a shapefile and give OSM permission to use it. I discuss the issue on 'imports,' Wikify the import, import samples for review, and then do a full import, again with careful conflation and redaction. Frederik's reply, if I recall correctly, was that everything but (1) was working against the vision of OSM, and damaging the community by importing outside data. To me, that sounded as if the true objective is that OSM should be the only medium over which it is permissible to share local knowledge, and that knowledge acquired other than by the personal eyeballs and instruments of the single individual doing the mapping is already tainted. The argument presented was essentially that if you import the local knowledge of anyone but yourself, you've damaged the community by (a) failing to recruit the person who collected the information as a mapper and (b) destroying a blank area on the map that would be a potential opportunity for a future mapper. In my view, I'd rather adopt a broader definition of 'mapper' - that the person with the boots on the ground collecting the data, the person who is collating, conflating, and redacting, and the person who is driving a graphical user interface and pushing data into OSM don't all have to be the same individual for OSM to be a cooperative assemblage of local knowledge. If my bad foot is acting up, and I can't get out with boots on the ground for a few months because I'm hobbling about with a cane, the guy who is out there gathering data is every bit a mapper, even if he gives me data in a raw form and I have to tidy it up for OSM. And I'm still a mapper, by doing so. By the same token, external data sources are assemblages of someone else's local knowledge that they were gracious enough to share with the community. By far the majority of sources that I've examined are of too low quality to contemplate importing wholesale, or are of features that are too hard to conflate, or are fundamentally organized differently from OSM (pixel-based landcover data, for instance). I don't import them. In fact, the only things that I've actually imported are that I use address point data to assign street addresses to buildings that I or someone else has already traced, and data giving administrative boundaries of parks, forests, nature reserves and government facilities. And generally my motivation for importing the boundaries was not to fill in a blank area in OSM, but that by importing them, I had an opportunity to open them up for collaborative repair! As an example, I've spent quite a few hours working on conflation for the myriad of islands and lakeshores in https://www.openstreetmap.org/relation/6362702 - the job is by no means done, but it's one heck of a lot better than it was. And not all the edits were mine! Others have carried on the work, even if the most recent edit was from me, revising the import to add some newly-purchased land. (Actually, I may be lying there - but I know that there are complex reserves like that one where others *have* taken up the baton.) I've actually done a little bit of correction even of these boundaries from field data. One friend of mine is a volunteer boundary monitor for the Appalachian Trail, and makes periodic inspection trips to ensure that the boundary markers for the Trail corridor are still in place and that adjoining landowners are not encroaching with construction, fences, and the like. He's shared data with me, and I've been happy to adjust the mapped boundary to conform with the stakes, blazed trees, and witness posts in the field, rather than the (badly) digitized data in the government's data set. I don't have a pipeline to push this information back to the government, but at least I can share it on OSM (and then I have to import the rest of that corridor section to keep the topology consistent). My anecdotal experience is that these administrative imports have triggered mappers to do useful work. I imported (with permission!) the boundary of Black Rock Forest https://www.openstreetmap.org/relation/7083920 - largely because there were bad topological problems with the West Point military reservation that I was correcting at the same time. At the same time, I dropped a note on OSM remarking that the forest had a trail network that was entirely unmapped. In all the time that OSM had been in existence, that had been a blank spot. Apparently the note and boundary served as some sort of trigger, because within a few months, not only the trails, but also the ponds, streams, and hills had appeared in OSM by the hands of others. > The irony of this is that you are telling this to someone who has done > 90 percent of his mapping work in areas far more remote and wild than > any place you have most likely considered to do an import in. So > frankly i have very little understanding for your special position > here. Have you considered that one of the reasons why the active > mapping community in Canada is so small is that you are discouraging > potential craft mappers (in particular also ones from abroad) with the > imports? I'm also a wilderness mapper - and have had a couple of trips that went in at least 30 km from the nearest drivable road or permanent habitation. About all I was able to gather on those trips was the alignment of trails and the location of points-of-interest; I surely didn't have time to go slogging through the quicksand on the shores of the ponds to map them, or characterize the bog-to-meadow-to-woodland transition of the innumerable peat lands, given that my stay was limited by the supplies I could carry. I was mapping trail for which no satisfactory map existed - and there are a bunch of trails up there that still fall into that category. I don't believe that importing the boundaries of the wilderness areas in any way discouraged others from mapping their content - and in fact, it may be encouraging it. (I don't know if the trail mapping that I'm seeing is an unrelated trend, or partly triggered by having a boundary of something to fill in.) > Anyway - lets not loose track of the subject here. I made kind of a > fundamental statement here in response to Peda's suggestion for > adjusting the rules for imports but i don't think this should > overshadow the specific discussion of the import in Sweden. >https://www.openstreetmap.org/relation/7083920 I mostly agree with you about the specific situation with the Swedish import, but note that the importer stopped it without prompting upon noticing significant issues with the quality of the imported data. Personally, I don't import landcover, ever. I use external sources of landcover information, to produce maps like http://kbk.is-a-geek.net/catskills/test4.html?la=42.2856&lo=-74.1710&z=14 . There's no need to have all that in OSM, and the pixel-based data doesn't match our format. If there is landcover information in OSM, I put that on a layer above the government landcover data set, so that it overrides whatever the government said. (Exception - I don't overlay natural=wood, landuse=forest, or landcover=trees unless there is leaf cycle information, because in my part of the world, the deciduous forest is relatively open, while the pine swamps and balsam-and-spruce Krummholz are dense and rough going.) In this way, I get for my rendering the best of both worlds: I have some sort of landcover information everywhere in the US, but if a mapper has gone to the trouble of mapping the detail of where the bald hilltops are, as at https://www.openstreetmap.org/#map=15/41.2800/-74.0072, I can show that, too. Not everything that a data consumer renders has to come from OSM! _______________________________________________ Imports mailing list Imports@openstreetmap.org https://lists.openstreetmap.org/listinfo/imports