[OSM-legal-talk] instead of replacing data can I just revert to the last known "clean" version?

[Frederik Ramm]

Maning,

On 12/16/11 08:26, maning sambale wrote:
> As what the subjects says, instead of removing and recreating tainted
> data, I think it's best (in some cases) to revert to the last known
> clean version.

This makes sense.

Sometimes you will not even have to revert to a "last known clean 
version", for example if you have a node where

* Agreeing user A creates the node with amenity=restaurant, name=foo
* Disagreeing user B adds cuisine=italian
* Agreeing user C adds wheelchair=yes

then it is sufficient that you remove "cuisine=italian", you don't need 
to remove wheelchair=yes also.

However, if you do that, then the object still looks "tainted" to anyone 
taking a quick look at the history (user B does not vanish from the 
history). Only closer examination reveals that user B's contribution is 
now void. And it gets more complicated:

* Agreeing user A creates the node with amenity=restaurant
* Disagreeing user B adds name=Fred's Pizza Place
* Agreeing user C changes name=Tom's Pasta Emporium

this node is clearly "clean" already, because it does not contain traces 
of B's work any longer. However a quite similar example...

* Agreeing user A creates the node with amenity=restaurant
* Disagreeing user B adds name=Freds Pizza Place
* Agreeing user C changes name=Fred's Pizza Place

,,, suddenly isn't that clear-cut anymore. Has user C really surveyed 
the place, or has he maybe just run a bot that used complex rules to 
"fix" names?

I think it would be good to have a tag that mappers can use to say "this 
object is clean, I have personally checked the history and/or reverted 
it to a relicensable state, any contributions by non-agreeing users are 
not present in the current version any longer".

Then, if you revert an object to an earlier version, you'd just add that 
tag to express then even though the object history does contain 
contributions by non-agreers, it can remain.

I am experimenting with using the tag "odbl=clean" for this, and will 
build support for that into the OSMI relicensing view. But the matter 
still needs to be discussed properly, and with OSM Inspector not being 
an "official" site in any way, it is not for me to say whether such a 
tag would be honoured when the day comes.

Bye
Frederik