Thanks! (Btw I think I mapped your place today: osm.org/node/1829661485 )

Hi homeslice, thanks for joining us -- I shouldn't have given up so quickly! And thanks for being clear about your goals and MO.
As I said, I've admired your sidewalks. But I hope I've also demonstrated the downside this mapping technique has on the pedestrian routing.
When sidewalks are fully mapped, of course, the pedestrian routing improves, and, as bhousel mentioned, it creates opportunities for detailed mapping of the sidewalk ways that may assist wheelchair users, etc. In the mean time, though, these islands with loose ends will break pedestrian and wheelchair routing for years to come. There are over 4000 intersections in Brooklyn, and a ballpark estimate is that it would be 6 or 7 years before you're done. Of course, the process may speed up as others chip in! But we're looking at years of broken routing nonetheless.
I'd ask you consider connecting the loose ends to the road network so the presence of the sidewalks does not negatively affect the routing.
Offhand I feel that capping the loose ends with the four-crosswalk way like I added at Fulton & Ashland is the best way to transition between a sidewalk-mapped block and its sidewalkless neighbors. And if it is in fact your goal to completely map the Brooklyn sidewalks, then it's something you'll want to do eventually anyway,
I'd love to hear other suggestions if you or anyone else has them, though, since I do a fair amount of this myself (connecting paths and sidewalks to the grid) and I want do what's in the best interest of the map.
Thanks again for joining us, J

Hah, thanks! So worried about getting the "ae" right that I flubbed the rest of the word!

Well I added my sample "cap" at the Fulton/Ashland place, changeset 61672263. If there's ever a consensus to purge these loose-end sidewalks that survives the edit wars, please feel free to get rid of that too -- it's only useful IMO as a fix to routing problem the loose ends cause.

Homeslice60148 has not responded to my invitation to join this discussion... Other than following homeslice around and patching the loose ends (which I'm not inclined to do) I don't think there's anything to be done about it, and the pedestrian routing will be damaged for the foreseeable.

(If I have too much coffee someday I might bring this up on one of the mailing list, with the aim of improving the Sidewalk_as_separate_way section of the wiki.)

J

Happy to hear it. And MikeN. by the way, I believe I misread what you meant by "end at the curb" -- yes indeed, the sidewalks themselves do literally end at the curb, crosswalks or no, and I can easily imagine a mapper choosing to map them in this literal fashion without realizing that it affects the routing.

BTW this changset (61631661) has been reverted by changeset 61662763 (restoring the deleted sidewalks), and the loose sidewalk ends have been tied into the street grid by changeset 61662897. This will improve the routing. It will still be a left and two rights to get from 691 Fulton to 239 Ashland Place, for instance, but I think we can all agree that it's an improvement over this:
https://imgur.com/50v4S2R
What I'd suggest as a standard would be this:
https://imgur.com/PZGVrHO
Routing would still have some little wiggles of course, no way to prevent that, but there'd be no real wrong turns or backtracking. (And I'll go ahead and upload this in a bit.)

I don't think these are being mapped this way because they stop at the curb -- there are clearly visible crosswalks at all of the hanging ends. I don't know whether homeslice60148 prefers to map this way or if it's part of the maproulette instructions (? I don't know how maproulette works. You, Mike, seem to have some experience though!) I messaged homeslice60148 regarding this changeset discussion, so maybe we'll see.
I don't think we're going to be able to come to any agreement on the sidewalks-as-ways issue anytime soon, but I do like the idea of basic guidelines for how the sidewalk ways should be tied into the street grid when they *are* mapped, to avoid routing problems.

These comments are not a personal attack and your defensiveness is uncalled for. I'm discussing the changeset. If you don't want to part of the changeset discussion, hit the "unsubscribe" button.

BTW here is a very silly example of screwed up pedestrian routing. (Wheelchair routing would have the same problem, I believe.)

osm.org/directions?engine=graphhopper_foot&route=40.60721%2C-74.01134%3B40.60740%2C-74.01154#map=19/40.60770/-74.01090&layers=N

Personally I only map sidewalks when I perceive that it would improve routing, but I can easily imagine that at some point in the future mapping every sidewalk will be the norm. I'm not inclined to fight against this -- but I do want people to map responsibly, so that the map isn't less useful after their contributions than it was before. Maybe osm.wiki/Sidewalks#Sidewalk_as_separate_way needs some good idea/bad idea examples for how to tie mapped sidewalks into the existing street grid.

None of these #maproulette #Long_Island_traffic_lights_(Brooklyn,_Queens) sidewalks have any wheelchair or smoothness info. I've actually never seen any such info on sidewalks here. (I have seen sloped_curb=yes on some *road* intersections, but that's obsolete now.) Yes, it might be added someday, but it might be added someday via sidewalk:* tags on the road as well -- that's the very first example given in the Sidewalks wiki page. Doesn't seem "too cumbersome." Regardless, the state these are being left in is an active impediment to both pedestrian and wheelchair routing. It can't really be justified by saying "maybe someday someone will want to add wheelchair tags."

There are dozens of these around Brooklyn and Queens, all added by homeslice60148 and tagged with #maproulette and #Long_Island_traffic_lights_(Brooklyn,_Queens)

Aside from the facts that 1) they're generally not in places where sidewalk mapping makes any difference to pedestrian routing, and 2) they're left unconnected on the ends, so they're actively problematic for routing... they seem really well done. So I've left them alone, figuring it's a work in progress, and I've never tried to contact the mapper.

Do you think these should all be purged?

I have no experience with maproulette, but is this something that can be fixed or prevented from that end?

User did not respond to changeset comments or personal message. Reverted in changeset 61505468 to fix corrupted building geometry, POI node created for medical centre.

Reverted in changeset 61505468 to fix corrupted building geometry, POI node created for medical centre

osm.org/node/5764129854 incorrectly shows a bank branch inside the Brooklyn Public Library in NYC. Telephone number has a Saskatchewan area code.

Hi, this looks like an accidental change to the geometry of these buildings. The old footprints were correct -- at least, they were far more correct than they are now.

Can you revert this? Thanks, J

Hi Beni_k -- this restaurant is already on the map ( osm.org/way/279772352 ) and that location seems to correspond better with the address 557 Driggs. It appears your node may be on the wrong side of the avenue. Please re-check and consider deleting this new node and making edits and additions on the existing way 279772352 instead.

Thanks, good catch!

Hi Mission_Control -- welcome to the map & to the neighborhood.
N.b., Slope Natural Food, which you added at 77 5th Ave, is actually already mapped one block north at 61 5th Ave. The shop at 77 5th Ave is currently vacant. I changed 77 5th to "disused:shop=yes" and copied your payment tags onto the existing shop node.
( osm.org/changeset/59984521 )

Hi, thanks... pharmacy=yes makes sense to me in the same way as adding bar=yes to a restaurant or bench=yes to a bus stop. Physically, the pharmacy counter is part of the same shop: same address, same phone number, and also serves as an additional checkout area for items from the main shop. It looks and feels like part of the shop, not a separate business operating from the same premises.

I saw pharmacy=yes suggested as a combination with shop=chemist in the talk page for pharmacy on the wiki. Browsing through taginfo, it looks like it's much more common to add amenity=pharmacy to the chemist object, rather than pharmacy=yes, so I've made this change (changeset 59951758.)

That said, if this Duane Reade looks and feels to you like two separate features, feel free to map it as such -- I won't take offense. (This approach does have some advantages, like simpler encoding of the opening_hours.) And if there's an existing consensus about mapping this style of shop/pharmacy that's been discussed and documented, please point me in that direction. I read through the wiki, the July 2016 thread on the tagging mailing list, and some related issues on the iD github page, and there's no authoritative consensus that I can see. But there are over 200 Duane Reade stores in the NYC area (and, including other brands, many thousands more in the USA) with essentially the same design, so it would be great to have them tagged in a consistent way.

(As for the ATM, tagging shops and amenities with atm=yes is a well-established practice.)