1ec5
commented
10 months ago Hi, first of all, I’ve noticed your many contributions while working on the “On this day” feature for the OSM Wiki and appreciate the effort you’ve been putting in. I share your frustration about having to remap the world, but I’m not so sure that becoming reliant on OSM data is the best solution.
OSM only maps what exists, this is their policy, I therefore propose to adopt the opposite policy: only map what no longer exists.
I think this is too restrictive. I have (re)mapped many things that still exist, because without that as a frame of reference, it becomes nearly impossible to map many things that no longer exist, or things that have changed during the course of their existence. Railroads change names, roads get realigned, buildings get expanded, and cities annex land. My city has a long and storied tradition of moving buildings on wheels, and it isn’t alone.
Moreover, dates in OSM are less useful to OHM than they may appear at a glance. OSM only accepts start_dates that are somehow verifiable through direct observation – that is, on a sign – not dates from published works or other research. On tourism=artwork, for example, start_date refers to the date when the artwork was created, rather than when it was installed at a given location, because the little placard beside a work of art typically indicates the date of creation. A shop sign may boast that a business has been in operation “since” a particular start_date, even if it started out in a different location.
Remapping what’s in OSM may sound like a daunting challenge, but to me it feels like the early days of OSM, when the map was also a blank slate – or, in the U.S., the expressionistic artwork known as TIGER. Like in those days, we don’t have to fuss over every micromapping detail quite yet, but we can if we want to. Therefore, different mappers have different personal priorities, for example:
- Focus on extant features that have known start dates. This is the easiest approach for someone who is skilled at mapping but less experienced in conducting historical research. They can cover a lot of ground in the present, but this coverage grows sparse as the end user travels back in time.
- Focus on all the features that have ever existed in a specific place. This is the easiest approach for someone who prefers to research local history. They can help us tell a comprehensive history about that place, but coverage of this sort is uneven across geographies.
- Focus on all the features that existed at a specific time in the past. This is the easiest approach for someone who likes to digitize old maps. It tends to create a lot of to-do items that require intensive research.
All three approaches are complementary. Personally, I switch between all three based on the sources I have on hand and my mood on a given day. We shouldn’t restrict contributors to one method that they find less engaging or more difficult. Instead, we should make clear that all three are welcome in OHM.
either we map what no longer exists but we obtain a mostly empty rendering, which is currently the case, or else we must then map everything, which is not humanly feasible.
I don’t think these two options are mutually exclusive. The default OHM style often looks barren, but that’s because it focuses on the data that’s in OHM proper. It’s just like how osm-carto has chosen not to incorporate Wikidata’s translations of place names, even though OSM has decided that most translations and transliterations should go there rather than OSM proper. The technology exists to create a mashup of OSM and OHM within certain parameters. With vector tiles, there are even more possibilities. For example, you could take the official OHM vector tile layer, filter out extant features (anything that lacks an end_date), and overlay it atop an OSM basemap, styling everything with dotted lines and translucent fills, color-coded by time period. Maybe a mashup like this could be added to the OHM website, just to show that open data projects don’t exist in a vacuum.
Hi,
I have been a contributor to OSM and OHM for some time now, mainly in Belgium and the North of France.
I mainly work on old railway and tram lines which sometimes involves having to re-create entire neighborhoods but this raises a problem, either we map what no longer exists but we obtain a mostly empty rendering, which is currently the case, or else we must then map everything, which is not humanly feasible.
OSM only maps what exists, this is their policy, I therefore propose to adopt the opposite policy: only map what no longer exists.
The databases for the two projects would remain separate but the software creating the rendering layers would use both databases.
The use of multilinestrings which we have already talked about a lot could help with this, I take the example of the N8 road in Belgium, it only took its current ref "N8" in 1986, the OSM objects would be tagged start_date=1986, we then copy and paste these lines into OHM, delete all the tags and use a multilinestrings relationship for the old situations, having done in 10 minutes work that would have required hours. The same process can be used with multipolygons.