OpenStreetMap
The process of improving Ohio's county boundaries has been ongoing for months. The work has been done mostly by myself and one other editor, using Potlatch.
Well, I recently tackled the boundaries in JOSM, and after a couple of nights of work, all the overlapping boundary ways have been merged, and proper relations created for each county.
The TIGER data in adjacent counties still needs to be properly joined over much of the state. I'll probably stick to Potlatch for that task. I can also improve the boundary accuracy that way. But now it will be easier to work with the boundaries as they are, with much fewer situations of half-finished counties with doubled boundaries that confuse other mappers. I've had people screw up my work because of that, but no more!
So yeah, the wiki says that JOSM is best for concentrated editing in a small area, but I find it's quite useful for manipulating very large structures when you don't want to load every little street in the region. The tricky part is getting JOSM to only load what you're interested in...
Discussion
Comment from Andy Allan on 14 May 2010 at 10:12
Good work on the TIGER fixup - it's a long, hard and inglorious task!
Comment from JohnSmith on 14 May 2010 at 17:35
The other benefit of JOSM is it now has a filter function to make admin boundaries, or any other object for that matter, unclickable or completely hidden, or you can do inverse filtering to only show one type of object.
Comment from Vid the Kid on 14 May 2010 at 20:56
@ JohnSmith:
Yes, I've heard about JOSM's filtering feature. I don't think it would have been very helpful on this task, however, unless I was willing to download the entire state of Ohio. (I was not.) But I do look forward to trying it out in the future.
Comment from orienteerer on 16 May 2010 at 05:28
How do you get JOSM to only load what you want?
Comment from Vid the Kid on 16 May 2010 at 07:57
@ orienteer:
I think I've heard of a way to download just a specific subset of data (perhaps using the xapi) but I don't know if that functionality is built into JOSM. In this case, I started by downloading data for a very small portion of Lake Erie that I knew contained some nodes that were part of existing county boundaries. In that way, I got a couple of the county boundaries downloaded. Since all the objects I was interested in were connected to each other (the ways may have been overlapping, but at least they shared nodes) but to very little other data, I repeatedly selected some nodes, and hit ctrl+alt+D (Download Referrers) to retrieve other ways that use those nodes, until I got all the old USGS boundaries. Some of these were also connected to boundary ways (and relations) that had already been manipulated in Potlatch. Once I got part of one of those, I just told JOSM to download the rest of the relation for that county. Finally, since I intended to merge these with the TIGER-imported Ohio boundary, I told JOSM to download every member of Ohio's boundary relation, selected all those members, then invoked the Download Referrers command again so that JOSM would be aware of other relations using those ways, such as Kentucky and Canada. That way, if I split those ways, the other relevant boundary relations could be updated accordingly.
So yeah, like I said earlier, it's tricky. But in some cases, it's doable. It helps when a relation refers to most of the data you want; I and other contributors have made large-scale adjustments to roads that way.