We share guidelines with our team based on local conventions and take action to rectify any concerns based on community feedback and OSM standards.
We received some community feedback here about mapping sidewalks separately along residential roads. I want to respectfully align with the community based on mapping practices; however, there are examples outside of my direct team where sidewalks are mapped separately along residential roads. overpass-turbo example. Can someone point me directly to community policy on when it is appropriate to map sidewalks separately along residential roads, like in the examples shown?
Thank you for the question. No, the guidelines do not involve visiting the areas being mapped; however, I recommend the following to my team:
Use street level imagery and aerial imagery to verify edits.
When in doubt about making an edit, leave it to the local OSM community to address.
With that being said, please feel free to map with us. It seems like you have a lot of expertise in this area based on your OSM profile. The more the merrier!
Adding sidewalks as ways in places like that has rather limited utility, and breaks some intuitive uses like being able to easily cross the street anywhere. (For example, getting from 37 Downsell Road to the allotments)
Thatās a shame. While itās certainly possible to have a go at mapping things remotely somewhere like this (click the edit button there and youāll see that everything is nicely separated and the area is flat), itās much harder here (click āeditā there, and you can see that the buildings overlap the roads in the imagery - either from the west or the east).
Unfortunately, a former Facebook mapper claimed not only that they could see a separate sidewalk there, using the Bing imagery (where the building at the west completely obscures it) they could also see that the surface was āpaving stonesā. Clearly something doesnāt add up here.
You claim that you have a āproject for Londonā. According to your OEG page London was āCompleteā in āNovember 2023ā, and then for some reason a āLondon, UK (cleanup)ā activity was immediately needed, which was then āCompleteā in āNovember 2024ā, but another one is now āIn progressā again as of last month.
If I was to give advice to anyone about āhow to map Londonā (or Manchester, also on your list) the zeroth item on that list would be āvisit the place, even if only to get a feel for how things are joined together thereā.
This is Facebookās third go at London - may I suggest that you need to figure out a way for someone to āactually go there and have a lookā, otherwise I can see that a fourth, fifth and more rounds of this will be needed.
PS: For completeness, Iām a DWG member but the āyouāll do a better job if you go there and visitā recommendation is made in a personal capacity. Iām currently handling a somewhat related DWG issue (not in London) so have not referred directly to that here.
I used to live just off the left of that map excerpt and am still local. These are the worst possible type of street for mapping with separate sidewalks, as theyāre Victorian terraces with small front gardens and very few driveways to help pedestrian routers cross the road without a very long detour. I did add a few separate sidewalks on similar residential streets just to the south of there a few years ago, but realised it was a bad idea and removed them.
Iād be much happier if some of this effort was directed at adding sidewalks and crossings accurately to busy main roads where it would actually help routing for pedestrians and users with mobility and visual impairments.
Coming back to another āWalkaboutā project, which I suspect is amongst @SomeoneElseās current DWG headaches, I came across this. Mapping crossings this badly, or having two crossing ways and two sidewalk ways join at the same barrier=kerb nodes is not helpful. You generally donāt need to traverse a kerb to turn a corner on the same side of the street and if itās done that negligently, weāre better off with sidewalk:$side=* tagged on the parent highway because that doesnāt involve lying to data consumers.
Is there any particular reason why some mappers associated with this project are replacing sidewalk=both with sidewalk:both=yes? Theyāre both valid, but Iām struggling to see the point. Is this a suggested tag āupgradeā which Rapid is trying to inflict on us?
Also, replacing crossing:markings=zebra;dots with crossing:marking=zebra when the presence of the TSRGD diagram 1055.1 dots was established from a survey or imagery isnāt particularly helpful. It might help if they familiarised themselves with Key:crossing:markings - OpenStreetMap Wiki first.
they could also see that the surface was āpaving stonesā.
@SomeoneElse I have commented on the paving stones tags changeset as of yesterday: Changeset: 161983054 | OpenStreetMap . The account you are referring to is also archived. Furthermore, the mapperās edits are not apart of this project.
The project we are engaged in is different from the other two London projects listed on our wiki. This project has a different focus, which is why it is important that I publish information about it here.
Respectfully, many people do not have the privilege to visit where they have mapped. When in doubt about making an edit, I advise my team to allow the community to address it. I wholeheartedly agree that on the ground knowledge is important.
Thank you for your feedback. I strongly encourage collaboration. Please feel free to map with us.
replacing sidewalk=both with sidewalk:both=yes? Theyāre both valid
If sidewalk=both is preferred, we can definitely use this tag instead. Iāll share the feedback on crossing:markings=zebra;dot with the team as well. Thank you!
No and I respectfully would not want it to be interpreted that way. āWhen in doubt about making an edit,ā do not make the edit. We are using this thread to intentionally connect with other mappers and asking for advice.
If that could be rephrased as ārecognise those places (such as London and Manchester) where remote-only sidewalk mapping is difficult or impossible and not map in those placesā that would be something that I would personally support.
Local communities rarely object that an organisation such Facebook has not mapped somewhere remotely, but they often do if they have and have made a bit of a mess of it.
The sidewalk:<side>= (I will refer to as āsidednessā) schema is objectively better than the sidewalk= (I will refer to as āoldā) schema:
Removes the possibility for ambiguity (sidewalk=yes and sidewalk=separate both are necessarily ambiguous with respect to communicating sidewalk presence by side)
Allows for much more flexibility (sidewalk:right=no+sidewalk:left=separate cannot be reduced to a sidewalk=* tag without losing information)
Enjoys support from powerful and widely used editors such as StreetComplete/SCEE [Edit to add: This last point is not entirely correct - see below correction by @rskedgellhere!]
Even if you oppose separately mapped sidewalks, the sidedness approach for roadway tagging of sidewalks is clearly superior to the old schema.
You could have checked yourself, rather than implying some kind of malicious intent on behalf of Meta/Rapid. I donāt see a validator warning or suggestion for changing sidewalk=both in the currently deployed version of either iD or Rapid. (Though, I would happily support such a change to add this. Thereās some additional info and discussion Iāve been involved in on this iD Issue)
They do. From their OEA page:
Our process includes a suite of OSM tools such as Tasking Manager, Map Roulette, Rapid, JOSM, and street-level imagery from Mapillary and Bing.
Additionally, they link to a video by @geocruizer which explains how to use approved street-level imagery sources to map pedestrian infrastructure.
Even cherry-picking one of the few segments around that location without directly-along, close-up, post-renovation street-level imagery (section of George St east of One St Peterās Square), the sidewalk is visible in both the Esri Wayback 2019-10-30 imagery layer (brightness set to +185% for screenshot) and in Bing Streetside:
Of course, I agree that the source=Bing tag on that specific changeset should have been more informative.
Still, the assertion that sidewalks cannot be mapped with separate geometry by remote mappers using approved sources of aerial and street-level imagery in London is yet to be supported.
As shown above, physically visiting a place in order to develop an understanding of its pedestrian infrastructure norms is not necessary; it is entirely possible to accomplish this to a reasonably acceptable degree with the use of the myriad open street-level imagery sources (Mapillary, Bing Streetside, KartaView, Panoramax, Mapilio, Wikimedia Commonsā¦)
Why does it not make sense for there to be ongoing cycles of mapping and refinement of pedestrian infrastructure in a major city? How is that indicating that the approach to organised mapping is wrong?
I donāt want to just reiterate the points already made in the Separate sidewalks (or not) near Ealing thread, but I find it very telling that this post of mine was never responded to:
Separately mapped sidewalks and tagging of sidewalks on roadways with the āsidednessā schema is good. Obviously, that doesnāt mean that every edit on a project to do this is good. Where and when mistakes are made (and to be clear, mistakes have been made, and not just in London ā I remember āfightingā a bit with mappers from Meta about unmarked crossings in Seattle, and then coincidentally meeting that Meta employee I had argued back and forth with at SOTM US ā25, which was genuinely lovely) they should be corrected and the processes used by organised editing teams contributing the offending changesets should be transparently improved. Thatās exactly what I see happening in my previous discussions with Meta employees as well as this thread and in the relevant changesetās comments, so I commend those working as part of the process.
Given that we (TCAT) run our own very similar ā though much, much smaller in scale ā organised editing effort via OpenSidewalks, I intentionally post this with my āwork hat onā and with the genuine and in-good-faith hope of moving the discussion forward and being a āgood citizenā in the world of OSM. If thatās not being accomplished, please do let me know. Thanks.
How do you know? Have you actually tried going somewhere and said to yourself āno, thereās information that I could get here that I couldnāt get by looking at a picture of it, so Iāll go back behind my computer and do thatā?
When I map things in OSM, apart from āIāll just update this thing with StreetComplete / Vespucci because Iām actually hereā itās usually from a combination of sources. There will be a GPS trace, there might be some photographs, thereāll be some notes (actually theyāll be attached to waypoints in the GPS traces) and thereāll various sorts of aerial imagery. Most importantly though, thereāll be the recollection of what it actually was like when I was there. I have learnt through experience that while aerial imagery can be useful, it can get out of date, and a path that looks as if it might join actually does not, perhaps because of a thin fence, or a layer difference, or something else.
In OSM we should use all of the sources we can. Sometimes (perhaps North Korea) survey is not an option, but in major European cities it absolutely is and an attempt to map without using the best source available is bound to lead to suboptimal results - as has been seen from the complaints in the forum threads that preceded this one.
Be that as it may, any organised editing effort that aims to replace one tag with a supposedly superior other tag without adding information, such as this change, should clearly include this tag replacement plan in the requisite prior discussion with the affected OSM community so that its merits can be discussed before instead of after the fact.
Those ambiguous tagging variants donāt exist in Greater London. Iām pretty sure that @RedAuburn and I made sure of that.
True, and Iāve only used it thousands of times before. Itās still completely and utterly irrelevant to replacing sidewalk=both with sidewalk:both=yes and doesnāt answer my question about whether this change is an editor driven āupgradeā or not.
Except that they will write sidewalk=both, sidewalk=left and sidewalk=right where thatās more concise. You could have checked yourself.
No, but as I actually use pedestrian routing when planning and sharing running routes in London, I oppose people doing it badly. Itās almost as annoying as tedious whatiffery.
This is the same user who repeatedly mis-tagged unmarked pedestrian crossings in the UK which were next to give way markings as crossing:markings=dashes and doesnāt respond to changeset comments or clean up their mess? This does not fill me with confidence. In any case, if you canāt recognise the UKās give way road markings for what they are, you probably shouldnāt attempt to map highways in the UK from aerial imagery.
