thanks for getting in touch with us. Sorry we had to give you such a hard start.
I’m glad we had been able to sort out the problems. Let’s work together in rectifying the problems left. Then we can move forward on improving OSM together.
I’m quite confident that you can be a valuable contributor to OSM Thailand.
As you had learned the hard way: Sometimes it’s better to start slowly getting used with all the tooling, process and conventions. I can understand it’s difficult to slow down when you are full of motivation and good intentions.
And yes: We are all people behind our usernames. The mappers I have meat are all friendly, even myself can be friendly from time to time
If you have the chance to meet with other mappers in Thailand, please try to do so. It’s great to get in touch with others. Share ideas, discuss about the tooling or even completely different things.
There is a large community in Chiang Mai, others are in Chiang Rai and biking in Laos. Bangkok also has a small community. Not sure about Isaan. Best would simply be opening a topic in the forum and ask.
I’ll cross check the commented changelists. Johnny and other certainly also will support in getting things fixed. I will try to get a list with changes you wanted to fix and a delta of what is fixed and what is still to do. Until that the overpass query is a good start, even if it misses a change.
I compiled a more readable list of things to review/repair. It is based on last week’s data.
It highlights:
deletions of name:th or name:en
modification of ref, int_ref, highway, place
It does not check for int_name as these have been reported fixed.
If something was modified by Balthus and is meanwhile fixed (regardless of last editor) it is not included.
If I shall exclude specific elements please send me a list of the OSM element IDs to ignore.
I noticed also multiple deletions of named junctions. Does it make sense to check for those as well? What other kind of “fixes” did you do? Did you also modify waterways? Shall I filter for other things?
What a mess … I had a cursory look at the huge list … seems like a bunch of edits that only Balthus knows why. If something was deleted, I guess we just need to know he went there and physically verified it wasnt there.
Some residential roads changed to service … well, OK, but only if he knows the area and realises the difference in what the road does.
I rode past a new sign here in March 2015 saying “รย.4036 Scenic Route” - I stored it as a waypoint which I only ever do next to the sign … he has now changed the route to the 4001.
Now if he had seen a different sign since, then fine. Just add a note. But if its come from some out of date Database, then Balthus, leave the bloody ref alone. I see he has helpfully added a fixme=check ref. Well thats for him to do… dont undo other peoples work and then introduce confusion.
And whats with adding the reference to a description tag … what the hell does that achieve ?. We normally use that for a physical description of the road.
If he is using the DRR & DOH sites, along with Google street view, I think thats breaks the terms of where we can get data from, and then to credit them as a source … is that not asking for trouble ?
My vote is that for any ref changes he had done, and any road status changes too, then the changesets shud be reversed. Re-run the report and we can see whats left.
When I was entering the data collected during my December holiday, I found out that Balthus had mapped many of the roads meanwhile. Though some of his “tracks” looked more like “unpaved residentials” to me, the majority of his edits, including upgrading or downgrading of roads, between Ubon and the Mekong seem acceptable to me. Nonetheless, I changed some of his edits, please make sure not to destroy that.
Job 02 : Removing int_ref from Rural Roads and fix source
Start: 27-Feb-2017 – Done: 3-Mar-2017
In this step I removed all int_ref values from my touched Rural Roads and fixed the source.
All information came in general by my personal survey. Additionally I did a confirmation check for every touched Rural Road in the DRR LMP database to verify my survey.
In only one case there was a difference – and Russ actually found it. This case was the above mentioned scenic route Chaloem Burapha Chonlathit Road which is 100 km long and stretches over two provinces (Chanthaburi and Rayong). I did this route at least 10 times – and by survey this road has only road signs with รย.4036 (Rayong 4036) – even in Chanthaburi. As I checked the DRR LMP – the red vector layer shows the Rural Roads – I noticed that from the Chanthaburi border they gave the road a new reference จบ.4001 (Chanthaburi 4001). I guess, I messed it up and should have waited for the «new» road signs to make any change – am I right? Sorry.
Of course I will revert จบ.4001 back to รย.4036.
Remark: int_ref
I defined a romanised version of the Rural Road reference mainly for two reasons:
It can be a help for people who can not read Thai script, that means could be helpful for most of the foreigner.
The romanisation by the renderer can lead to conflicts, that means different roads could have the same romanised reference. Eg. Chonburi ชบ and Chanthaburi จบ are rendered both as ChB.
Instead of deleting the value in this removing step I just saved the value in the description key for a possible future use. As you seen in the OSM wiki http://wiki.openstreetmap.org/wiki/Key:description this information could be a help for end users who cannot read Thai. In the Wiki the use of the key description is quite general – it states it’s just additional information for the end user whatsoever. Russ, I do not see this physical exclusion.
But if there is a consensus to delete this value in the description key I will certainly do it – no problem.
The remaining int_refs reported had been checked manually by me and all looks fixed, so I removed that from the reporting.
I think what now is missing is a review of some of the changes. If you do so, please send me a list of elements you consider OK. I will then filter that out from the list. You can simply mark the line in the browser and copy paste into a message. It should simply contain the element reference, then I’m able to pick it up.
Many of the deletions come from removing the tags from nodes on highway=services and creating new elements.
If others assist in reviewing what’s left by sending IDs of checked elements I can provide daily updates. Should bring down the list to zero quickly.
Originally it was highway=tertiary
Now its changed to… highway=unclassified,source=DRR LMP, Bing Hires"
So I take a further look as I have ridden that road before and know it well …
1/. Stephan originally created it as tertiary but with no ref.
2/. Balthus checks the DDR site and despite it appearing on there as the NK.2024, he changes it to Unclassified !!
3/. I know its a major road … My Garmin GPS has shown it as the Nong Khai 2024 for as long as I can remember
So, its a classified road, or tertiary in our books … why the change Balthus, and then note the source as the DDR site, which clearly u mis-read ?
Many of the items on the list need checking carefully and basically I am not going to look at his list as every edit just seems random, and its up to him to go through the lot. Im still for reverting all road edits where he even mentions the DDR or DOH, or anything where he has changed the status.
Balthus also did send a list of IDs he checks. I wonder if I should simply acknowledge these with the concerns brought up by Russ. Currently not included because of this.
@Balthus: If you modify something tagged, then it is important that other mappers can understand why you are doing something. Document it in your changeset comment. Eg you did a more recent survey on the ground.
@Russ: Can you come up with a detailed description on how to detect what to revert?
How to make the review process easier? Shall we consider highway reclassifications ok for residential<->service, for track->residential/unpaved, track->service?
Finding what to detect is difficult …
Virtually every change I look at prompts questions as to WHY ???
For example, if I open up the first page of you latest list, fourth one down (Node 1651244385) I see Cafe Amazon deleted ? Well thats strange, open up the link and the PTT is still there. So whats the chances of that happening in reality ?
PTT and Amazon are inextricably linked in real life … have they really closed this one while keeping the station open ?
Rocketman added the Cafe five years ago, and I think he is an experienced mapper… I guess Balthus could not see it on Bing, so he deleted it.
Fine if he physically surveyed it, fine if he contacted Rocketman who fessed up to a mistake … but I have no idea who is right.
I have added complete new PTT service areas that don’t show on Bing … I hope Balthus has not removed them just because they don’t show on the aerial photos ?
So, Yes, there are so many edits I just cant work out why he has changed things. And I afraid giving you criteria is not easy, short of reinstating every deletion, and reverting every road status change.
Yes - for someone else is probably impossible because I did following wrong:
1). I did saving of my edits time-based instead of object-based, which makes a review VERY painful. That means a changeset contains many edits together. This is VERY unpractical – and with this review I have learned this. I won’t do this again.
2.) Although all of my edits were in general based on personal survey on the ground, I did not mentioned it in the source – which was wrong. For many roads I fixed this but not for all.
http://www.openstreetmap.org/way/50592797#map=14/17.8468/102.7020
I did this road in Jan-2017 and could not see any road sign with a reference number, plus it didn’t looked like tertiary to me – so I checked my survey with DRR LMP which states it as a Local Road and not as a Rural Road. For me this Road is probably an outdated Rural Road from before 2002. But please prove me wrong.
I «improved» quite many of this big PTT service areas as they are important to me too. Most deletes there were based on spelling and drawing buildings. I hope I did not delete your new PTT service area which were not on Bing or did I? Please inform me if I did.
This review is an elimination process to find out what is ok, what was clearly wrong, and what is problematic. The first results of this triage I already sent to Stephan – but I am not finished yet. As a end result only my problematic edits will remain. For these we need a general solution then – e.g. a revert.
Thank you for your patience until we can finish this triage.
As a survey can also be doubted I have started to contribute photos of road signs and way points to Mapillary.
@ Stephan
Yes, I learned this the HARD way. Even for me it is very difficult to comprehend my edits afterwards. My edits should be transparent for everyone else. I also learned that adding is less problematic than changing. Changes must be comprehensible for others.
I reviewed some of them. Before there was a PTT highway services area painted and the corner nodes of that area had been tagged with petrol station, Café and 7-Eleven/Jiffy. Balthus created new nodes, probably closer to the center of the buildings involved and moved the tags into these.
Not sure how much better the data got due to this but certainly it didn’t get worse.
In that context I spotted some mis-taggings. The key is all lower-case. So it is “fuel:e20”. I fixed those already.
So it’s time to cool down a little. Let’s thank Balthus for his co-operative behavior. Among a few thousand changes, some may be wrong. Of course, every mistake is bad, but let’s not exaggerate.
Separating overloaded nodes/ways into individual nodes/ways is correct, though at first view it may look like a deletion of data.
After my Thailand holidays, I found some of his edits near Ubon, which may not be present in your lists because I did further changes to some roads (hence I am the “last changer” of them). E.g road 2112 from Khong Chiam to Kemmarat was tagged as a tertiary, Balthus changed it to secondary. I agree with that. primary could be possible, but secondary looks better, tertiary is wrong.
That road south of Khong Chiam https://www.openstreetmap.org/way/388895392/history was tagged as a secondary, and Balthus changed it to unclassified. I agree with that, also tertiary could be acceptable.
There were some more changes in category which I wanted to do, and found out that Balthus had done them already.
I hope Balthus will stay with us and contribute eagerly to the map of Thailand.
And not to forget: @Balthus: welcome to this small group of mostly Farangs creating the map of Thailand!
@Bernhard: I am parsing the full history. So I also detect changes if later someone else touched the object.
In case the current state it equal to the state before Balthus edited it (looking at specific tags only) then it would not be listed.
And a new version containing other reviews filtered out. This new list also excludes changes I globally considered OK after spot checks. Much smaller number of elements left for checks.
Please browse through it and do spot checks if something looks suspicious. http://downloads.osm-tools.org/balthus-20170307.html
How confident are we that changed reference numbers are now in the cast majority correct? If so I could also filter them out.
By spot checks I found some corrections. These had been added as ref=2222 and later been change by Bernhard to 4005.
way 198481274
Road 2222 vs. Ub 4005 is the road from Ubon to Phibun north of the Mun river. Ub4005 is what I found on the ground there. 2222 is also used by Google. I cannot exclude a recent change of the reference number in either direction (2222 being the correct number but old signs still there, or Ub4005 being the correct number changed from 2222 recently).
As we saw in a different thread, reference numbers are a mess…
Together with Balthus I fixed another few dozens of things.
As reference numbers seem to be mostly OK, I now consider changes of the ref as OK. Further on a change of the highway type is also considered OK if a ref was added as well.
The list is boiled down to ~700 ways. A lot of the remaining items are modifications of the highway type. Lots of them looking fine. So best would be to open the list in JOSM and check out an area you are familiar with. If it’s to be ignored select the ways and copy the ID to a message to me.
Maybe a good time to state it clearly: While there had been bad edits (yes, plenty), a very large number of Balthus’ contributions is good and very valuable for OSM in Thailand.
@Bernhard. Thank you for your welcome :)) I also apologise if I messed up edits of you. Through this review I learned a lot and I hope I will not mess things up here anymore.
Regarding อบ.4005 (UBN.4005) I can confirm your survey. I have been in Phibun in last December when I rode the 2222 coming from Khong Chiam. I did not make the อบ.4005 fully – I only did about 1 km to check the road condition – and turned back for the bridge crossing the Mun river.
Btw. I just did a check now in the so-called «outdated» DRR LMP database. The vector layer for the Rural Roads (red color) confirms our survey. Remark: As you see in the first section of check boxes on the left side the various base layers are coming from external sources – they serve as an orientation only – the DRR LMP can’t be blamed for the quality of these bitmaps!
Balthus, briefly, Yes I apologise if I was a bit harsh in my comment to you, and now knowing that a lot of your edits come from a personal survey on the ground … well that takes precedence over any other source.
I will say its nice that you are continuing to work with us to fix the edits where necessary, and perhaps like others, when we see edits from a new mapper without explanations, its easy to jump to conclusions.
For the future, just leave a brief "note=* " tag for the important changes, or the ones where you are changing the work of others.
A couple of habits I have, are :
1/. When changing a single node into a building, then change it to a way first, then move the now 4 nodes to the building outline, adding more if the shape is irregular. This preserves the node history, keeping a record of who first added the POI. When you delete and add a new building, and retag, then the history is lost.
2/. If for example you want to a delete a feature, such as a Gas Station, then leave a single node on it with the note=* tag explaining something like … “note=Station deleted, noted derelict in Feb 2017”. This avoids confusion as many mappers will see it on the Bing aerial and add it again without knowing.
And please be careful when just changing roads … for example you have changed the route where Hwy 242 connects the 211 at Tha Bo http://www.openstreetmap.org/#map=15/17.8339/102.5944.
I doubt if it splits in the manner you show, but more importantly, you have not deleted the relations on the old road when you changed it to unclassified status.
And then you add “fixme=verify”… I would suggest you don’t change anything that then needs verification.
After a painful wrist inflammation I am back to computing and back to this review – sorry for the delay to everyone.
@Russ. Thank you very much. I think I understood your two habits – much more practical and transparent for other mappers.
And yes, in future I will be much more careful about changes of classification and reference. I have learned this in my review.
Yes, this fixme=verify comes from this review and means a verification by someone else could be helpful and that I am now not 100% sure about this change anymore. In future I try not to make changes that needs verification – I promise. I also try to upload more of my GPS traces and make photos with road signs for Mapillary for better verification of my mapping.
You don’t need to upload photos to Mapillary. Writing a comment that you have been there and take on the ground information from your photos should help others to see that your change has a reason.
If you can contribute photos to Mapillary that certainly is useful. Also as those pictures might include additional details you not thought about and others can add. Especially with the last months update of Mapillary they now allow you to get to the full resolution image uploaded it actually gets possible to read street names. And incorrectly blurred areas can be corrected to be un-blurred. Mapillary is now actually useful.
The “fixme” tag needs to be handled with care. You already got the idea that all data needs continuously be verified. It is a similar story like these “under construction” banners of websites which later got replaced by “always under construction” banners.
With most likely your changes we see a roughly 15% increase of the fixme tags in Thailand during the last month.
With osmose running for Thailand on my server we have a good statistic on this:
You can review these on a map eithern on http://osmose.osm-tools.org/ (lots of checks available) or specifically the “fixme” by running a query on overpass.
