[OSM-talk] Follow up on last summer discussion about the Automated Edits Code of Conduct and the DWG

[Victor Grousset/tuxayo]

Hi,

On 2017-04-21 08:18, Roland Olbricht wrote:
> Thank you for keeping track of the issue. But I deem the summary
> reflects neither the current situation nor the fidings of the discussion.

You are right, it was a legacy of how this page started.
But now, it's misleading and the intro isn't enough to clarify that.

So I changed the name, the new URL is
https://wiki.openstreetmap.org/wiki/User:Tuxayo/Automated_edits_code_of_conduct_and_DWG:_Follow_up_to_mailing_list_discussion_and_proposals

Is that correct for you now?

> Some key points:
> 
> * There is no consent on what an automated edit is or not.

Indeed, I thought it was not related to the topic but in fact it is.
It seems a lot of disagreements are whether an edit is automated or not.

> It is pretty clear that your example (changing all phone~"^http://" to
> "https://" worldwide) is an automated edit. The grey cases are things
> like the French buildings import, the MapRoulette challenge in the
> Antartic region, and even the edit without local knowledge of Passau
> main station (hence a pretty small changeset) of our company.
> 
> All of these edits have at least made some data worse and have therefore
> been discussed and partly fixed, partly kept for a reason. The fact that
> the word "automated" did cause confusion gave rise to the
> https://wiki.openstreetmap.org/wiki/Organized_Editing_Policy

Thanks a lot I didn't know there was such a policy.

> The two extreme positions are
> - Any edit without local knowledge is by its nature flawed.
> - We regulate only edits run by a bot.
> 
> I personally (or we as a company) do not endorse any of the two extremes.
> 
> They key point is that to be productive you should:
> - define and publish your own criterion (e.g. one of
> -- changesets of unusual large extent
> -- unusual high activity per tag and day
> -- changesets having "revert" in their comment)
> - give it a specific name and set up a watch tool for it

These are interesting ideas for monitoring tools.

> * The DWG is not so special as you might think
> 
> The DWG members are indeed special in dedicating huge amounts of time to
> fix human misbehaviour, and we should be grateful for that. The DWGs job
> is communication, not pushing around data.
> 


> Most of the actual reverting is done by mappers outside the DWG. 

That's good to hear, how do you know that? I though a lot of people
would report issues to the DWG instead of reverting themselves.
Maybe it's only the case so automated edits?

> Also,
> DWG members do not have any special rights. Moderation (and possibly
> redaction) is essentially done by the sysadmins, not the DWG.

Aren't DWG members moderators? Which means they have the permission to
block an account.
https://wiki.openstreetmap.org/wiki/Web_front_end#Moderators
https://wiki.openstreetmap.org/wiki/Data_working_group#User_Blocks

> I agree that from outside, the DWG activity is hard to judge. The
> problem here is that nobody has found a magic solution how to make DWG
> activity public without asking the DWG for substantially more work,

Would an issue tracking system suits this situation?

> damaging the reputation of involved mappers, or both.

Oh, good point. That indeed seems to make this impossible to solve
without magic :(


> I therefore would suggest to make clear-cut rules:
> 
> a) If you can decide freely what to map, where to map, and how to map
> then OSM will trust all your edits that are based on local survey. Happy
> mapping!
> 
> b) If you are directed by an organization (regardless whether you are
> paid or voluntary) then use a dedicated account and put a line on your
> user profile, e.g.:
> http://www.openstreetmap.org/user/drolbr_mdv
> That organization should have a corresponding Wiki page, e.g.:
> http://wiki.openstreetmap.org/wiki/MENTZ_GmbH
> 
> c) If you run a software where you do not approve as a human every
> individual edit (every single change of a tag or change in geometry or
> topology) then you need to follow the Automated Edits Code of Conduct
> https://wiki.openstreetmap.org/wiki/Automated_Edits_code_of_conduct

That's great. It seems very clear and I don't see much ambiguity.

> This still leaves open the case of Armchair Mapping of all shades.

Indeed, this shows that Armchair Mapping is orthogonal to the 3 above
categories.

> An example with net benefit for OSM is MapRoulette. Therefore I would
> suggest to ask Martijn first for his best practices and then start to
> make rules on that.

Ask about what exactly? About how to avoid the issues with armchair mapping?


Cheers,

-- 
Victor Grousset/tuxayo