[Tagging] Feature Proposal - Voting - Rivers Classification

[Christoph Hormann]

On Tuesday 17 October 2017, Kevin Kenny wrote:
>
> I concede that 'relative road importance' strains that definition.
> But I fail to see where any conceivable renderer would be able to get
> the information if we don't tag it. [...]

The problem you are probably referring to here is that OSM-Carto (and 
other styles) render highway=motorway and highway=trunk from z5 but do 
not render highway=primary until z8 which makes the map look ugly in 
between with road segments ending in the middle of nowhere because the 
classification as highway=motorway/trunk is usually not a measure of 
structural importance but based on local physical characteristics and 
official road classifications.

Concluding from this that you need an additional tag for roads 
indicating a subjective measure of importance to be able to produce a 
good looking road map is wrong though.  This is something you can solve 
pretty well through analysis of the connectivity in the road network.  
And doing so instead of having a static importance tag in the database 
would allow you to adjust the method of analysis and thereby the 
results to the specific application - like 'i want a map showing only 
the frequently used road connections' vs. 'i want a map that also shows 
rarely used connections to the remote parts of the country' where there 
is much less traffic overall'.

Even if you are fine in principle with having tags that are not 
verifiable you should be aware that having an importance tag would 
still imply there is only one correct way to measure importance and all 
maps should base decisions on this measure.  In other words: Tagging a 
subjective importance tag would mean you try to tell others what they 
should consider important.

-- 
Christoph Hormann
http://www.imagico.de/