[Tagging] Non-orthogonal crossing=* tag proposals: crossing=marked/unmarked vs crossing:markings=yes/no

Paul Allen pla16021 at gmail.com
Sat May 25 12:20:53 UTC 2019


On Sat, 25 May 2019 at 08:03, Nick Bolten <nbolten at gmail.com> wrote:
> What do you mean by a crossing with traffic signals AND with road
markings?

>
> Status quo, per the wiki: tag with crossing=traffic_signals,
> hiding/erasing any information about markings that would be communicated in
> other values.
>
> Under the new proposals: tag with crossing=marked (or
> crossing:markings=yes if it ends up the proposal) and crossing:signals=yes.
>

Here we seem to be in fundamental disagreement.  A crossing with traffic
signals is a crossing
with traffic signals independent of road markings: the interaction of
pedestrians and traffic is
determined by the status of the lights.  At a crossing with only markings
the interaction of pedestrians
and traffic is determined by the markings.  Traffic regulations for the one
say that the behaviour
of pedestrians and traffic is controlled by the lights; traffic regulations
for the other says that the
behaviour is controlled by the road markings.

Please let's not drift off into semantic arguments about "controlled."
Ultimately it's all about
traffic regulations and ultimately laws do not control behaviour as people
can, and do, break
those laws.  Maybe we should be talking about "passive" and "active"
instead.  Or signalized
and unsignalized.

Yes, it's of interest that in some areas a crossing controlled by lights
also has road markings but,
unless you can show a case that those markings make it a different type of
crossing, they're
cosmetic enhancements.   And yes, I understand that those "cosmetic"
enhancements may be
important to those with visual impairment, and worth mapping for that
reason, but they do not
significantly change what type of crossing it is.  Don't get hung up on
"cosmetic" implying
"trivial" or "insignificant" - I'm not saying that, I'm saying they don't
change the fundamental
type of crossing or the way pedestrians and traffic are supposed to
interact in a given jurisdiction.

I have yet to see anyone present a case where the presence or absence of
road markings at a
crossing controlled by traffic signals requires different behaviour by
either pedestrians or
traffic.  Perhaps such cases exist (Poland is a possibility, awaiting
clarification) but until then
the defining characteristic is that the lights tell pedestrians and
motorists what to do.

Road markings alone is a more difficult.case because different
jurisdictions assign different
behaviours to them.  But an important characteristic is that there are no
traffic lights.  In the
UK, pedestrians have right of way at such crossings.  In other countries
they may serve purely
to warn motorists to be more cautious about pedestrians attempting to cross
there.  It is
debatable whether this should be handled as auxiliary tags
(pedestrian_right_of_way=yes/no)
or (as with many tags) it is something we don't map because we try to avoid
mapping
legislation and instead say "It's a marked (unsignalized) crossing, it's up
to you to figure out what
that means in this particular jurisdiction."

Acid test: explain to a child how to cross the road.  It is going to be
along the lines of "At this type
of crossing you wait for the green signal (or whatever) before you cross."
and "At this type of
crossing there are no lights, you behave in this (country-specific) way."
I have yet to see anyone
say that "At this type of crossing you wait for the green signal UNLESS
there are these road
markings, in which case it's completely different."  If that is the case
somewhere in the world
then we'll have to find a way of mapping it.


> > Have you ever seen a crossing with lights AND zebra stripes?  Which of
> the two takes
> precedence?
>
> Neither. They are separate properties of the crossing and can communicate
> different information. We can describe the number of lanes a street has as
> well as it's speed limit without having to decide which takes precedent,
> let's use that same idea for crossings.
>

Does the presence or absence of those road markings fundamentally change
the interaction between
pedestrian and traffic?  Can we say "This crossing with lights is of type X
because it has road
markings and that crossing with lights is of type Y because it doesn't have
road markings; the
behaviour at X and Y crossings differs in these significant ways"?

> However, if you include the zig-zag lines before and after the crossing
> (...)
>
> Maybe the proposal should be updated to be even clearer: a marked crossing
> is one where the pedestrian crossing space is, specifically, visibly
> outlined with designated markings.
>

Which seems to be precisely the opposite of how most people interpret it.
For me, at least, the
visible outlining is cosmetic because it doesn't alter the rules of
engagement between pedestrian
traffic.  Worth mapping for the benefit of the visually impaired, but not
by redefining current usage.
Current usage has marked crossings meaning "not controlled by lights."  So
then we need
marked_and_not_controlled_by_lights and marked_but_controlled_by_lights.
Which is fine,
as long as you don't redefine current usage, because that would cause major
problems if
it went through (which it almost certainly would not).  If you don't make
the distinction then we
have the situation that the visually impaired in the UK (and other
countries) may be misled into
thinking that the markings mean they have right of way at what are actually
cosmetic markings
at traffic lights.


> > then you have the dangerous situation that the map leads people to think
> that a light-controlled crossing (...) is a marked crossing (like a zebra)
> where pedestrians have priority.
>
> With orthogonal crossing tags for markings and pedestrian signals, such a
> crossing could and should be tagged as having signals. The situation
> described appears to be a tagging error: someone said the crossing was
> marked when it wasn't and also neglected to tag the pedestrian signals.
> Situations like this, where signals get neglected, are actually easier
> under the current schema due to markings being mappable from aerial imagery
> while signals usually aren't.
>

Explain how your proposal would significantly reduce errors.  Aerial
mapping a new crossing with
stripes is going to result in a marked crossing either way.  Aerial editing
an existing crossing could
also result in errors either way, but possibly worse with your proposal.
Currently if I edit an existing
crossing because I see stripes in aerial imagery I see from the tag list
that it's already been marked as
having lights and would have to change that to being uncontrolled, so I do
further checks (has
the type of crossing changed, can light-controlled crossings have stripes
in this jurisdiction).
With your proposal I see a "markings" checkbox and tick it without noticing
the "lights" checkbox is
already ticked."



> > But I suspect this is Nick;s interpretation
> of what a marked crossing is - there are some marks on the road (I can't
> make sense of his
> proposals without that interpretation).
>
> My interpretation is the boringest one: a marked crossing is a marked
> crossing. It's described roughly the same way by transit agencies,
> Wikipedia, dictionaries, etc.: crossing with visual markings designating
> the pedestrian space for crossing the street.
>

Wikipedia?  Like https://en.wikipedia.org/wiki/Pedestrian_crossing  It
makes a distinction between
signalized and unsignalized crossings.  Which is what we have in current
usage.  The tag
values are unclear and misleading without referring to the documentation,
which is the case of
many tags, but this is how most people interpret matters:
crossing=traffic_signals means
signalized (irrespective of road markings); crossing=uncontrolled means
unsignalized.  That is the
current usage, and the current implication of the wiki; redefining it would
cause big problems.

Pelican crossings tend to have a dotted line outlining the pedestrian space
> of the crossing...
>

It's still a signalized crossing, not an unsignalized one.  Those markings
are cosmetic.

Outside the UK it's common to find pedestrian-signaled crossings with
> virtually any marking style. Some are shown on the proposal page.
>

Again, those markings are cosmetic.  Important to some people but do not
affect the rules of
interaction of the crossing.

I have absolutely no objection to introducing a way of tagging cosmetic
markings.  I even agree
with you that we should find a way of doing so.  The problem comes with
your apparent
insistence that we redefine the current meaning of a tag.  But maybe I've
misunderstood
what you're proposing.  Redefining a tag a big problem for more than one
reason.

Suppose we wanted to replace landuse=grass with landcover=grass.  Most of
us here actually
do want to do that.  It would require a mass edit, and it's very hard to
get consent for that even
when it's a 1-for-1 substitution with no exceptions.  And we have to get
editor presets changed
(which can be easy or impossible, depending on the views of the maintainers
of the editors).
And then the big problem is that the carto guys have a rule of "no
aliases."  They appear to
be inflexible on that, even in the case of a transition like this.  The old
landuse=grass will
continues to be rendered; the (proposed) new landcover=grass will never be
rendered.

That was the simple case.  What if we have tags P, Q and R, where some P
become X but other
Ps become Y; Q and R become Z in most cases but in some cases become Y.
That requires
a manual edit in each and every case (assuming the proposal goes through,
which is very
unlikely).  On 2.5 million POIs, that's not going to happen.

I'm not saying I'm happy with these obstacles to cleaner, clearer tagging,
because I'm not.
But they're there.  And I also think the distinction between signalized and
unsignalized is of
greater importance than whether or not a signalized crossing has some form
of road markings.

Come up with a way that retains current tag meanings (as commonly
interpreted) and which
allows cosmetic markings at signalized crossings to be mapped without any
confusion between
them and the markings at unsignalized crossings and I'm all for it.

-- 
Paul
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20190525/261f270e/attachment-0001.html>


More information about the Tagging mailing list