Does anyone even check what HOTOSM contributors leave behind?

Posted by ff5722 on 4 September 2017 in English (English)

To name a few things in the above area:

  • residential roads tagged as motorways, primary or secondary

  • the same road went from motorway to service to residential to primary

  • island roads (not connected to rest of network)

  • roads ending just before the crossing

  • roads split in arbitrary segments all over the place

  • ways with a node for every half a metre length

And how about this?

Location: 15.192, 108.783

Comment from SomeoneElse on 4 September 2017 at 12:45

Any chance of a link to the area? I'm aware of a suspicious "trunk road adder" in China (who isn't actually a HOT contributor as far as I am aware) and it'd be useful to see if the trunk roads above (a) are the same user and (b) match anything on any imagery.

Comment from Zverik on 4 September 2017 at 14:02

@SomeoneElse, the link is in the message footer, the one with "Vietnam" in it.

Comment from SomeoneElse on 4 September 2017 at 14:22

It's a different user. What I'd suggest rather than (or perhaps as well as) writing a diary entry about a problem like this is to actually try and help the mapper make better contributions. Currently they've created 13 changesets, and no-one has commented on any of them.

Given that they're a brand new mapper I'd suggest a "hello and welcome" comment in the first instance, then trying to explain where it's appropriate to use "motorway" and where it isn't.

Oh - and thanks for patching up the data :)

Comment from Jean-Marc Liotier on 4 September 2017 at 15:34

I feel your pain man... And then they trumpet : "58736 buildings and 2382399 kilometers of roads added !". Quality control - what is that ?

Maybe the initiators of Tasking Manager jobs are unwilling to frustrate novice contributors by having them correct their errors. Maybe they do not want to do it themselves... Cleaning up after them is infuriating.

Comment from Glassman on 4 September 2017 at 16:33

Why is HOT singled out? New mappers make mistakes all the time. I just fixed, and commented on the changeset, a street tagged as a gate, instead of a node on the way where the gate was actually located. HOT tasks have a validation step. You could spend some time validating tasks.

Anther approach, which I'm in favor of, is to build better systems to prevent mistakes in the first place. Take my example above, why doesn't the editor complain about adding a gate to a highway=residential way? Or your example of the polygon house tagged as a railway.

The problem isn't HOT. HOT only stands out because so many new users get their start on HOT projects.

iD's new feature of asking for a review is a good start. I believe that [OSMCha]{ will be update in the near future to flag changesets asking for a review.


Comment from Jean-Marc Liotier on 4 September 2017 at 17:07

Indeed mistakes are normal for new users and HOT has no monopoly there (and new users don't either - experienced users such as me sometimes make mistakes of considerably larger scale). What sets organizations such as HOT apart is that they are skilled at corralling lots of new users into projects to which they claim leadership - which is all a good thing for everyone involved. But I believe that the branding of these new user's contributions (witness their ubiquitous use of hashtags in changeset comments) means some responsibility upon the leadership who organizes them. That leadership is no hierarchy and the users take no orders... But there is leadership in organization nevertheless and with that comes responsibility for the outcome. In technical terms, it means that I believe the leaders to be ultimately responsible for the quality of their project's output: they cannot foist that upon the unsuspecting new users who, contrary to them, have no idea about what a sufficient quality assurance process is. The tasking manager has a validation step built upon its process, so all that remains is a will to enforce that step measured using the excellent array of validation tooling such as Osmose or the JOSM validator: enticing new users into contributing and branding their contributions comes with a responsibility to coach them.

Comment from rab on 4 September 2017 at 18:40

It is not only a problem of new mappers. This Problem is in deed very HOT specific. HOT is encouraging new mappers, to map places, where nothing is to find comparable to their normal environment. And in some boring tiles, where nothing is to map, even experienced mappers start to map animal traces as highway=unclassified. Project managers and requesting organisations are most of the time neither involved in mapping nor in validating. No critical feedback, only success stories all the time.

Comment from ff5722 on 4 September 2017 at 18:46

@Glassman HOT makes tasks in areas with few active local mappers. After the HOT task is over, the area is mostly abandoned. The contributions through HOT are also highly concentrated, making solving the mistakes an gargantuan task. Together, this means that these mistakes will remain on the map for years, in this case, since 2013.

The quality of the road data added is pretty much worthless too, in the time it takes to classify the roads, fix crossings, extend roads, simplify or detail curves etc, i could easily redraw the road network from scratch.

Comment from butrus_butrus on 7 September 2017 at 07:42

"ways with a node for every half a metre length"

There is nothing wrong with it, the more detail in OSM the better!

Comment from dcp on 7 September 2017 at 15:53

I have only contributed to one or two HOT requests and I did not return as I thought (maybe wrongly) that the quality of the data being added was mostly wrong and/or worthless to the HOT mission in question and to OSM as well.

I also considered the HOT specific tagging lacked the connection to the HOT team request and therefore there was very little chance of it being corrected or removed.

Here is one example:

HOT requested the users to identify possible helicopter landing pads. This is almost impossible to do using Bing as you cannot judge the ground inclination but I tried anyway. If these pads had a specific HOT tag then they could easily be removed once the HOT task in question is declared over by the HOT authorities (or an OSM user volunteer).

Examples of specific HOT-tags:

Comment from dieterdreist on 7 September 2017 at 16:29

Yes, new mappers make more mistakes than experienced ones, but HOT has a systematic problem: they encourage people to do armchair mapping in areas they mostly never have been to. Remote mapping is generally more difficult than mapping what you know, so it shouldn't be the first thing you do when you come to OSM. Also HOT areas are typically lacking active mappers who could assist newbies with their first steps.

Obviously, errors like those reported in this diary post will likely not happen with anyone just slightly responsible and careful. Uploading something like this to a shared database is either vandalism or complete incompetence and reluctance.

Comment from SomeoneElse on 7 September 2017 at 20:30

@butrus_butrus "ways with a node for every half a metre length" are a real problem if those nodes don't add any extra value (e.g. if it's a straight line), or if they're in the wrong place (which can happen if the accuracy of the underlying data is good enough to say within 1/2m in any given direction.

With anything in OSM someone somewhere will always want to come along and improve it later, and if 500 nodes are used when 5 would do then we're making that person's work much harder.

Login to leave a comment