In general, to understand why any mess happens, you can try to poke around in the data and the editing "history" of the data. This hopefully will help you to understand, but should at least enable you to find some users who you could ask about it. Maybe you know this stuff already but... Steps to investigate weird data:
(Note: There are various types of situation where the web interface doesn't work well. An alternative is to use JOSM. Display the 'history' panel. Select the object and click the history button to fetch the data into that display) So in this case http://www.openstreetmap.org/browse/changeset/4618366 User Nakor added a bunch of nodes ways and relations with the changset comment "Washtenaw county boundaries". That might not help you decide what to do about the mess, but it might help you understand it. Also you could send a message to Nakor to chat to him about it. Remember to be open minded about people's motivations. Don't accuse people of anything in an initial message, just explain the weird data you've come across and ask them what they think. ... Alternatively... If you have solid local knowledge of admin boundaries in your area. Be bold! Just wipe the messy data and map it the way you know it to be! answered 01 Feb '11, 12:41 Harry Wood 3
Slight postscript to that: "Be bold" works if you understand OSM. If you're a newbie, don't be bold. Be slightly reserved. :) It's very easy to misunderstand an error message and start wiping data which is in fact perfectly valid.
(01 Feb '11, 12:46)
Richard ♦
|
You should describe a little bit more what the problem is.
Well, I am an amateur also, but I do sympathize with your problem There are a lot of boundaries in France which have got rather messy. I have looked at one tiny piece of the mess referenced by you. In particular: Way id=58358759 19 nodes; Data set: FC9FAC; User: [id:293105 name:NHD edits]; ChangeSet id: 4BF4B6; Timestamp: 2010-06-13T19:14:35Z, Version: 4 tags: "source"="TIGER/Line® 2009 County Subdivision Shapefiles (http://www.census.gov/geo/www/tiger/)" "boundary"="administrative" "admin_level"="8" relation referrer: 135130 903166 nodes: 394522563 ....etc.
You maybe could try to send a query to the originator "nhd" to see if there is a response and point to one obvious bit of the mess to ask what has gone wrong. Superficially it looks like they have been inserted by an automatic process from the above URL. Some of the bits of this mighty multipolygon look correct i.e. seems to me that the points might well lie on a city boundary. Other bits seem very unlikely to be correct. So, I have not answered your problem - other than patiently re-working the multipolygon by hand, but at least you got some sympathy!
On reflection over dinner, some of the "administrative boundaries" surround objects such as "Gerald R Ford Presidential library". Could it be related to the ownership boundary of the object rather than administrative? Also, are "Ann Arbor" and "Ann Arbor Township" two separate administrative bodies? Just wondering.
If I remember right, this is one of those really odd situations of enclaves within exclaves within enclaves. I'll see if I can dig out a reference, and add a proper answer.