GRAB - Here we bloody go again !

Fresh on the heels of the Facebook, cock-up … we now have the Grab team coming into Chiang Mai, and wildly editing stuff…
And of course we find it by chance…

Just happened to visit a Powder coating Company I use, and you know how you subconsciously remember stuff… well back home, I just wanted to amend its tag … but WAIT, I see a new residential road added, that connects and links across the Middle Ring road, which I know wasn’t there today !

ak_ram, who states he works for Grab, has used his massive 1 week/40 edit experience to put in a new road right through undergrowth and then tops it off by changing a Secondary link to a residential. Way 641423275 if anyone wants to check.

Meanwhile renuka1, another clown in the Grab team, decides to change one of Dave’s personally surveyed Soi’s (Way 390556246) to a service road. Of more concern is that he claims nearly 2000 edits to date !
WHY ? … Why f**k with our data, why a service road, what relevance does it have for a Grab taxi ?

All of that in an area of about 1 sq km, I stumbled across ;https://www.openstreetmap.org/way/390556246#map=18/18.83351/99.00352
Just what else has been done since these fuckwits descended on Chiang Mai?

So, GRAB, if you are reading this PLEASE, PLEASE … STOP TODAY … Get some advice from the mappers who live here and have more experience than u.

And instead of us all ignoring this latest threat, Stephan, Johnny, you are the computer Gurus so please help me in reverting ALL of ak_ram’s edits until he proves himself. Its only been a week. I can make individual comments on the changesets, but I need the DWG to act fast with this GRAB menace until they can prove themselves.

Please, please do tell the people that are making these edits that there are problems.

Right now at http://resultmaps.neis-one.org/osm-discussion-comments?uid=7795623 there’s nothing apart from 1 bit of (since deleted) spam and 2 tests.

Edit: Thanks for commenting on http://resultmaps.neis-one.org/osm-discussion-comments?uid=9035937 BTW.

That way looks OK to me: I think "unclassified"as originally tagged by Dave is not as good as “service” for such a small road just leading into the backyard of a vehicle inspection shop.

Re way 641423275, Bing and ESRI show some structure which could be interpreted as a track from the residential area towards the main road. I would not draw it. A different imagery shows a lot of vegetation there. I don’t know the current state there.

That’s what I see in the imagery. Of course, on the ground knowledge is preferable.

I can tell you that the road in question was mapped by me from a photo taken from a motorcycle drive-by with a camera. I didn’t look at the way very closely but because it had a name, no obvious houses, and could perhaps connect with the residential streets in the trees, I opted for unclassified. It’s a tough call either way because the soi is so small. Usually in Chiang Mai, any road that has a name is residential even if it’s a tiny soi but in this case tagging it as a service road isn’t a horrible choice.

Creating those residential ways near Russ’s powder coat shop look like the work of a pure amateur, however. Again we see ways that were drawn hastily, connected through 90-degree corners, and that look like they’re only rough tracks through a field. The only cure for it is to write changeset comments asking them what reasoning they used to add such “ways” to OSM.

I fear that this is the future of OSM. Its very openness and the assumption that everyone who edits the map has the goal of making a quality map worked for a long time. I bring up the example of Wikipedia again because it too was once very open but because of vandalism and concerns about data integrity, the organization decided to close its ranks. They made it more difficult for a casual contributor to add information. They made it so difficult that many people, myself included, stopped contributing anything. Lots of good information and help was lost when they did that. If there is no vetting process at OSM and no training period for new mappers, we’ll just have to learn to live with this sort of thing until it becomes such a problem that OSM resorts to the same prohibitive stance.

Dave

Dave,
Having had the benefit of riding down it, it does connect to the residential Soi beyond, but impossible to know from the aerials as the corner is under a tree (I have now added the connection).
But now I see more roads by ak_ram, to which I have added the following to the changeset comments

OK, Mate - Im asking DWG to formally revert all your edits …
In the same area:
Way: 641423287 - You have imagined a road here. This is a field, there are no houses. There is no road.
Way: 641423288 - This is a drive to a building company. You have continued it through their yard, cutting their building, then out through the back wall, and along a tree line.
I am having reservations about both your eyesight and imagination. Your seniors must also suffer the same conditions if they are approving this.
Let me tell you Chiang Mai is quite well mapped, so please don’t make up roads just so as you can add something.

Now ak_ram did go back and delete his earlier residential that went through the bushes, but I’m afraid it prompted another note:

Hi Akram, Thanks for revising the way, but please understand that at OSM, we simply do not assume that roads connect when we can’t see the connection.
Yes, its a judgement call always, but in a case like this, it would be better to draw half the road where you can see the asphalt, and leave a “fixme” tag, for local mappers to review and visit the area if required.
The very words you used … “maintaining connectivity” imply you feel all roads have to connect somewhere, which is not the case.
I have gone back and corrected that area, but can I explain another error you caused … when you just deleted the road, you destroyed the actual connection of the U-turn to the Dual carriageway. To any GPS it would not use the U-turn as it does not see a connection. You should have re-joined the nodes (which are shown in Potlatch 2 by a red circle over the nodes in question).
As I said, please take some time to read about editing OSM before making important changes on major highways.
Its all fixed here now, for u.
Russ

So, Andy@DWG, can I formally request all of ak_ram’s edits are reversed immediately. I think if we start to do this for any mapper as soon as we find evidence of an uneducated approach, then we might just stave off the Wikipedia approach that Dave describes, and ultimately we will have to go down.
Russ.

Hello All,

Thanks RussMcD for taking out time to share the feedback to our team member
We respect your time and feedback and ensure we are calibrated with local policies.

As per the conversation on the change set comments, we had ak_ram to stop editing and is put back to training with the immediate effect after receiving your feedback(same is communicated to you through change set comments)
Meanwhile all his edits are QC-ed by our internal quality team to ensure map quality is not degraded.

Further if you notice any mistakes, please do let us know and we will correct accordingly

Thanks again for the time

Grab Team

I applaud @Russ McD for actually going into the field to verify them, and @AlaskaDave for speaking up about the slippery slope OSM is heading down.

Everyone, if you have to fix/revert damages caused by Grab OSM “contributors”, I would like to ask you to also add following hashtag in changeset comment:

#WhatInGrabsNameIsThis

(Pronounced: What in Grab’s name is this?)

I’m now busy fixing Grab-induced damages in downtown Bangkok, after-the-fact, case-by-case, anywhere that I could do ground survey. Huge time sink, I know; but it doesn’t sit well with my conscience to leave them as-is.

If you happen to see supposedly-straight urban roadways/sois being slightly twisted like earthworm, or excessive node/curving in supposedly 90-degree turns; you can assume it is their doing-- targeting these would give you much more hits than misses.

I will post my writeup on GrabOSM Thailand complaints thread later. (I originally plan to post it immediately after my rant, but I got winded up busy fixing the errors instead)

Sorry that a delay in my part happened to give an impression that their “uneducated” edits are acceptable…

Note: The statistics of #WhatInGrabsNameIsThis exclamation could be viewed on Pascal Neis’ site.

I second that. The reprisal should have been Revert, Block, Ban (in that order).

Otherwise, on-the-ground contributors will wind up in an endless spot-a-difference game. If they were not dedicated OSM hardcore. they would definitely give up.

One of the problems I saw with FB’s contributions, and now the impact Grab is having on the Thai OSM, is neither organisation really took the time to explain what, and why, they were contributing to OSM …

I mean with Facebook, even the latest diary entry from them reveals a process in AI, and how they conquered data entry to add ways on a scale no-body had done before. Its all about computer geekery. Not once did they really look at the accuracy of the tagging. Sure, they eventually connected nodes accurately, and way hay, even taught the computer to draw bridges … but they never looked a the real mistakes of correct and appropriate tagging.
Which is why we are now still left with virtually everything tagged as a residential road, from tracks across fields, to peoples private drives, and Yes, even an airport runway !

So moving onto Grab …

For many living in Asia, both Grab and Uber have made a dramatic contribution the manner in how we travel “over here”. Uber’s software was pretty good, but for reasons only known to them, they sold out to Grab. At first we thought there would be no difference, but soon learned the Grab software is a long way behind that of Uber, and one area was the poor availability of, lets call them, waypoints. Grab insisted you needed a location in their map, presumably Google. The frustration of booking a Grab, only to have your confirmed pickup location appear hundreds of meters away, was annoying and led to many cancelled rides when the driver couldn’t find you.

When Grab introduced themselves to OSM, I thought great … a real opportunity to involve their organisation with local mappers, and as we live here, surely we know the places people want taxis from.

No, what we get is a bunch of kids playing with the map. No real explanation of what they want to achieve. No introduction to us (apart from Mishari). No experience on OSM … And from the apologetic grammar used in the Changeset comments, I doubt if they are even from Thailand.

This would be a great opportunity for Grab to align themselves with OSM, but instead, what we are seeing is another major organisation, allowing a bunch of amateurs loose on a training ground they call OSM. They are all being paid, while us volunteers, mop up the mess.

I suggest if any Grab senior management read this, how about taking some of that money and inviting us to a presentation where you can explain what you want, and we can help you achieve it. Otherwise you will just have a load of pissed off mappers deleting data as fast as you are putting it in.

I presume like airports, is good to have exit gate numbers mapped, and I presume the same would apply to many buildings. Every time I call Grab to my Condo, they drive to the small Soi behind, then leave in frustration as they cant find it. If through OSM, we can help them by plotting entrances, or drawing the small service roads that take them to the front door, I’m guessing we would surpass Google’s maps in terms of helpfulness to both drivers & passengers.

But no, if you carry on adding roads that don’t exist, and messing with the tagging of existing roads, how is that going to help Grab ?

We can be on your side, but they way you are starting off is just going to alienate your organisation from ours.

Russ.

@Russ McD

Can you link any specific edits causing problems? I reverted or reported to DWG many problematic edits, but without a link to specific edits with ignored changeset comments I am not planning to spend my time on this specific problem.

If that is true then reverts of all edits of all their accounts are in order. But at this moment I am unable to confirm is it correctly describing the situation.

@Mateusz,
Hi,
I’m just about to shoot off to Laos for 10 days so wont have time to dig my comments out, but if you can search for all comments on changesets in the last 2 years, by user RussMcD, then you might get a idea of some of the issues.

Hi Russ and team

Thanks for flagging this and my apologies for our error. I’m Ajay and I lead grabs regional map operations. Would be good to touch base with you and learn more as you suggested. Please let me know how. We’d love to be more involved in the community there and map together.

To add to this thread, I found a Grab user that has restored a demolished city block and connected roads in my neighborhood which are separated by a wall. These are just the immediate changes that I have noticed.
https://www.openstreetmap.org/changeset/62994668

Secondly, unrelated but I’ve noticed that Grab has started using OSM data without attribution in the same way that I discovered Uber did a while ago.

Regarding your first paragraph, if we quickly revert every changeset once we find a few errors, then the message might get though.
If we just highlight it on the Thai forum then take no action, they will carry on.
I do get the impression that this team are struggling to find meaningful new ways, and are resorting to making them up to justify their existence… and perhaps salaries too.

I offered to be a consultant with Grab so that I can spend time to help put their contribution processes in order, adjust their KPIs to reflect quality contribution, make sure their contractors know what they’re doing, review processes and procedures, conduct training etc. however they made it clear they don’t have a policy to take on consultants so I’m not sure how else to help. I suggested to them that best thing they can do is to remove the remote mappers but this doesn’t seem to be an option and I cannot figure out what Grab has to gain by continuing to flood the map with low quality data.

Would the best way forward be for Grab to fork OSM and work on a forked version? I’m not sure if it’s fair for Grab to mess up the map and leave the maintenance to the community.

This like this quote from the Phoenix Project book:

Grab’s given us a free puppy and it’s running around taking a dump all over the place.

Best Regards
Mishari

Can someone recommend what’s the best way to revert edits? I usually use JOSM, but it seems a bit complicated. Is there a better way?

Well Ajay, I’m guessing you are in India, and we are in Thailand. Will your Company pay you to travel over here ?

If not … well where do we start ? To summarize, we are seeing many examples of your team adding roads that simply are not there. How do we stop that ?

Dont forget, that your efforts follow that of Facebook where they saturated the map with bad info… many of us have tried the Mr Nice guy approach, and all we get back is profuse apologies, but no real reverting.
Some of now are just deleting massive amounts of data where we see continual bad mapping, and you must agree thats a waste of time for all of us.

I presume your team is paid, and I presume no mapper like to end his day with no ways added … so I for one am convinced Global Logic might have presented Grab with a good looking proposal to increase OSM coverage, and are now looking to add anything they can, just to justify this. Am I wrong ?

You have to stop this, and stop it today. We dont need thousands of private driveways added, or tracks across fields where Im sure no Grab taxi will be pleased to go. So please STOP TODAY.

Contact me privately if you like, or via this forum.

@Russ Ajay is based in SG and in all fairness he has offered to come down and talk and to work with the community.

The ideal way I see this working is for Grab to have a dialog with the community, we establish shared goals and Grab deploys resources to achieve those goals.

At the moment I’m not sure if there is any possibility of a shared goal. In fact I’m not sure what Grab is trying to achieve, and whether OpenStreetMap is the platform by which to achieve it.

Hi Russ.

Yes we’ve stopped working on Thailand. I’m happy to come over to Thailand and have an open dialogue with the community either on 15-16th of January. If the dates work, I can set something up for us in the office and we can chalk our a plan to collaborate better. Looking forward to seeing you in person.

Hi Ajay,

Thank you for your cooperation, we really appreciate it… Will you be able to make it to Chiang Mai? Most of the active community lives up there. I can fly up for the day if its on the 15th.

In the meanwhile it would be good to have more insight into what exactly you guys are doing. Scope of work, training manual, objectives, KPIs, any documentation that you have that explains why the mappers are working the way they are would be a great place to start.

Best regards
Mishari

At present, both dates work for me. … but in CM of course.