[Rebuild] Do I win a prize if I am the first to post?

[Frederik Ramm]

Hi,

On 01/16/12 09:46, Simon Poole wrote:
> I would like to take one step back and ask, what is the value of keeping
> the history of edit-tainted objects and is it enough value to justify
> expending a lot of effort to keep it in the DB (in the current DB, it is
> a given that in some form it will stay available for people really
> interested enough to process history files)?

History information is quite valuable to many; think of e.g. Muki's 
various rather scientific approaches at determining "quality" in OSM by 
looking at the intensity of edits.

> We are talking about something around 5% of the current highways and
> routes (the impact on other objects is < 1%) and the number is likely to
> decrease up to the switch over date.

The more edits something has (the more important it is), the likelier a 
contribution by someone who disagrees.

> I believe that we will be far better served by avoiding complicated and
> in the end unsatisfactory history retention schemes and should simply
> synthesize new v1 objects (keeping the current id to make things simple)
> for anything that has tainted edits and leave it at that.

If my reasoning is not totally off, then that would likely destroy the 
history of not just "any" 5% but of the 5% most important objects. And 
object going back 3-4 years and having been touched by 20 different 
mappers can tell a lot about OSM and we should not sacrifice that lightly.

Not saying we shouldn't do it *at all* but I think that if we do cut 
history then I would be in favour of cutting *all* history and setting 
up a separate service where you can inquire about pre-changeover history 
(under CC-BY-SA) for *all* objects, so that it becomes clear that you 
*have* to look there for pre-changeover information, rather than keeping 
most history and throwing away the 5%.

Bye
Frederik