[Talk-us] Bike route relation issues

stevea steveaOSM at softworkers.com
Sun Jan 11 19:54:07 UTC 2015

Kerry Irons writes:
>By the logic that I-5 in Oregon is tagged as a bike route, then all 
>roads in the US that don't prohibit bicycles should be tagged 
>likewise.  Obviously that "logic" is incorrect.  There is no body, 
>official or otherwise, that calls I-5 in Oregon a bike route.

Agreed:  see below about the map referenced by Clifford Snow which 
only notes that I-5 "is" an Interstate highway.  No suitability or 
legality for bicycles is expressed (though it may be implied) by 
Oregon's DOT map.

The legend on Oregon's State Bicycle Map, shows Interstate Freeways 
simply designated as such (and "diminished" by map color semiotics -- 
making them gray), no suitability or legality of Interstates for 
bicycles is expressed, though it may be implied by being a "lesser 
semiotic."  (As in, "poor choice upon which to bicycle.")  The map 
legend also denotes "Highway Shoulder Width 4' or More" (prominent: 
thick with red casing), "Highway Shoulder Width Less then 4'" (yellow 
and thinner) and "Paved/Gravel Road Without Shoulder Data" (thinner, 
less prominent lines yellow with gray casing or gray and very thin). 
Importantly, no specific mention is made about the legality of 
bicyclists on any particular road.  So I come to a conclusion that 
Oregon's DOT makes no assertion of bicycle legality on any road, AND 
does not express any particular bicycle routes, at least with this 
particular map.

Let us recall that it is longstanding correct data entry in OSM to 
enter physical infrastructure tags for bicycles (such as 
"cycleway=lane") as well as logical infrastructure tags for bicycles 
(route relation data such as network=rcn).  Both might be determined 
from either "on the ground" real world data such as paint on the 
asphalt (physical) / a "Local Bike Route Number 44" sign (logical) OR 
from published/printed (by a government official body) data such as a 
map of a local or state bicycle route network.  However, in the 
latter case of describing logical infrastructure, actual signs make 
route data unambiguous to put into OSM, whereas a published map 
without signs is a bit more controversial.  I argue that a government 
body which says "a logical bike route exists on these segments of 
physical infrastructure" (but without signs) means that OSM can 
correctly contain a bicycle route relation reflecting this.  This is 
the "on the ground verifiability" issue regarding signed vs. unsigned 
(logical) bicycle routes.  We should not confuse this with using 
proper tags (cycleway=lane...) to describe physical bicycle 
infrastructure, or whether bicycling is legal on a particular segment 
of physical infrastructure:  these are different but related issues.

James Umbanhowar writes:
>The GDMBR issue seems to be a conflict between tagging for the 
>renderer and tagging for the router...My opinion is that the road 
>ways themselves should be tagged as unpaved (or tracks as many 
>already are).

Agreed, though this does not seem a conflict between tagging for the 
renderer and tagging for the router:  tags highway=track and 
surface=gravel suffice to describe physical infrastructure, route=mtb 
and ref=GDB suffice to describe logical infrastructure.  These 
accurately and sufficiently tag, and renderers get them right (well, 
they do or should).  Additional tags (width=...) might not render, 
but if accurate, can be helpful.

>The I-5 thing seems strange.  That is not a separate "bike route" 
>but rather an interstate highway that allows bicycles.  bicycle=yes 
>on all the component ways should be sufficient.

I do not agree:  again, I find no evidence (from the Oregon DOT map) 
that bicycles are explicitly designated "legal" on I-5.  It may be 
the case that explicit statute specifies bicycles are allowed on I-5 
in Oregon, but this map does not explicitly do so.  Again, please 
note that no specific "bike routes" are designated on that map, 
either.  It simply displays some highways as Interstates and some 
highways as containing wide shoulders or narrow shoulders.  While not 
complaining about Oregon's DOT helping bicyclists better understand 
where they might or might not ride a bicycle in that state, I 
characterize these map data as "early" or "underdeveloped" w.r.t. 
helpful "bicycle routing" by a DOT.

And Richard Fairhurst asks:
>  > What does the community think?

There are many issues here.  One (e.g. in Oregon re: I-5) is whether 
"any" road which is legal for bicyclists should be 1) tagged with 
"bicycle=yes" and 2) be part of a bicycle route relation.  From our 
United_States/Bicycle_Networks wiki, if a road or cycleway is tagged 
with a (local) "Bike Route" sign, "without labeling or numbering of 
routes, ways marked as bike routes should be tagged lcn=yes, either 
directly or as members of a route relation."  This makes sense, but 
it is not 1) above, it is more like 2).  If a government body has 
posted Bike Route signs, it is clear we want lcn=yes.  If a 
government body has published a map explicitly denoting a bicycle 
route (whether numbered, named or simply identified), I believe we 
can reasonably enter that into OSM, even without signs.  (I realize 
this is controversial w.r.t. OSM's "on the ground verifiability" 
maxim).  But if a government body has simply made bicycling legal on 
a particular CLASS of roadways, I don't believe we need explicitly 
tag it that way, lest we clutter the map, as bicycling is 
specifically allowed on many (perhaps even most) roads in the USA. 
Exceptions reasonably arise, for example, on sections of freeway 
(motorway) where bicycles ARE allowed, where it seems to make sense 
to tag bicycle=yes as it may otherwise be reasonably implied that 
freeway (motorway) is bicycle=no by default.

James Umbanhower's point about bicycle=yes on I-5 in Oregon (if 
indeed they are actually allowed/legal) seems correct to me as well. 
At least in the 50 states in the USA, it makes sense for OSM to start 
with the assumptions that bicycles are legal on all roads and that 
bicycles are not legal on freeways (motorways).  We should then tag 
exceptions where there are exceptions.  If this is I-5 in Oregon, 
then let us tag bicycle=yes on those segments.  AND (also), if there 
is an explicit (non-private/public) ROUTE, then put those segments 
into a route relation.  But I see no evidence to do this in the 
instant case.  (We can certainly be persuaded otherwise with 
convincing data).

In short:  bicycle tagging is reasonably separated into physical 
tagging and logical tagging.  The former denotes physical 
infrastructure and/or (sometimes) whether bicycling is allowed/legal 
on a particular physical facility.  The latter denotes whether a 
bicycle route logically exists on a set of physical facilities. 
Sometimes signs exist to denote a logical route (in which case it is 
unambiguously a bicycle route), sometimes no signs exist, but a 
government body publishes a map or description of the route, in which 
case it really does "exist" in real world, but may be difficult or 
impossible to determine by "on the ground" verifiability.  This 
latter case still does stir controversy in OSM, but we do continue to 
discuss this in ways which are productive and helpful for OSM's 
longer term future.


More information about the Talk-us mailing list