Accidental vandalism by KT47


KT47 is a new user who have accidentally deleted a bunch of highways in Thailand.

He have 43 changesets, all created within the last 24 hours, each changeset contains multiple pages of changes. All contain new additions to the map, some contain deletions of important highways.

Before you all go crazy on him, please be aware that this was a mistake. I have already been in touch with him, and the deletions are a mistake - he is not sure how it happened, and he will make no further edits until we figure it out.

I don’t know how to go through 43 changesets with multiple pages of edits to find all the deletions, so I have decided to contact the Data Working Group for help.

It seems that when the highways are deleted, the nodes are left behind. Please don’t delete them or try to connect them. The less we touch this data, the easier it is for automated tools to revert it. Lets see what the DWG have to say first.

Here is a list of his changesets:

Here is an example of a changeset with deletions:

Here is an example of a deleted highway:

Also if someone have experience with JOSM and an idea about how this could happen, please post it here or let KT47 know. He think that it could be related to how he uses the validation tool.

I saw that and also (politely) contacted him and got this reply:

Hi Tom,

I’m currently working with the data working group to sort out these edits.


It’s not clear from Digital Globe’s web site that he had license to use their data. Also, it’s not very good. Many of the little roads either don’t exist or are just cow tracks. A track in a stadium is shown as a res road. So even if DG’s data is available, I don’t think it should be used unless it is carefully edited.

Johnny, please let us know when reversions are completed. I had a similar experience where someone deleted and redrew a whole city. Only a small portion of the data could be recovered.

Regards, Tom

See this:

TBH, I am suspicious of vandalism.

What, how to contact the working group about a revert?

I sent Data Working an email re. these 2 users with links to these 2 threads.

I will not make any ‘repair’ edits for now…

This is the message I sent to the guy:

Hi. I noticed you’ve been adding in Residential Roads, Chiang Rai. I recently mapped this area by survey on motorbike. This is a very isolated, rural area - what you have marked as Residential Road is not - it is dirt track or paths. It is not a good idea to draw traces from satelite photos in such areas without knowledge of the area. I live here - I ride and map the area by offroad motorbike. Please do not edit any tracks or paths you can see as marked as source ‘GPS’ around Chiang Rai - or add in residential or other road types in isolated areas - unless you have been there to physically survey. Much of the area around Chiang Rai is very undeveloped. Thanks.

I haven’t heard anything from the DWG yet, I’ll see if I can find a way to get some focus on this.

My mistake, I wrote to instead of I have resent my email.

Similar users:
MGL25 was mediatrixlopez
C-Son was CyrusWesson

‘Thailand Road Geometry’ again… Definitely dodgy now.

I am still adding names to the above list, and they ALL have deletions.

Multiple users with the same m.o. - this does look more like organised vandalism.

Some of those users are editing now - or a few minutes ago.

A member of the DWG just confirmed that they are discussing the issue now.

(some of this was said earlier on #osm on IRC - apologies if anyone sees it twice)

The first thing that I’d suggest in cases of problematical edits is to comment on at least one changeset for each user. Here they appear to be engaged on the same task (based on the types of edits and the changeset comments) but it’s possible that they’re not physically close together or actually communicating with each other, so in the absense of a message to each one might be unaware that there is even an issue. I’d also suggest using those comments to try and “draw the mappers concerned into the conversation” - don’t just say “X is wrong”, ask questions about it - What was the source? What licence was it under? What do the changeset comments mean? - that sort of thing. Additionally I’d draw each mapper’s attention to this forum and suggest that they post in here to introduce themselves to the community. The advantage of communicating via changeset discussion comment and this forum rather than private message is that messages are public; where there are multiple accounts involved it’s easier to communicate that way.

Once they’ve been contacted and have had a chance to reply we can then decide where to go from there. I notice that a few of the accounts are already reverting their own edits. Based on a small sample, their “editing daytime” seems to correspond to the American west coast, so I guess we’ll have to wait for 16 hours or so to see what happens next.

  • Andy (SomeoneElse) from the Data Working Group

I have created a new thread that I’ll be pointing users to in their changeset comments. Please let me know if you have any additions or suggestions to the first message - I’d be happy to update it.

Just to point this particular changeset was reverted around 7:00 local time.

Curious and checked user’s edits. There are changeset comments in the lines of “Egypt Road Geometry”…

Hey all , I’m Sadi, a product manager working on location infrastructure at Facebook and wanted to address the issues raised in this thread.

My team is focused on improving information about real-world places to help people on Facebook. We use OSM on Facebook in some countries to support that, but as we become more active with OSM we’re still learning how we can best contribute to the community

In the last few days, my colleagues working on Thailand accidentally made the edits that Johnny Carlsen flagged above. We have stopped making all edits until we fix our processes and tooling, and will help the data-working group revert these accidental changes as quickly as possible. We’ve reached out to them over email.

I welcome any feedback and input on how we can best support the OSM community, and look forward to working closely with all of you

Hi Sadi,

Thank you very much for the feedback, this answers a lot of questions.

As you can see in this thread there is a few things that concerned us about your edits:

  1. Deletion of major highways and other existing minor roads.
  2. The source of your edits (DigitalGlobe), do you have the rights permissions?
  3. Changing the type of roads from path/track to residential where we have already done on the ground surveys.
  4. The quality of the edits. All the additions to the map are residential roads, whether or not what is in fact on the ground (if anything at all).

If you are going to help us improve the map of Thailand, we would prioritize quality over quantity. Besides the obvious mistakes of deleting roads, the quality of your additions are still very low. Roads are quickly traced (low precision), they are added from anything on the imagery that could look a bit like a road (streams, tree lines, hedges, very minor paths), not a lot of critical thinking seems to go into the edits - I am guessing it is all about speed. There are still many big roads that are easily identifiable missing, I’d suggest you start by adding all these big and obvious roads and maybe leave the smaller ones that can be difficult to identify correctly from your imagery. It is possible that the roads you are less certain about could be added as highway=road, which means we on the ground should go and verify.

Now I can’t speak for the whole community, these are just my thoughts and comments. I invite others to join in.

TBH, in isolated, rural areas (quite a lot of Thailand), if uncertain then better to add no more than track - if anything at all. Believe me, a lot of this stuff is often not easy to access re. physical surveying - and the number of actual surveyors/mappers = very few!

FYI: I am starting to receive messages from the DWG that they are reverting edits.

Thanks for the detailed feedback Johnny, really appreciate it, and totally agree with you on quality.

We’re working with the Data-Working-Group on reverting these edits for now, as we work hard to improve our tools & processes to ensure high quality contributions.