[Imports] Uganda Bureau of Statistics educational facilities import

[Christoph Hormann]

On Wednesday 30 December 2015, Rafael Avila Coya wrote:
>
> It is a really big number (24,587 nodes to be exact). I understand
> your worries. The biggest I remember done in a similar way (manually
> with the Tasking Manager) was 14,020 place nodes for Liberia [1], and
> I remember you had your concerns about having enough volunteers to
> accomplish the job [2]. The job was done in about 1 month by (mainly)
> only three volunteers (you can check the Stats for that job). One of
> those volunteers was me.

And that was pretty close to the estimate Frederik gave during 
discussion (i.e. 30 person hours) although no review has been done to 
date if the import has been performed with the diligence assumed to be 
necessary.

IMO for imports of this size where the import encompasses more than just 
a handful of changesets an afterwards review should be mandatory to 
verify if the import actually went according to plans.  This should IMO 
include at least:

* what data was ultimately imported.
* which mappers imported what data.
* what manual changes were performed in comparison to the input data.

> I don't see any restriction on mapping remotely, even in areas where
> one haven't been before. Another matter are the opinions about that,
> and there are for all tastes. But I can't see anything in that
> respect that goes against the import guidelines.

I said this several times already - planning imports is not about 
following the letter of the rules, import discussion has been made a 
requirements for the import process because the community realizes that 
you can't fully cover all problems that occur with imports in a fixed 
set of rules.  As far as discussion here is concerned the guidelines 
only set a baseline, they do not relieve you from the responsibility to 
abide by the community wishes brought up during discussion

I have not reviewed the import plans in detail yet but at the first 
glance it seems clear that:

* There is no data source that allows verification of the data, even 
high resolution imagery will usually not allow you to identify schools 
or even characterize the specific type of school.  This makes the fixme 
tags rather pointless.
* There does not seem to be any validation and assessment being done on 
the quality of the data.  IMO this is not acceptable for an import in 
OSM.  OSM is about mapping reality and we have no way to verify if this 
data represents reality except the ensurance of the Ugandan 
authorities.

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