Actually the source was mostly Maxar, as it seems to be more recent than Bing here

Looking at the aerial, I agree. I have changed that section of the "purgatory trail" to be highway=unclassified, surface=unpaved

dpp, you introduced some major errors in this changeset. For example, you changed "ele=1181' MSL" to "ele=1181". The original is in feet (that's what the ' means) and by removing it, you have changed it to meters, without doing the feet-meters conversion.

In the future, be very careful about pressing 'Q' by accident, and please review your changes carefully before uploading!

Ricky, it appears that in this changeset, you selected a political boundary and accidentally "squared" it ("orthogonalize shape") which completely messed up the geometry. I would attempt to undo your changeset, but it seems you touched 607 ways and 4138 nodes, so you changed a lot beside this boundary way. I will attempt to revert just this way, to fix the damage.

Alifatri, I don't know why you did this, adding hundreds of inefficient, colinear "garbage" nodes to this way, but PLEASE stop. Perhaps you are using the FastDraw plugin and need to learn how to use its simplify feature? I hope you did not click a mouse 1210 times to make this way? You used 1210 nodes to draw a way which is perfectly modeled, with even more precision than the imagery, with less than 400 nodes.

Bonya, there are major issues with these streams:
1. You are using 4x too many nodes! Run simplifly before uploading, so we don't have to do it for you.
2. You used the wrong tag: these are clearly stream, not river.
3. Most of the stream directions are obviously wrong (uphill).

I am trying to fix it all, but please stop making so much cleanup work for us to do.

These are not tertiary roads! Only a substantially improved, wide road that leads to a significant settlement is a tertiary. Otherwise it is unclassified, or residential, or track. Please fix the 17000 "tertiary" roads you have created in Myanmar. If you need help, let us know.

Maximus, the ways in this upload are badly overnoded, e.g. streams with 480 nodes which are fully represented with only 180 nodes. It saves the OSM database a lot of inefficient churn if you trim out the noise _before_ you upload; otherwise, we just have to come along afterwards and delete it. Let me know if you need any help on how to do this.

Actually, this is NHD in Minnesota, just across the border (and sharing an edge with) the Canvec in Ontario, hence the mixup.

Hi Ikiya. I'm glad to hear that local mappers are paying attention! That is very good.

I can further explain the edits here in English, and you can forward them to the talk-ja mailing list? I do this kind of standard OSM cleanup for dozens of countries around the world and there is nothing Japan-specific about them.

For example: "Shikoku, tidy coastline to 60cm". In this case, a mapping account named "Laharl" did a raw import of coastline data, with source given as "GSI/KIBAN Block". However, this import was badly overnoded in some places, with hundreds of redundant and colinear nodes on small lengths of simple coastline. To fix the problem, in JOSM one can set the advanced property simplify-way.max-error to a very small value like 0.6 (60 cm). This removes the noise and redundancy, without moving the coastline at all cartographically.

The political boundaries are similar; they are legal entities which are not actually sub-meter precise, so the sub-meter node bloat is due entirely to GIS processing artifacts. These are easily removed with a simplify-way:max-error of a meter, although I used only 80cm in this case to be even more gentle and minimal change.

Please see these diary posts for more information, about Japan (@bdiscoe/diary/43428) and the general problem of overnoding (@bdiscoe/diary/37421 and @bdiscoe/diary/37473)

Omino, since you haven't responded in a while, i went ahead and changed your "dam" features to reflect what is actually present in relatively recent imagery (DigitalGlobe Premium is more recent than Bing, here)

Omino, could you explain what this is? You added a "water control area" and "gogo dam" water area, but this is downstream from where the actual and expanding gogo dam is. No changes appear here, even in very recent imagery. You may want to update or delete these non-existent "water" features?

dougcb68, these changesets contains intensely overnoded ways, at least 3x more nodes than are appropriate for rough natural features like natural waterways and coastlines. Please use "simplify" on your ways before uploaded to save a lot of work for people to come along afterward to delete your nodes! Feel free to ask if you need help knowing how to do this.

Partly used DigitalGlobe standard.

I actually used mostly 'DigitalGlobe Standard' for this changeset, because it was much more recent than Bing imagery, which doesn't even show MEX 24 built here.

Thanks. It can be hard to tell sometimes, but please feel welcome to add access=private to any pools you like.

For what it's worth, the reason I write "clean up" instead of "simplify" in the changeset comments is because I'm doing more than just running simplify; I also use the JOSM validator to find and fix other issues, including broken topology, overlapping ways, places where the import collided with existing data, etc.

Michael, I completely agree. Unfortunately, it's a big planet, and by the time my code discovers an issue, the importer is generally no longer paying close attention. In this case, this data was uploaded by 'huaraz' more than a year ago, in changesets such as changeset/40512747 (In the case of NHD, it can many more years, even back before OSM _had_ changesets!) As mentioned in the comments on the diary posts I shared, there are brief guidelines like osm.wiki/Import/Guidelines#Consider_simplifying It could be worth trying to communicate this to huaraz. Perhaps you could help?

For what it's worth, there are several more million bad nodes here. It is going to take a lot of 10,000 edit changesets to clean up the mess. Alternately, we could convince users (like 'huaraz' who did this part of the Canvec import) to clean up the data before importing it.