Please visit the Community link above where this is discussed in great detail. Believe me, I strongly believe in the important role of OSM to be the best pedestrian (and other) router and expended a lot of effort to ensure that the tagging reflected the situation on the ground, and also reflected the availability of accessibility features. The only dispute is whether inconsistently adding little lines on the map alongside every road improves or worsens the ability to compute pedestrian routes.

What local confusion has this caused? The tags on the object made it clear it was on private property and may not be available at all times. It's there, and I feel it should therefore be on the map. Maybe the tagging scheme needs improvement?

The. Data. Is. Wrong!

If you can't see why getting ways wrong creates a fundamental problem with one of the primary use cases of a geospatial database I don't think it's worth continuing the conversation.

Hey, OpenStreetMap, give me a walking route from 23 Balmoral Gardens to 23 Jersey Road on the other side of the road.
osm.org/directions?engine=fossgis_osrm_foot&route=51.502463%2C-0.326765%3B51.502593%2C-0.327133#map=19/51.502408/-0.326554

Yeah, I'm going back to Google maps....

Also if you are going to map sidewalks as separate ways, shouldn't you then remove the sidewalk=both tag from the road, or at least tag sidewalk=separate?

ndrw6, but people are pointing out how this makes the data harder to work with.
It would help if amalash would reply and acknowledge the incorrect mapping and explain the use case they think they are serving, but since this changeset comment they've ignored they've gone on to:
Map sidewalks to nowhere (which can't even be seen in the aerial imagery) in Saskatoon, Canada:
way/1413229098
Regina, Canada:
way/1413221099
ttps://www.openstreetmap.org/way/1413221103
way/1413221105
Toronto:
way/1413211813
North Battleford, Canada:
way/1413183824

I suppose at least in Canada it might be a less confident assumption that there will be a sidewalk, but mapping a sidewalk to nowhere doesn't feel of much use to anyone.

Hi rphyrin,

Yes, absolutely there are circumstances where mapping sidewalks separately records very useful information, and other places where it adds no useful information, makes the map look crap and incomplete, and can actually complicate routing decisions.
I would argue that this particular mapping does the latter. I have married separate sidewalks in scenarios like your example when the situation justifies it.
The mapping is wrong for a start, because footpaths end in the middle of nowhere and don't rejoin the highway even though there are pavements available to continue a journey.
We don't have jaywalking rules in the UK, you can cross a road anywhere, but groups of people for whom specific crossings infrastructure is important are people in wheelchairs who need to know where drop kerbs are and people with visual impairment who may depend on tactile paving. I've already recorded these types of crossings in the area this contributor has added to, they have added more crossings but not even indicated whether they have accessibility features.
It feels like someone helicoptering into an area, adding some incomplete data wrongly to a couple of blocks and then disappearing again, and it demotivates me from contributing further because who wants to contribute time and effort to a crap map? I might be more sympathetic if they could justify why they have added sidewalks to two or three blocks in random cities around the world?

Hi shyamalashanthi!

Welcome to OpenStreetMap! My name is Steven and I do a lot of mapping in Ealing.

I see you've been mapping pavements in Northfields and though you are free to map whatever you like, wanted to check whether you understood how pedestrian routing worked and whether adding pavements was the most productive thing that could be mapped:
- Firstly it is safe to assume that any road will have pavements on both sides unless explicitly tagged in OpenStreetMap, and if you check any of the roads you are adding pavements to, you'll see that it already has "sidewalk" tags indicating for routers whether there are pavements. For example, Erlesmere Road has tags "sidewalk=both":
way/5052558#map=17/51.501547/-0.324901&layers=T
whereas Belvedere Road nearby has "sidewalk=left" because it only has a pavement on one side:
way/165299914
I think one of the best aspects of OpenStreetMap is its power to provide pedestrian and cycle routing, but mapping sidewalks separately adds noise without adding much signal. Mapping footpaths where they *don't* follow roads (like across parks and common land) feels much more valuable. Any pedestrian router will assume that a road has a pavement either side.
- When you add sidewalks manually and separately, they start appearing on the map and does the absence of a sidewalk rendered on the map mean there isn't a sidewalk? Are you ready to add sidewalks everywhere?
- If you add pavements manually, you should be putting a lot of effort into ensuring the crossings are correct. For example, this footway ends without connecting back to the road in two places. It effectively creates a path to nowhere:
way/1411745327

I feel like there are more important worthwhile things to map. For example:
- adding addresses that are missing from houses and offices adds those addresses to Nominatim, the OpenStreetMap address lookup.
- adding buildings indicates how built up an area is, and allows others to add addresses if they can't add houses themselves
- if you are interested in pedestrian routing, then you could add crossing points to roads, for example I have added crossings on some road junctions as a trial:
node/10860481957
By adding the "implied" crossings, it means I can record whether they have accessibility features like drop kerbs and tactile paving. This adds genuinely useful new information that a pedestrian router for a blind person or wheelchair user could use to select a pedestrian/wheeling route. It also means you can use "Overpass Queries" to report on how accessible an area is showing at a glance how rare accessibility features are.
- On pedestrian and wheeling routing, you could also record the surface type and smoothness of paths along canals and across parks. This could be used to inform a router for someone in a wheelchair about the relative accessibility of particular routes so that they can avoid deep gravel paths or really rutted surfaces, for example. See the canal towpath as an example where I've tried adding this information with "surface" and "smoothness" tags:
way/4800268#map=17/51.500338/-0.337583&layers=D
Let me know if you'd like to discuss this further - and welcome once again, I'm so happy to see a new mapper in the area!

Steven

Ah, brilliant, thanks for the tip about measuring the dimensions of the panels. I didn't know any of that. I'll have to revisit the ones I added in the last week or two to check I haven't undercounted lots!
I didn't hear about loss of high resolution imagery, that would be a shame!

Hi! Thanks for your awesome work tagging so many solar panels - I've been trying to add more myself to keep the count up to date.
On this changeset I noticed a module count that looks several times too big. I'm not sure:
1) Whether to cut them down to a single node with a more accurate module count
2) Preserve all the nodes and reduce the module count accordingly
or 3) wait for you to point out something I missed!
These are the nodes in question, all attached to the same building:
node/8188418038
node/8188408404
node/8188418053
node/8188408408
node/8188418033
node/8188418066
node/8188408386
node/8188408410
node/8188418036
node/8188408385
node/8188408370
Thanks,
Steven

"Hanwell Island" returns results from across the Internet including Council publications, which makes it at least a colloquial name even if it is unofficial. I looked up the tagging scheme and there is a "loc_name" tag for names that might be used other than an official name - feel free to update to that. I don't know if will display loc_name on the map though.

I don't remember! It was probably closed at the time I went past, or had sufficiently varied stock I couldn't decide. I can't even find it in a web search.
I think "fixme" is what Everydoor adds when you choose something that isn't a recognised tag. Even without a shop type, it adds value for someone searching for the shop via Nominatim, no?

This was tagged correctly as a service road. If there is a width restriction it can be tagged as such but accessibility by emergency service vehicles is not a criteria for tagging service roads.
---

Published using OSMCha: https://osmcha.org/changesets/164580738

This is an example of a side road crossing I added. I think it gives enough information to a pedestrian router and in an imaginary situation (?) where you could tell a router "I'm in a wheelchair" it could slightly alter routes to avoid places where there are high kerbs.
node/33965546
There are a small number of local residential roads where the sidewalk is only on one side or there are no sidewalks at all so I tag those, and I think that gives routers enough information to know where to tell pedestrians to go. (way/165299914 and way/5052147)
Sorry, long comments for a changeset, maybe I should have contributed to the Community Forum thread!

Hi again

I didn't mean to imply you were doing anything wrong, I've deleted at least one decorative sidewalk myself. There have been discussions in the talk-gb mailing list (are you on there?) about when it's appropriate to map them separately and when they just add clutter and no useful information. There's someone in Acton creating lots of decorative sidewalks and not joining them to roads too, which is really annoying to me but doesn't seem to be affecting routing at least. I had a look at that MapWithAI task but I don't like needlessly adding sidewalks when there's a reasonable assumption that there are pavements on both sides of the road.
Regarding unmarked crossings, I mapped the status of every street corner drop kerb and tactile paving within a couple of blocks of my home and being able to report on it in Overpass Turbo is really useful.
I've also tried using the pic4review site to create a mission for others to use Mapillary imagery to upgrade the tagging of other unmarked crossings (https://pic4review.pavie.info/#/mission/2738). Not sure about the data quality that comes out of that yet - you have to really check the mapillary image location to identify where in the picture (if at all) the referenced crossing is.
Is the forum more active than the mailing list?

Hi Robert,
Is there a place I can look up what the waymap-project is?
I have been creating "unmarked crossings" in my local area in order to record the presence (or absence) of tactile paving and drop kerbs/flush crossings. I like the idea of OSM being able to help create pedestrian routes for people who depend on that infra.

Why did you delete the Blenheim Centre in this changeset?

- Steven

Are you sure about this one? Signage and a row of bollards are visible in recent Mapillary imagery:
https://www.mapillary.com/app/?lat=51.48957330000002&lng=-0.2868645999999444&z=18.49032465646991&pKey=2000001437030905&focus=photo

Thanks. I know it's controversial to delete someone else's work but I've left it for months and they've shown no sign of adding more.
There are specific circumstances where adding sidewalks as separate ways adds useful information, but this isn't one of them.