[Imports] San Francisco building heights import

Christoph Hormann chris_hormann at gmx.de
Fri Dec 2 17:27:56 UTC 2016

I am sorry for not having the time to follow up on this in more depth.  
The remarks in my reply from Nov 17 mostly still stand.  I am not deep 
enough into the matter to reliably asses if the concerns raised there 
are warranted but i also have not seen any replies convincing me they 
are not.  I admit though this goes quite deeply into the technical 
foundations of LIDAR data processing and due to the lack of more 
detailed information on the process issues raised were fairly vague.

This touches the subject of how much you can, when planning and 
organizing an import, trust assertations of external experts (usually 
data providers with their own interests not necessarily in sync with 
the OSM community) regarding properties of their data and suitability 
of the processes used.  In normal mapping we do this to some extent 
(think of GPS device producers, aerial imagery providers) but these 
cases are either technically simple or easily accessible for in depth 
reliability checks and cross comparison with other sources even for a 
layman.  This is different when you have a complex and largely 
intransparent and not publicly documented process, no access to the raw 
data (raw LIDAR by the way means point cloud data, not some kind of 
gridded product) and at best an overall statistical error assertation.  
You can do spot tests on the final data all you want - it is still 
something fundamentally different from normal mapping in OSM, which is 
done with full understanding and control over the whole process from 
the primary data.

Ultimately i think OSM can only work if everyone entering data - either 
as a normal mapper or through imports (and in that case both the ones 
preparing and the ones actually entering the data) is accountable for 
the data he/she imports.  And being accountable has two components: (1) 
the acceptance of responsibility for what you do and (2) the ability to 
substantially carry this responsibility by having a sufficiently 
well-founded ability to assess the quality of the data you enter with 
respect to the real world situation.

The main aim when planning and performing an import must not be to add 
additional data but it needs to be to allow you and your fellow 
importers to responsibly and productively participate in the community 
process of open map production.  If i read people justifying imports in 
general independent of how they were made because they add useful data 
this appears quite misguided to me.  I could easily produce many 
gigabytes of useful information i could enter into OSM but i won't do 
that if i am not convinced this is valuable for OSM as a community 
*independent of the value of the information itself*.

All of this does not mean you should not import building heights for SF.  
But discussion needs to stay clear of the fruitless level of overall 
quality assertations, the pointless concentration on the value and 
usefulness of the data and the sweeping dismissal of concerns as 
unfounded.  I have seen a lot of positive things in the planning of 
this import but when i read some of the replies to Frederik i am once 
again flabbergasted by how people can dismiss concerns just because 
they apparently don't understand them.  It is completely fine to 
articulate diverging experiences and opinions but in return you also 
need to respect the views of others and that those might be valueable - 
even if you do not comprehend them.

Christoph Hormann

More information about the Imports mailing list