[Tagging] Data redundancy with "ref" tag on ways vs relations

[Georg Feddern]

Am 31.07.2012 10:33, schrieb "Petr Morávek [Xificurk]":
> Kytömaa Lauri wrote:
>>   Petr Morávek [Xificurk] wrote:
>>> 2) A relation exists with member ways without ref tag. This means that
>>> the route is essentially mapped and any further editor is correcting
>>> errors, that he found. Then someone comes and adds a ref tag to one of
>>> the ways - why?
>> He drove by, and saw a different ref sign next
>> to that road from one intersection and the next
>> one. He has no knowledge of where the ref on
>> the relation comes from, and if it is still valid on
>> all the ways currently part of the relation. He
>> only knows that this way has a ref=new value,
>> and it's all he can enter. Locals will eventually
>> notice, and resurvey the whole area - they'd
>> have to resurvey anyway.
> If he knows for sure, that on that road from point A to point B is
> ref=42 and not ref=56 as the OSM data says, then the user should fix it
> as I wrote in previous email. Remove the ways from the current relation
> and add the correct ref tag to the ways themselves, or create a new
> relation for them. Problem fixed! Note, that if the edit was mistake
> after all, then QA tool analyzing road network should catch that.
>
> If he's not really sure about his data, but wants to alert locals "hey,
> here may be something wrong", then he should use FIXME tag as for any
> other dispute.

Well, I am quite happy with relations and I would like to use them 
"only" as you want, too.

On the other hand I have the practical expirience, that OSM lives from 
"new" contributors
- and "new" contributors have a quite simple approach:
    - E. g. they add nodes, where already buildings exist, because there 
is no icon in their editor, they expect there.
    - They simply do not see relations in their first periods of mapping.

I am quite sure, they would add those ref tags as described above, 
simply because it is their first view.
(I have no practical expirience of that right now - but simply because 
the refs are still on the ways yet here in Germany.)

As a QA contributor you just have two opportunities:
- Remove all ref tags from the ways and organize in relations - again 
and again.
- Tag all ways with the ref tags appropriate. (Double the data with 
redundancy).

You will not get a consistant and straight solution anyway, I am quite sure.

> If I understood your scenario correctly, it can be summarized as: "Let's
> use conflicting ref tags for road disputes instead of fixme tag."
> Personally, I don't support this view.

I would understand as:
Live with the (mapping) reality and use the (QA) possibilities it opens up.

Georg