That suggests “using deep learning for an area in Thailand” (and we know what happened last time). I’ve suggested on the talk page that any import should be discussed here (at least) before it takes place. Last time the import was (putting it politely) “not exactly announced before it happened” so I’d suggest keeping an eye open for strange new roads added by strange new accounts.
I agree that any proposed additions should be carefully reviewed. The Wiki indicates that this will be the route taken before new data is added and I for one, would love to have the benefit of this good looking aerial imagery. However, a quick reading of the allowable uses causes me to wonder if the license will really work for OSM. The writer seems to think so and has put a considerable amount of effort into the proposal.
Considering paragraphs 1.1 and 1.2 under the heading Data seems to confuse the issue. Also, I wonder who is offering this service? Digital Globe, Facebook, or ??
This could be a huge and valuable addition to OSM in Thailand. Let’s investigate further and keep the list up to date with any information we uncover.
This was our next step after posting the description to the wiki. We have been chatting with a few community members over the last few months gathering feedback and local knowledge of Thailand. We have also shared tiles we edited to gather feedback on quality and tagging, and are happy to do the same for anyone else.
Dave, to answer your question, Facebook is offering this service. We have a team dedicated to OSM ready to answer any questions through this forum, or you can email us directly at firstname.lastname@example.org. We will get back to you within 24 hours.
Thank-you for bringing this to our attention. The team is investigating this now and will get back to you. We have a team dedicated to OSM ready to answer any questions through this forum, or you can email us directly at email@example.com.
The first edit listed above is http://www.openstreetmap.org/changeset/46144791 . The changeset comments there are “#nsroadimport#thailand” which are just hashtag gibberish and in no way meaningful. The account used http://www.openstreetmap.org/user/VLD004 has no information on it - it looks like it has just been created as a throw-away account like a spammer would use. It’s basic politeness when creating accounts to do work for a corporation to explain what the accounts are created for, who the user actually is, and who the corporation is.
To pick up a few points from your post above, “chatting with a few community members over the last few months” is not an appropriate mechanism for informing the community, nor is providing an impersonal email address firstname.lastname@example.org a good idea.
Please STOP all Facebook import activity in Thailand and elsewhere until you have discussed fully what you are doing with the community both on the “imports” mailing list and (about edits in Thailand) here.
Andy Townsend, on behalf of OSM’s Data Working Group.
We share a common goal: to make OSM better with high-quality edits. No edits have been made since last week when this feedback was brought to our attention. As I mentioned in the response, any further edits we make will include changeset comments to state we are adding roads.
We just replied to Frederick with this invitation as well, but given your expertise and experience in OSM, we’d like to speak to you and your OSM Data Working Group partners in more detail about our approach and process. We’re new to this community and want your direct feedback and suggestions. We’ll follow up to schedule a time.
As you say, OSM is a community. We’d expect mappers working for Facebook to be open and honest about what they’re doing and to engage with the community, and being responsive when other mappers ask them about edits that they have made, just like any other member of the community. An individual mapper can see commented changesets in their history list, and third-party tools such as http://resultmaps.neis-one.org/osm-discussion-comments?uid=5314363 also exist to see what questions have been asked about what changesets.
The Data Working Group works on behalf of both the community and the OSMF. Although we may be able to help with procedural points (as here) the primary communication is with the community both internationally (e.g. the “imports” list) and in national forums such as here. The DWG may also of course be involved with removing failed imports (as we did with e.g. some of last year’s Facebook imports in Egypt) and blocking importers who insist on importing without following the rules (which we’ve not had to do yet in this case). Ideally we’d like any imports to be performed according to the guidelines with community approval so that we don’t need to get involved at all, so we look forward both to your “proposal to import” on the imports list and to the replies from your mappers to the questions that they have been asked in changeset discussions.
Andy Townsend, on behalf of OSM’s Data Working Group.
Thanks for the clarification. We appreciate the guidance, especially with the number of lists and guides all over for OSM. We will tackle the questions including yours We will also make sure to follow the changes you mentioned and move forward with the e-mail to the import list and country discussion post.
Editors will also add their company affiliation to their profiles first thing Monday
I obtained a license key and TMS Url from the Facebook group, installed it into JOSM and went exploring. I was disappointed to learn that while the members of the FB group have access to high quality satellite imagery from Digital Globe (DG), ordinary OSM users do not. The TMS imagery we are allowed to use is what’s been derived after their post-processing of that high quality DG data. See the sample (I obtained the link from their Import page in the Wiki) image below for a look at the layer we get to use:
What this means is that we must use our available satellite imagery, either Bing or Mapbox, and alter its transparency such that their derived layer, placed beneath the Bing or Mapbox layer, is visible through it. In other words, it might be a small help in some areas but mostly you will have to rely on the usual satellite imagery to do your work. There is no way the layer they provide us can help determine if a road is paved or unpaved, 1-lane or multi-lane or even if it is actually a road. Below is a screenshot I made of an area I was investigating as a test of the layer. You can see the thin white strokes don’t quite match the Mapbox imagery beneath it. Is that because their imagery is more recent or is it an error in their post-processing? There’s no way for us to determine that.
I’m very disappointed but not surprised. My quick reading of the terms of the EULA the other day suggested to me that ordinary folks like us were excluded from deriving information from Digital Globe’s data. That turned out to be true. I doubt the Facebook TMS layer will be of much help to us and furthermore, without access to the same imagery the FB group is using there won’t be any good way, short of a personal field trip, for us to verify or validate their work.
Agreed. It’s merely a “track” in the middle of nowhere and connected to nothing. How they decided to map that way as a track instead of an unclassified and unpaved highway, is unknown. The changeset comment is not very helpful either. It’s the same old stuff again “#nsroadimport#thailand”. Useless.
How do we stop this train before it crashes and creates another big mess?
One answer is to make the DG data available to skilled OSM members. If we had access to their high-resolution data it could make a huge difference in our ability to add details to the area. But I’m afraid that won’t be possible. DigitalGlobe sold those rights to Facebook and they are not likely to give anybody free access. DG isn’t interested in making roads more accurate, only in making money to cover their investment. Can you blame them?
Oh well, after all that it’s just another day in the 'hood. Copyright is a double-edged sword. It protects the owner and impedes the spread of knowledge at the same time.
As we explained in e-mail, we know it’s not ideal, but we cannot share our imagery because of our agreement with DG. This said, we are going to share more about our process and quality checks via the import list.
Thank you for your feedback. We based our tagging on a few things;
Mishari and I had some of the initial communication with Facebook about their import. We have raised our concerns and answered questions to the best of our knowledge. After what happened with their initial set of edits, they really do want to be extra careful not just with the data, but also with the community.
The way I see it, is that we have an extremely technically talented group at Facebook to our disposal, they have access to resources that we do not have (Like High-Res DigitalGlobe imagery). Instead of dismissing them as Stephan is doing, I’d much rather try to find ways to use those resources.
I don’t think Stephan is fair in saying “Go away and play somewhere else”, especially because he doesn’t seem to communicate his real reasons for saying so. If I understand Stephan correct, then even if Facebook get a 100% hit rate on the geometries and tags, he still wouldn’t want their data.
If I understand Stephan correctly, then he is afraid of the effect it will have on community growth. If all roads are correctly mapped, it gets less fun and more difficult for new mappers to get started. Just like we are seeing in the US with the Tiger import.
I can understand his concern, but we must also realized that one day the road network will be complete with or without the help from Facebook. Should we then all slow down a bit to avoid ever completing? I am personally adding data as fast as I can, I want OSM to be the best map out there and I think Facebook can help.
Stephan let me know if I have presented your opinions incorrectly.
I do not know what the correct way forward is, but I’d much rather work with Facebook on a solution than just dismissing them.
I want to second Johnny on this. The guys at FB have been quite nice and enthusiastic and I think we should discuss ways of incorporating the work done by FB’s team. I’m willing to say that what has transpired has been to a degree my fault as I didn’t have time to think through my feedback to them.
It doesn’t even have to be direct imports but some ideas I have are:
done in grids and validated by the community similar to hotosm
.osm files that can be validated and incorporated gradually.
generate an enhanced maplayer that can then be traced manually
We should consider this a good sign that OSM is so awesome now that FB wants to be involved. Now we just need to find out how we can work together.
Again a sign that you didn’t understand the fundamental principles of OSM as a community.
It’s not just a storage of free (as in beer) geodata.
It is a community gardening the data. This is one reason the data is of so high quality that even companies chose to use it as a data source.
You think there is a tiny group of “leaders” you have to convince then you have “permission” to do your things on OSM. This is not true.
Neither DWG, nor Johnny, Dave or Mishary have the power to grant you permissions of any sort.
OSM is a community. Your edits have to be considered conforming to our community standards.
Part of this process is transparency. You don’t silently develop something in your office and suddenly present your result. You have to include us in all your steps.
You plan to use “secret” data sources not available to others. This makes it harder to verify your edits. Handing out API keys to your edit interface is not a solution.
Yes, I might be hard by saying “go away”. But I believe this is reasonable as long as you chose no to cooperate in an open and transparent manner with the community. This doesn’t say I disagree with having roads mapped or support in completing OSM data. I disagree with the process you have implemented doing so.
I have the feeling you understood the problem. It’s not about having data in or not. It’s about how it is done. It is a meta layer to the actual editing.
It is about finding a way on how to best integrate this with the community. To strengthen the community and grow it.
Referring to a previous comment by Johnny. Yes, even a 100% correct import would be something I see skeptical.
OSM had not yet enough time to grow to master the challenges of gardening the data on that completion level.
With the inflow of occasional mapper using maps.me we already see shortcoming in how the community can integrate this. We have to work on these challenges.
AI has the potential to be useful here. It could be a companion to newbies supporting them in getting started and avoiding common mistakes.
So yes: I see a potential here in multiple areas on how Facebook could support us here. But let me stress it again: It’s not about working in secrecy, it’s about openness and acceptance of community feedback. Just pretending to listen to the community and then continuing as before is not.