In cases where contributors reclassify road types (e.g., service, residential, unclassified, living_street, road) primarily to improve routing outcomes, would you consider this equivalent to “tagging for the renderer”?
These are often low-level or ambiguous road types, and the changes may be driven by how routers interpret them rather than by ground truth or local signage.
If this is considered a form of tagging for a specific data consumer (in this case, a router), should the Tagging for the rendered article be expanded to include this practice?
Historically, OSM was oriented toward tile generation and visual rendering, but now that many users rely on OSM as a data source for routing and analysis, the concept of “rendering” seems broader. I’m curious how the community views this shift.
I think your suggestion makes sense if the classification is incorrect – “lying for the renderer”.
As an example, changing a public road from highway=residential to highway=service to discourage people from using it as a shortcut would be “lying for the router/renderer”. Adding an access=destination where there is not a legal restriction would be a similar lie.
yes, and in the same way would be good if changes are correct, bad if someone breaks data to force specific behavior in router and dubious if change is not obviously wrong but motivated mainly by trying to force specific routing
Similarly mistagging flowerbed as landuse=industrial to force purple rendering is bad, while mapping industrial area as landuse=industrial - motivated by getting it rendered - would be fine and commendable.
If someone spots open residential road mistagged as highway=path or highway=construction and edits it to fix routing - great!
if someone retags open residential road as highway=footway to influence router, it is a really bad edit. The same if someone uses bicycle=no on roads where cycling is legal but unusual/dangerous/rare.
(Quoting myself here; your question was one of my first ones when I joined OSM!) The bad thing is actually better termed “mistagging for a renderer (or router)” - the vast majority of OSM edits are for the benefit of more useful renderers and routers and that’s a good thing, so long as the data is accurate!
But we can consider a reasonable abstract router, or consider all router we know.
For example the wiki for highway=tertiary describes how tertiary roads are linking unclassified roads and settlement to major roads. They should form a network. That is why e.g., Osmose has markers for “Broken highway level continuity“:
The classification of a highway should normally be consistent along the entire path. For example a highway=tertiary should remain highway=tertiary until it intersects with a road of higher classification.
That can be seen as a way of helping routers.
Or we can add tags, that does not harm.
maxspeed:practical is disputed according to the wiki. But I have used it on a few occasions. E.g., when I have been driving on a road that is smooth asphalt, not too narrow, outside urban areas and has no signs limiting speed. But it still slow because there are a lot of intersections and driveways with very bad visibility because of vegetation. Even if I accurately mapped every tree and hedge, routers would not be able to determine how fast I could go. So I set maxspeed=40 so routers will not use the road as a shortcut. And if routers do not like that, they can just ignore maxspeed:practical.
I think it is fine as long as it is a correct data and not narrowly tailored to specific one.
matkoniecz_bicycle_router:dziamdziarandzia=16b would definitely not be a valid tag
but for example using maxspeed:conditional in syntax that happens to be currently supported only by a single router is fine (as long as tag is not made worse and less readable to get it supported)
