[Talk-us] Undiscussed mass-revert by user Nakaner-repair

Sat Apr 21 03:04:21 UTC 2018

Thank you Michael for your thorough history of the situation. I think it
clearly explains what has happened and is still happening.

I personally am not sure how to go forward. I suppose that there will
always be mappers who are not following good practices and I suppose
that those who are paid to map are likely to create new accounts and
continue regardless of the negative feedback they receive.

I am not too worried about this type of activity occurring near me as I
have RSS feeds setup to monitor changes and have no reservations about
cleaning up after obviously bad edits especially after posting a change
set comment and getting no response.

That may also be true of other areas that may be affected, but there are
large swaths of the US where the mapping community is thin and nobody is
performing a oversight function. So relying on local mappers to perform
QA is not, at present, a viable path forward.

And, basically the mappers in GB and DE who detected the issue were
performing a QA function. Since the mapping community in those countries
is larger and stronger it does not surprise me that this issue was
detected there first.

It seems a project wide method for reducing this type of activity is
in order. I am not worried about paid versus volunteer: I couldn’t care
less if a mapper is paid as long as they make good edits.

I will be traveling to northern India in a while and, as usual when
traveling to a place I haven’t been before, I have taken a look at maps
based on OSM to get a feel for the area. My impression is that the
completeness and accuracy of OSM data in that area is worse than where I
live and usually map.

This gave me the idea that maybe a way to get new mappers, paid or
volunteer, to learn the craft might be to restrict them to editing only
in the geographical area their registration IP address indicates they
are in until they have some number of edits. In this case that would
mean they’d be editing in India or Pakistan until they’ve got some
experience. Two birds with one stone as it were: First their local area
would become better mapped. And second, when they start editing in more
distant regions they might value the account they worked to setup enough
to respond to comments, etc. so that it won’t be blocked.

Within the OSM profile is a field for where you are located. If the IP
address is not feasible, we could restrict edits to a radius around the
declared location and not allow the location to be changed frequently.

I suppose the HOT team might not like that setup as it would cut into
their new volunteer base. But perhaps any geographical restriction on
editing could be cut short if the new mapper has an experienced mapper
who is listed as a mentor.

Just tossing out ideas for a possible general case solution. Feel free
to shoot them all down. :)

-Tod
n76 on OSM

> On Apr 20, 2018, at 4:00 PM, Michael Reichert <osm-ml at michreichert.de> wrote:
> 
> Hi,
> 
> Am 20.04.2018 um 17:13 schrieb Ian Dees:
>> I noticed that user Nakaner-repair just reverted 1000+ changesets
>> throughout the United States without any discussion in the local community.
>> Nakaner-repair points to a thread in the German forum [0] that seems to
>> indicate that they think these edits were made by paid mappers. Having not
>> heard from those suspected paid mappers, they went ahead and reverted
>> without discussion from the local community.
> 
> TL;DR A group of mappers (presumeably from UTC+5) started adding
> highway=service and doing wrong quality assurance on roads in Europe and
> the U.S. in mid/end of March. British and later German mappers found
> these strange edits last week, some German searched for more accounts,
> SomeoneElse and myself wrote changeset comments and SomeoneElse (DWG)
> blocked many on them.
> 
> Unfortunately, the 0-hour user blocks are not as useful as they are
> usually (and I expected them to be). They have been ignored and they
> continued editing or created sockpuppets. Longer blocks were ignored and
> they continued editing after the block. Changeset comments were not
> answered or the response did not answer the question.
> 
> Since this week, they don't do any QA on roads any more and only add
> highway=service in the U.S. They create new accounts if their old
> accounts are blocked. This pattern now repeats day by day and the last
> resort was to revert their work because that causes financial damage (I
> hope they get paid).
> 
> Please see some inline comments/responses on Ian's questions below.
> 
> -----------------
> The full story:
> 
> On 2018-03-29 Will Phillips writes to the Talk-gb mailing list that he
> observed "a series of edits with users removing or changing access and
> oneway tags". He describes the quality of these edits as "very poor". At
> that time "none of them has yet responded to changeset comments".
> https://lists.openstreetmap.org/pipermail/talk-gb/2018-March/021259.html
> 
> SomeoneElse suggests him to write an email to data at osmfoundation.org
> (the DWG).
> https://lists.openstreetmap.org/pipermail/talk-gb/2018-March/021260.html
> 
> I am not subscribed to Talk-gb and did not notice it at that time.
> 
> On 2018-04-09 tux67 creates a new thread on the German forum because he
> found two users (sri_harsha and Premsakhare) editing roads globally
> without local knowledge. He asks other mappers to review their edits.
> https://forum.openstreetmap.org/viewtopic.php?pid=693849#p693849
> (contains changeset links)
> 
> Premsakhare had received multiple (!) user blocks at that time. The
> oldes block was created after the discussion on Talk-gb.
> https://www.openstreetmap.org/user_blocks/1831
> He had been blocked because he did not answer changeset comments asking
> for the sources being used. The block was removed automatically when
> Premsakhare read it. Premsakhare uploaded more changesets but did not
> answer the question, SomeoneElse blocked him again, the block expired,
> Premsakhare started editing again, SomeoneElse blocked him again, …
> https://www.openstreetmap.org/user/Premsakhare/blocks
> 
> One hour after tux67's posting on the German forum, user whb commented
> that he observed multiple mappers editing roads and intersections
> without local knowledge. He found patterns (editor, locale, mapped
> features) making it possible to group them. He suspected that they are
> accounts of employees of "Mapbox or at least any other company".
> https://forum.openstreetmap.org/viewtopic.php?pid=693862#p693862
> 
> I asked Mapbox a few hours later. They denied (and I believe it). Mapbox
> has been stating clearly if an accounts belongs to them.
> 
> Other users comment on the thread sharing the results of the reviews
> they did in their local mapping areas. We (users at the forum) came to
> the conclusion that they resolve some kind of routing "errors". However
> these errors are not always errors. They erroneously
> 
> - made oneway streets bidirectional against local surveys and Mapillary
> (e.g. https://www.openstreetmap.org/changeset/57929974 and
> https://www.openstreetmap.org/changeset/57930358)
> - added/deleted turn restrictions
> - removed access restrictions or added them although Mapillary imagery
> was available at some locations
> (https://www.openstreetmap.org/changeset/57959190
> - edits as if OSM were a map for cars, not for any type of vehicle
> 
> I have not fully checked all changesets but the users reporting that
> changes are trustworthy and experienced.
> 
> User wambacher assumes that the mappers come from India or Pakistan
> because their user names look like other user names from that region.
> https://forum.openstreetmap.org/viewtopic.php?pid=693896#p693896
> A few days later a mapper wrote a changeset comment using the Bengali
> script.
> 
> The origin has not been proven but the times of activity are quite
> interesting. They seem to be located in UTC+5 (or near that time zone)
> and have five days of work per week.
> 
> On evening of 2018-04-09 and the day afterwards further user accounts
> were found which had similar editing patterns.
> 
> Most changeset comments have not been answered at all. If any answer was
> written, most questions has been left without any answer or indicated a
> lack of training. For example, see
> https://www.openstreetmap.org/way/26996470/history and the comments at
> https://www.openstreetmap.org/changeset/57793900
> 
> On 2018-04-11 a group of mappers (myself, tux67, whb) asked a well known
> person to run some queries on their database to search for other
> accounts which belong to this group using quite fancy criteria. A huge
> thanks goes to all those we found before because they were used to
> optimize the query. :-) We found many of them. A similar search was
> repeated later.
> 
> User list on 2018-04-11	22:16 UTC:
> https://forum.openstreetmap.org/viewtopic.php?pid=694208#p694208
> 
> I asked the DWG to ask the OWG to look for patterns in the email
> addresses and IP addresses used for signup on 2018-04-11 but have not
> received any response. I myself doubt if it will really bring us forward
> given the recreation of sockpuppets we can observe. But I am still
> interested in the results.
> 
> At that time I started posting a changeset comment to any account I
> found. The comment asked who pays/directs them and to add that answer to
> the user profile page. In addition, I asked for the host of the iD
> editor they used because they used a quite old iD at that time.
> 
> I usually received no answer at all. SomeoneElse (DWG) blocked all of
> them who did not answer my question and uploaded more changesets more
> than a few hours after my comment. He might have overseen some users but
> he kept an eye on the blocked users and re-blocked those who continued
> after their first 0-hour block.
> 
> On 2018-04-12 mvvieka answered such a comment
> (https://www.openstreetmap.org/changeset/57994668). That's the whole answer:
>> HI , i'm not working for any organization. Im just a freelancer and
>> im just making edits as per the available imagery .
> 
> Another mapper wrote
> (https://forum.openstreetmap.org/viewtopic.php?pid=694422#p694422):
>> Freelance Map Analyst currently working with Point Inside
> 
> I continued searching for new users this week and it began to become a
> regular task like writing welcome messages to newbies.
> 
> I found out at that time that some users downgraded motorways to trunks
> in the U.S. I planned to write to Talk-us because I myself am not sure
> what mappers in the U.S. define as highway=trunk.
> 
> List of users as of 2018-04-17:
> https://forum.openstreetmap.org/viewtopic.php?pid=694878#p694878
> + https://forum.openstreetmap.org/viewtopic.php?pid=695171#p695171
> + https://forum.openstreetmap.org/viewtopic.php?pid=695215#p695215
> 
> I reverted all changesets of all known users in Europe, Africa, Asia and
> Australia/Oceania on 2018-04-18 because the edits in that area where
> mainly routing "QA" edits containing many errors (see above). That
> revert was driven by their mistakes, not of them being unresponsive. I
> decided to exclude North and South America at that time because the
> edits there were mix of bad QA and addition of highway=service (good)
> and the trunk vs. motorway/primary topic can be highly controversial
> (see recent discussion on Talk-us).
> The first (non-US) revert: https://www.openstreetmap.org/changeset/58217008
> 
> Once the mappers received a user block by SomeoneElse, they stopped
> editing but new accounts were created on the next morning (UTC+5). This
> happened every day this week. The choice was to either accept that OSM
> is overrun by a army of sockpuppets who ignore us at all or to make the
> time/money they invest into editing a waste of time/money by reverting
> their edits even if it removes good contributions. That's sad but what
> is the alternative?
> 
> It is up to you to decide if you want to be overrun or if working
> communication with other mappers is important to you.
> 
> To see more accounts appearing next week, just watch Pascal Neis' list
> of new contributors and the list of my changeset comments. Btw, many
> spammers pop up there, too, but that's a different topic.
> 
>> Some questions:
>> 
>> Was this action made under the auspices of the Data Working Group?
> 
> No. Neither I am a member of DWG nor the changeset comment contains a
> "on behalf of DWG". However, I requested requested lots of user blocks
> in this case already.
> 
>> Has the "directed mapping" policy been approved by the OSMF?
> 
> The policy itself has not been formally approved. However, the survey
> [1] shows that the community expects organised editors to inform the
> community that they belong to [organisation]. There is no formal policy
> yet but that doesn't mean that they can do what they want to do. If it
> is against the interests of the community, it's against the existing
> unwritten rules. The policy will mainly document good practice and
> established rules.
> 
>> I'd be interested in seeing all of these reverts reverted (at least in the
>> US) until discussion can take place.
> 
> I will pause reverting their edits in the U.S. Please keep in mind that
> some users of that group did "road QA" in the U.S. which might be as
> good as their QA in Europe. If you ask me to revert it, I will do it.
> 
> Best regards
> 
> Michael
> 
> [1]
> https://wiki.osmfoundation.org/wiki/Data_Working_Group/Results_of_Organised_Editing_Survey_2017
> 
> 
> -- 
> Per E-Mail kommuniziere ich bevorzugt GPG-verschlüsselt. (Mailinglisten
> ausgenommen)
> I prefer GPG encryption of emails. (does not apply on mailing lists)
> 
> _______________________________________________
> Talk-us mailing list
> Talk-us at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-us