[Osmf-talk] Iterating the Alternative Plan

Wed May 17 22:01:09 UTC 2023

> From all the data I’ve seen, addressing is the biggest gap between OSM
and any other widely used map. It’s arguable that PoI’s are the main gap,
however, fixing addresses would tie many PoIs to a location via their
address, but if we focused on PoIs then it wouldn’t help addresses. I’d be
super interested if anyone has a bigger “hole” in OSM than either of these.

addresses (at least at street level, not house): likely the major source in
most countries might be the government itself (often some ESRI shapefiles).
While it would still be a good idea to optimize for conflation with help
from people of other countries, let me give a hint of what helped in
Brazil: people from the OpenSreetMap Brasil setup years ago a layer to be
accessed via iD and JOSM. No idea about agreements, but the end result is
stored on MapBox and is free for mappers. Maybe similar to *every* country?
Must have others which do not have street names, but exist reference
government files!

PoIs: while some kinds of PoIs the government could still be the primary
source (but often is poorly geocoded, so metadata is rich, but position
not) this is likely to truly require survey, in particular the first time
(because at least help know the location). Using Brazil as an example, some
PoIs (n speciali healthcare and schools) there's decent metadata with poor
location precision (in the USA, it seems that they're more lucky on exact
location; I mean, for PoI with gov have data, this will vary a lot between
countries). But anyway, for PoIs sometimes the very specific types are
still feasible with Open Data (but may need more cleanup) but it is better
to assume they would require survey. We could focus on how to use more of
these apps, and maybe also tutorials on how to create plans to survey
several places which may have a PoI or not. And, obviously, as the person
travels, leave some promotional material to invite people to "put their
city on the map" (or better wording).

Bigger hole? Aka natural=wood? Ok, if you really think about it, maybe
there's other "holes" on huge chunks of space without anything mapped, but
often would be wood, types of terrain, desert, etc. Oceans (to my
knowledge) already are mapped. Maybe there will be people wanting to map
these large chunks of blank space. But addresses and PoIs are still
reasonable to do it and, by far, PoIs the largest effort (which also
implies increasing mappers over time). But different kinds of features
requires different approaches if want to speed up.

That's my 2 cents for now. But these don't require changes in the site,
maybe some planning ahead and incite mappers to go out and map. But the
idea of looking at how countries are doing and helping others make sense to
me. I cited Brazil, but sure there are others being creative.

Att.
Rocha

Em qua, 17 de mai de 2023 17:02, Steve Coast <steve at stevecoast.com>
escreveu:

> Dear all,
>
> Summarizing here some takeaways from the public list and private
> conversations on changes to make to the plan and next steps, this is really
> a grab bag of things I’d love feedback or even help on:
>
>
> * Engineering changes in general:
>
> Questions about the role of OSMF and the website are intertwined. I’ve had
> feedback to focus on “one line” changes like turning on map notes by
> default. I’ve had feedback to write code code myself, to have the OSMF do
> bounties, to run bounties myself, to giving up entirely on OSMF and having
> a parallel organization focused on technical support for all the background
> and foreground work that happens. All of these have pros and cons. A
> recurring theme is “why work with the OSMF” - I argue it’s important as
> mappers and others look to it for leadership and authority. It’s a common
> theme on engineering changes - people look to OSMF to authorize a change
> because there are downsides and someone has to take the responsibility for
> them. For example, if we make the map “worse” by putting map note pins all
> over it, some people will complain and it shouldn’t fall to any particular
> engineer to deal with that. I think OSMF has a role and we should find a
> way to build a smart, simple, actionable plan for the map.
>
> * Remove the conference and local member payments back to OSM
>
> I’ll remove that from the wiki version of the plan.
>
> * Should we focus on addressing?
>
> Is addressing this really the biggest problem, and therefore should it be
> the focus?
>
> I really don’t care what we focus on, but I do care that we focus and make
> the map better.
>
> From all the data I’ve seen, addressing is the biggest gap between OSM and
> any other widely used map. It’s arguable that PoI’s are the main gap,
> however, fixing addresses would tie many PoIs to a location via their
> address, but if we focused on PoIs then it wouldn’t help addresses. I’d be
> super interested if anyone has a bigger “hole” in OSM than either of these.
>
> * Changing the map to not show roads which don’t have addresses:
>
> One person suggested don’t make the map just go blank, instead we render
> roads with addr:complete as they look today, but if they don’t have that
> tag, then render them translucently so that they are still there and usable
> but obviously wrong and need work. I’ll change the plan to this, it seems
> like a great compromise. Another suggestion was to make roads translucent
> that have any address data near them, but aren’t marked complete. This
> seems like much more engineering effort than a simple rendering switch, as
> some process has to be built, run and maintained to find these roads and
> then integrated in to rendering.
>
> * Other simple changes:
>
> There are a variety of third party volunteer, supported and charter tools
> (maproulette, Neis’ OSM stats, OSMCha, mapwith.ai… ) which would have
> vastly more power and usage if they were integrated in to osm.org. These
> can be
>
> •       general tools like broad statistics
> •       specific to users like what they mapped recently
> •       specific to an area you are looking at like if there are map notes
> near where you are looking.
>
> The near-zero cost and maintenance way to integrate these is simply
> provide links out to them in the right contexts, I’m sensitive to not
> giving anyone any additional work, especially maintaining anything. But the
> feedback is - what things could be integrated, for as low cost as possible,
> so they aren’t off to the side that most people don’t know about?
>
> * Broadening the conversation:
>
> I had feedback to talk to more communities and I started on twitter[1],
> happy to do more.
>
> …
>
> I’m sure I’ve missed something, if so please let me know.
>
> Best
>
> Steve
>
> [1] - https://twitter.com/SteveC/status/1658208914360664065
>
> _______________________________________________
> osmf-talk mailing list
> osmf-talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/osmf-talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20230517/d4d7e66e/attachment.htm>