I’m posting this here because a couple of changesets and mappers have contributed to the current situation, and commenting on one or the other won’t bring everyone together.
To cut a long story short, what seemed to happen was that someone tried to add a new relation but broke some others (it happens). This also may have broken the coastline. A different mapper fixed one of these. A different mapper again then fixed and fixed one relation but broke some more relations. I’ve also made some of the relations valid polygons again (and I don’t think I’ve broken anything, but could be proved wrong on that).
With regard to this specific problem, I’ll comment on each of the relevant changesets suggesting that people get together here to work together to resolve things.
Taking a step back, there absolutely is an argument that some of the 7 relations here and the 66 relations here belong in OpenHistoricalMap rather than OSM. This has been mentioned before.
What seems to be missing is anything to make it easy to move data traced from out of copyright maps into OSM, into OHM. The last post in that thread seems just to suggest that “if you want, you can do all the work that you’ve already done, again!”, which not unnaturally seems not to have had any effect.
I certainly hope this isn’t the takeaway when someone tries to woo contributors away from Google Maps or Waze toward OpenStreetMap. It takes work to migrate content from one project ill-suited for it to another project that’s better-suited for it. In an open-source project, the work needs to be done by volunteers, not The Powers That Be. I’d also suggest that OHM is generally on even less of a deadline than OSM, so if this problem is a matter of urgency for OSM, hopefully the much larger OSM community can recognize this need without sniping from the sidelines. The OHM community will try to help where we can, but you’ll understand that we’re stretched a little thinner.
Regarding tooling, you’re right that there’s currently not much besides JOSM to do the heavy lifting of transferring data. We approached the Rapid team several times over the last couple years and they showed interest in setting up a more user-friendly workflow (limited at first to avoid licensing concerns). Unfortunately, they never followed through before getting defunded. Such is life in historical GIS and the digital humanities.
(@b-unicycling@PluMGMK, in case you or someone you know would be interested in mapping old census and parish boundaries in OHM in a more constructive environment, we can follow up on the forum.)
I suspect that there would actually three bits to a “worked example”:
Identify that sources are compatible (here I suspect they were 100% traced from OOC maps, so that should be both compatible and easy to state)
Do the actual transfer - which may involve ways being added to OHM to support the relations, where a way in OSM had some other source. It would be great if the add to OHM could be done on a category by category basis and “somewhat automated” to avoid having to do too much repeat work.
Identify the usage of the data and help those consumers switch from OSM to OHM.
In addition to the other people mentioned, I believe that @brianh (who may not be here yet) is one of the largest users of the data.
Both of those examples have extremely limited classes of data compared to OSM; I suspect that most people “moving” are doing so because “all the other data of that class” are in OSM already !
Since you were asking about a way to migrate data with a minimum of effort, do you know of an automated tool for determining the sources that contributed to a particular geometry or tag? Manual inspection would probably require more work than taking the sources we already know about and tracing them…
The short answer is “no”, but it probably helps to look at an example (I picked that one mainly because it was the one it was one of the most recent I looked at). Two of the five changesets have source tags on the changeset “Census of Ireland 1911 online and OSM townlands” and “Census of Ireland 1911 online and OSM townlands and 1901 lines” and the relation also has a source of “Census of Ireland 1911 online” and there’s a note “after further research a small assumption has been made for this boundary”.
The townlands are here, and the history of the central one (as an example) is here. Changeset sources for that include “Bing, GSGS and OS sheets 23 and 26 Paper 1937 and 1910”, “SI 13 1986 EDs and DLR Polling scheme 2009 with permission”, “Logainm.ie”, “Bing;note 586001” and “replaced gap”.
Let’s assume that we have permission for the use of all OOC maps and similar OOC sources, and that the small number of “sweat of the brow” contributors give their permission too. GSGS is I think “Geographic Section General Staff [of the War Office]” and is OOC. note 586001 was an anonymous answer but doesn’t affect the topology. The Logainm licence isn’t OSM specific, but unfortunately the linked page from our wiki isn’t there any more. That leaves the Bing licence. That is OSM specific - is that a problem, or have discussions about using Bing-derived features in OHM already happened?
Edit: In addition to Bing I forgot the “SI 13 1986 EDs and DLR Polling scheme 2009 with permission”. Clearly whatever was given as permission would need to be reviewed to make sure it applied to OHM too.
That’s my understanding of this issue, but I’ll double-check with the other advisors to make sure there’s a paper trail or followup, especially since the services named in the license agreement are being sunsetted in the next few months.
Fortunately, this anonymous sentence is ineligible for copyright as a rote statement of fact. We’d cite this statement as a matter of course.
If I’m not mistaken, the Dún Laoghaire–Rathdown County Council publishes the polling districts as boundary descriptions accompanied by reference maps. We’d need to ascertain which was used in the original changeset a decade ago; if it was the maps, we could ask again for permission, or we could verify that the data is consistent with the boundary description and carry on.
Legal boundary descriptions are edicts of government, which under U.S. copyright law are ineligible for copyright protection. (OHM is subject to U.S. copyright law.) OHM mappers tend to map boundaries based on legal descriptions where available, since reference maps tend to suffer from irregularities (a point that I’m currently going back and forth about with the U.S. federal government). Granted, legal descriptions are a luxury that aren’t always available, especially going back beyond a certain time period, but we’d want to rely on them when we can.