It's not a coincidence that there are few 'failures' in Belgium. Over the
years I added 70k of them and cleaned up the existing ones as I got along,
maybe 1k. As I chugged along, I didn't read the wiki every time it changed
and when I went to have a look in neighbouring countries, I noticed the
mapping was done differently there. Of course, now I am the one who is
doing things differently, but the way I see it, I'm doing it in as simple a
way as possible, while still complying with what matters of PT v2. Anyway,
I didn't feel like adding 140k objects, so there will be relatively few
stop_position nodes compared to the number of platform nodes. If you look
more closely at the data, you will find that platform ways are actually
mapped, where they exist. It's also not that I'm not mapping any
stop_position nodes. Where lines start/end and where a node is needed to
split the way anyway, I am mapping them consistently.


2018-04-16 19:50 GMT+02:00 Roland Olbricht <>:

> Dear all,
> there is another proposal, distinct from Polyglot's proposal that is in
> RFC since some days:
> stop_positions_and_platforms
> Before I tell a personal opinion, I would like to present some related
> facts. The basic idea is to look how the tags are actually used.
> Taginfo tells us
> that there are more platform tags than stop_position tags
> and both are so many that you cannot reasonably view them all on a map.
> About two thirds of elements with public_transport=platform
> bear also highway=bus_stop
> form#combinations
> but there are at least more than 100'000 nodes (not ways, not relations,
> hence not fully mapped platforms) that are public_transport=platform but
> not highway=bus_stop.
> These are still too many to reasonably show them on a map (and to not talk
> what should happen about them). For you convenience, I have added a bbox
> limited request for those that you need to move to your area of interest:
> From my fist impression, most of them are simply not tagged with
> highway=bus_stop. However, the first hit in Rome is a tram stop that for a
> good reason is not tagged as highway=bus_stop.
> It is much less compelling to retag public_transport=stop_position as
> highway=bus_stop. Less than one third of them is tagged as
> _position#combinations
> And here, many belong to railways. Again, a bbox bounded query:
> Another matter are definitely wrong public_transport=stop_position,
> because they are off any highway, railway, or ferry. There is a surprsingly
> large number of them. I have made an area related request because we need
> the statistics in a moment:
> Running this request for Germany, France, Belgium, the Netherlands,
> Austria, and Switzerland shows that more than half of all global uses of
> stop_position are in these six countries (go to "Data", read the first
> entry). This is not an arbitrary selection - all these six countries have
> about 1000 to 2000 stop_positions per million inhabitants, but for
> comparision UK has only 150 stop_positions per inhabitant.
> However, the number of failures are significant, between 5% and 10% (with
> the exception of Belgium). The uneven distribution suggest that these
> relates to specific users, but the total number of users is more than 500,
> suggesting that there is a problem to explain the difference between
> stop_position and platform.
> All in all, I do agree that there is definitely room for the improvement
> of public transport mapping. But making the tagging of 500'000 nodes
> deprecated and neglecting both problems with that and the more presseing
> problems is probably not a good way to go forward. Given that Ilya refuses
> for unknown reasons to read this mailing list, please discuss the matter on
> the discussion page if he wiki page.
> Best regards,
> Roland
> _______________________________________________
> Talk-transit mailing list
Talk-transit mailing list

Reply via email to