On 24 December 2011 23:03, Frederik Ramm wrote:
> Dermot McNally wrote:
>> Another mapper walks by, notices that the place is a pizzeria and adds
>> back an identical tag. Are we clean or dirty now?
>
> Dirty, because the very same situation could arise with a non-agreeing
> mapper adding "cuisine=pizza", the agreeing mapper "cleaning" the
> object and a third mapper reverting that last action. I have no way of
> telling apart a revert to the non-agreeing mapper's version and a true
> remapping from original sources. I'm open to suggestions but I can't
> see an easy way out.
Yes, I see the problem - I pose the question because it's interesting
given that the desired end-game is a node that is both clean _and_ has
cuisine=pizza. But you go on to cover this case...
> These changes carry with them the slight complication that they make
> tainted-ness dependent on the current version of the way. This means
> that an object that was previously untainted could now become tainted
> again, by exactly the process that you outline above (re-adding of the
> cuisine tag). That would be a very good use case for odbl=clean, or
> maybe we could introduce something that users can place in their
> changeset comment saying "all edits in this changeset are remapping
> from original sources", or we could even say: Whenever the changeset
> has a source tag we consider this to be original sources...
This was the issue I had in mind, and yes, odbl=clean will fix it. For
anybody who hasn't read the LWG minutes, LWG is in favour of
respecting odbl=clean come the switchover phase. We've asked for
community feedback on this (and on the principle that
moving-nodes-cleans-their-position, which we also favour) since we
want the decision to be an accountable one having regard to all valid
legal and ideological points that should be considered.
Now for a horrid twist to the thought experiment - odbl=clean is, as
you have described its use above, a nice solution to the problem of
wanting to cleanly reapply cuisine=pizza without wiping history. But
what if things happen like this?:
1. Agreeing mapper maps the restaurant and names it
2. Non-agreeing mapper adds the cuisine tag
3. Agreeing mapper removes the cuisine tag and sets odbl=clean. He or
she does not have enough information to assert the cuisine tag and
chooses, on balance, to lose the tag for now.
4. Well-meaning new (therefore agreeing) mapper sees the node, notices
the cuisine tag in the history and reapplies it without having
personal knowledge to back this up. odbl=clean is still set.
This is very similar to the case where the cuisine tag is reapplied
without us having odbl=clean set. Certainly, we can point out that we
expect good faith and due diligence from mappers. But if we are
prepared to consider the object clean in this case, why not also in
the case where the cuisine tag has just had a temporary "holiday" from
the object even if odbl=clean has not been set?
Dermot
--
--
Igaühel on siin oma laul
ja ma oma ei leiagi üles
___
legal-talk mailing list
legal-talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/legal-talk
