I'm guessing that way/815289737/history was an aircraft at some point (not a fence) but it doesn't seem to be on any imagery currently available to OSM. What was it?

Re "And for some reason you are fighting only with me.":
I'm only dealing with this because firstly, OSM's Data Working Group received specific reports about this problem, and they all mentioned you by name.
Secondly, I actually did have a look at other additions of "tracktype" in the mountains (you can see my changeset comments) - the person there said they'd been hiking locally, and inferred tracktype from that local knowledge.

Re https://map.land.gov.ua/?cc=2657643.816380124,6197966.900528279&z=16&l=kadastr&bl=ortho10k_all , (a) what is the licence associated with that site and the imagery displayed on that site and (b) what does it prove anyway? The fact that there was a break in the trees an unknown number of years ago does not prove the existance of any sort of track or path that is suitable for OSM.

For completeness, this revert doesn't mean that the DWG has taken a view on what the "correct" boundaries are here. As I said in osm.org/user_blocks/5487 , there's definitely a case to be made for Somaliland operating "on the ground" as a country, but that is a case that needs to be made to the rest of the OSM community.
- Andy (from the DWG)

Re the licence, just to transfer some comments from Slack, so that everyone can read them.
Someone mentioned https://github.com/facebookmicrosites/Open-Mapping-At-Facebook/wiki/Esri-ArcGIS-FAQ , and someone else mentioned osm.wiki/Import/United_States_Addresses .

https://github.com/facebookmicrosites/Open-Mapping-At-Facebook/wiki/Esri-ArcGIS-FAQ links to https://github.com/facebookmicrosites/Open-Mapping-At-Facebook/wiki/Esri-ArcGIS-FAQ which doesn't mention New Jersey, so per the second comment I guess this must have come from osm.wiki/Import/United_States_Addresses . Confusingly the licence for that osm.wiki/File:Usdot_nad_license.png says in the text "this is CC0" but the page links to https://creativecommons.org/licenses/by/4.0/ - I'm guessing that that last link is a cockup!

Hello,
I appreciate you've spent time tidying up the stuff that was imported here, but I'm concerned that as we don't know the source of the original data, we have no way of knowing whether it is licence-compatible with OSM (all attempts at communicating with the original importer failed).
Some of it is obviously still not useful as it stands (way/1007307802 does not connect to anything), but at least the "here is a footway" is verifiable from imagery in a way that house numbers and postcodes aren't.
Best Regards,
Andy

I'm just concerned that you're spending a lot of time editing things that I'm probably going to have to delete shortly - I don't want you to waste your time.

From another quick sample, way/962057588 is another one that looks like it just doesn't exist.

You've removed tracktype here, but way/999523559 and way/971767658 don't appear to exist at all (other than "someone could drive round a field here").

Looking on the available imagery, way/1006191185 does not look like a track at all. Likewise, relation/6093419 does not look at all like allotments.

Apologies about the size - these changes are all in the USA apart from 1 changeset in India (which I'll look at separately)

Hello Jeet Dev,
As you can see from node/9182612241/history you have removed lots of keys from this place. Can you explain what the problem with them was?
Also, can you please link to the data source that you have referred to in your changeset comment here?
Best Regards,
Andy

Hello again,
No reply, so I've filled in the gap here in changeset/114394801 , after talking to the person who originally added it on IRC. I've also re-added a stile where the old one node/2698217399 was.
Best Regards,
Andy

Hello again VARVAR8,
In addition to the things previously mentioned, could you please try and use changeset comments that explain to other mappers what changes you're making? Here you've used a bunch of hashtags and then just "challenge_19281" which doesn't really say anything. See osm.wiki/Good_changeset_comments for details.
Best Regards,
Andy

Hello Bharat Patil,
The fact that you're adding 10,000 nodes at a time looks very much like an import. These must follow osm.wiki/Import/Guidelines , which includes discussing it with the community beforehand. Also, you'll need to explain a bit more about your edit, not just one imagery source. See osm.wiki/Good_changeset_comments . For example, what plan for conflation did you have?
Best Regards,
Andy

Just for info, I've mentioned this changeset in a diary comment: @martin-kokos/diary/398149

Hello VARVAR8,
Just a reminder that the questions that I linked to in changeset/107276082 .
As I said there, it's looking increasingly likely that we'll have no option to revert your MapWithAI additions, and for information the DWG has had another request this morning to delete them.
Best Regards,
Andy

Where I've seen similar issues previously is around Marble Arch: relation/5640188 . There the data wasn't invalid and I had to explicitly exclude that type of junction from processing. Here the tags on the multipolygon members just didn't make a lot of sense, in addition to the geometry being unlikely. I left it as a multipolygon with one member to make the history easier to see, but arguably that should be tidied now. I saved the previous state at https://gist.github.com/SomeoneElseOSM/d3e8a4951b013cb8cbeeaf77706615ba . With osm2pgsql 0.96.0 the order of the ways in the relation relation/5660722/history had an effect on which appeared and which didn't. Note that even the "non-road" in that relation had access tags: way/380116486/history . Is there an easy overpass way to find "multipolygon relation with certain tags with constituent ways with certain tags"? It might be useful to look for more.

Previously part of Tower Bridge Road and two parts of Decima Street were part of the multipolygon. This didn't look right - most importantly because the imagery shows Tower Bridge Road as quite wide here and the pedestrian area much narrower than previously drawn.
Also the packaging of the outside roads into a multipolygon meant as far as I can see they were no longer valid geometrical objects in their own right - I was seeing one (arguably correctly) not shown and one (arguably in error) appear.
What's missing now I suspect is the "logical linear footway" through the pedestrian area, and where it connects to the sidewalk (or perhaps a crossing) to the south, and the connections to the crossings at east and northeast.

> But my decision is partly based on experience gained when hiking in this locality.
That's great! If you're familiar with the local conditions then yes - you'll have a pretty good idea what the tracktype is.