[Tagging] Clarification unclassified vs residential

[Fernando Trebien]

On Tue, Feb 26, 2019 at 9:40 AM Paul Allen <pla16021 at gmail.com> wrote:
>
> On Tue, 26 Feb 2019 at 12:17, Fernando Trebien <fernando.trebien at gmail.com> wrote:
>>
>>
>> I don't think a uniform, worldwide highway class standardisation based
>> on road attributes is possible and satisfactory. But I think a
>> functional one would be, at least as a guiding principle.
>
>
> What we currently have doesn't reflect reality too well, even in the UK.  It makes the
> assumption that the width/capacity/speed of a road correlates well with its classification.
> Of course, we have lanes and speed limits to refine matters, but there is still the implicit
> assumption by many mappers that a primary route is "better" than a secondary route.

This leads me to this question: if we can map road attributes using
specific tags (width=*, speed=*, lanes=*, surface=*, even divided=*
which is currently represented by geometry), then why highway=* has to
correlate strictly with them? I think highway=* is intended to
represent something else, not what mappers commonly think, though the
two would be correlated.

> It's sort of true, in the UK, most of the time.  But it is possible for a primary route in the UK
> to have fewer lanes or lower speeds for part of its length than a secondary route between the
> same two locations.  Unlikely, but possible.  Road classifications in the UK are essentially
> hints to the routeing algorithm in drivers' heads.  A primary route from A to B is generally
> preferable to a secondary route because of a combination of factors including speed, width,
> straightness, length, junctions (lights or roundabouts), surface, and signage.  On any single
> one of those metrics the secondary may be better than the primary, but overall the primary
> is preferable.  A secondary route in one locality may be better in all respects than the primary
> in a different locality but that route is a primary because it is the best route (for some values
> of "best") betweentwo important locations.

There's a very interesting similar situation near the place where I
live in Brazil. There are two main routes between a metropolis of 4
million people (Porto Alegre) and the second largest city in the state
(Caxias do Sul), which are 2 hours apart from each other: the shorter
federal highway, and a longer string of three state highways. Both are
paved, but the state highways are divided and higher speed, while the
federal highway contorts through hilly terrain. As a result, the state
highways are the main route between the two, so, there's consensus
that in this particular case the federal highway is not as important
and the state highways should be classified as more important than
that one.

A while ago I crunched some numbers and revealed that paved federal
highways highly correlate with the functional trunk class as published
by some local authorities, though the correlation is not exact. So if
we were classifying based solely on attributes, we would have achieved
a result that is not in line with consensus in this rather significant
case, though it would mostly agree in general.

-- 
Fernando Trebien