[Tagging] Access restrictions and expressway=yes

[Kevin Kenny]

On Mon, Feb 22, 2021 at 9:23 AM Fernando Trebien <fernando.trebien at gmail.com>
wrote:

> So, my point is that using only the physical characteristics for
> highway classification in OSM ends up indirectly encoding into the
> road network the regional and local wealth gaps, not reflecting their
> socioeconomic importance, [1] which is the concept adopted by HOT. [2]
>

Here's another case of something that I've started to believe is an
extremely general problem: hierarchical classification.

There's one classification that's road 'quality' in some sense:  how
capable a vehicle is needed to drive on this road? At what speed is
interurban travel likely to take place? How 'good' is it from a routing
perspective? That's what gives rise to contentions like "non-hard-surfaced
roads are tracks",  "an expressway must have fully controlled access", and
so on. It can also be taken to a ridiculous extent - downgrading roads that
are major routes to insignificance because they lack some surface
characteristic that the mapper thinks (or some Wiki page edited by
$LC_DEITY-knows-whom states) that those characteristics are typical of the
road type in question. This is nevertheless the approach that tends to be
preferred by data consumers such as routers, who want to assign a routing
penalty to 'worse' roads.

There's another classification that's road 'importance': to what extent is
a road a lifeline for the local economy? Is it the main route between major
cities, the only route that serves an isolated village, or an insignificant
road that gets tractors and logging trucks in and out of an agricultural or
forestry area? The 'economic and social' importance metric is what's
favored by HOT.  That's also been taken to a ridiculous extent. I've seen
mappers argue the idea that a major surface street maintained to a standard
that would be 'trunk' or 'expressway' elsewhere can be at most 'secondary'
if it parallels a freeway. (In my area, there are some of those, maintained
to the high standard because they are reliever roads for the freeway in the
event of trouble and because the freeway exits are many miles apart, so
there is often heavy surface traffic paralleling the main route.)  If this
particular stretch of the concept is applied too vehemently, it also winds
up confusing routers.

In sensibly administered areas, economic importance and road quality are
likely to correlate strongly, so the two camps are likely to coexist
peacefully. Nevertheless, regionally they are affected by the overall road
quality of the region.  Near me, in town a 'highway=tertiary' is likely to
be either two lanes of macadam with wide parking aprons, or even have four
traffic lanes, with traffic lights, turning lanes, and similar features. In
the mountains, a tertiary road may be surface=compacted and two narrow
traffic lanes, because some of the mountain terrain is seasonally subject
to mudslides and washouts, and a compacted road can be repaired and
reopened much more quickly and cheaply than a hard-surfaced road can be
rebuilt. If you have to rebuild the road every spring, it doesn't make
sense to pave it!

Of course, the mapping also correlates to the local level of economic
development. A wealthy region may be able to pave even some of its
agricultural, forestry and mining roads. A poorer region may have
questionable soft-surfaced roads serving even as the sole access for
relatively large towns.

Minh hints at yet a third hierarchy: political importance.  HFCS is about
funding, and its allocation is a political process. Ordinarily, the
bureaucrats actually do a reasonable job of managing it, but it accumulates
anomalies. Often, a road will have a higher HFCS classification simply
because of an inversion of the first two metrics - it's economically or
socially important out of proportion to its construction and has therefore
acquired a higher classification in hopes of funding improvements.  At
other times, the classifications are adjusted to accommodate out-and-out
political needs. The road that serves my brother's farm was suddenly
upgraded to a two-lane paved road from a compacted road on which vehicles
could pass each other with difficulty. The reason? One of the landowners
served by the road was elected to local office.  Similarly, there was a
year or two, about forty years ago, when all Interstates in Wyoming had
their HFCS classification downgraded to 'freeway' because Wyoming had
refused to adopt a national speed limit.

If I were starting from a blank canvas, I'd favor a classification based on
a road''s physical characteristics, HOT and HFCS notwithstanding. That's
what's most observable in the field, and it's got considerably less room
for subjectivity than the current definitions (Witness the endless
arguments over 'what is a trunk?' 'what is an expressway?' 'is a rural road
with these characteristics track, service, residential or unclassified?')
This approach comes close to the ideal for those who consume data for
routing and navigation. There are cases of inversion, where for instance
the local authorities have an interest in routing traffic onto a 'worse'
road to meet some other constraint, but on the whole it works for routing.
The chief drawback to the 'physical characteristics' approach is that in
areas with a highway network that comprises 'poorer' roads from a
routing-preference perspective, the renderer gets too sparse. In
particular, at low zoom levels, even the most major local roads between
large communities may disappear, while areas in more highway-intensive
regions will be unreadably cluttered.

But I'm not starting from a blank canvas. Moreover, it appears that for
many mappers and users, the rendered map, not the quality of data for
routing or statistical analysis, is the important aspect. I won't denigrate
this perception by calling it 'tagging for the renderer': having a readable
rendered map is also important - and so there has to be tagging that
supports rendering.

Where do we go from here?  I don't claim to have the answer - except to
counsel everyone (including myself) that your view of the 'top level'
classification of roads - or indeed, any other mapped object - is not the
one true hierarchy. A view that 'land cover is what is important; land use
is secondary', or 'the social and economic importance of a highway are what
must govern; physical characteristics are secondary' - or the inversion of
either view - needs always to be tempered with the realization that others'
needs are not yours. If we think strictly in terms of a hierarchical
taxonomy, we'll always reach these impasses.

-- 
73 de ke9tv/2, Kevin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20210222/272ba521/attachment.htm>