[Imports] Value of imports in general (was: Import of addresses in Poland)

Frederik Ramm frederik at remote.org
Wed Feb 5 19:47:52 UTC 2014


   I'm replying to you here and to those who without further comment
"+1"ed your message. I know that your background is the French cadastre
import but I'm trying to remain generic.

> The question is more to know if address data is something valuable for
> OSM  or not.

OSM is about community first, and about data second. (This is trivial
and not up for discussion - if we lose the data but still have people we
can re-create the data, but if we lose the people then we're done. If
anyone is of the opinion that data is more important than people, don't
read on - we live in different universes.)

Whether or not importing a certain data set is a net gain for OSM must
be judged in the light of the question: Will this increase the number of
people who participate in OSM, or at least make life better for those
who already do?

That's why in this discussion I asked about local communities working
with the imported data.

If the answer to my question above is "no but the map will be more
useful", then that is not sufficient reason for an import - I can then
simply mix-in the external data into Nominatim or a tile serving database.

> Of
> course, the community already answered to that question : it's much
> easier to do this merge and consistency check only once for all
> consumers in the same database. 

Very often, we don't see "the community" importing, but a very small
number of indiviudals, often without even talking to the community
before they dump their supposedly helpful data ("sure the map is better
with my building footprints than without?????!!!!") onto OSM and leave
us to sort things out.

These people don't mean evil - they just don't know enough about OSM to
see the error in their ways.

That's why in this discussion I asked whether this import had been
discussed in the community, or whether it was just one guy trying to be

> If someone find a mistake or something
> outdated, it is fixed once and for all consumers at the same time.

Sadly, no. We have many, many imports in OSM that just rot along, and
where precisely nothing is fixed, for years. One quick look and you find
lots of places where anyone can see that something is wrong, but it is
*not* fixed. This is particularly bad in the USA where in some places,
three different layers of non-matching imports are piled atop each
other, with nobody giving a damn about the obvious problems - if you, as
a newcomer, were to look at that map, you'd probably say: "Uh, this
doesn't look like anybody in OSM is interested in a good map. Cleaning
this up is too much work for one person. I'll come back when it is fixed."

We have imports where stacks of house numbers are at the same location,
making editing *more* *difficult* for people. We have imports where
what, in reality, is a nicely lined-up row of residential buildings, is
a jumble of rectangles at varying angles - often more work to get them
right than to trace over imagery in the first place. Imports like these
make OSM worse, not better.

That's why we have to discuss each import individually and try to
ascertain that it will make OSM better, not worse.

Imports can be valuable when they fall on fertile ground, when there's a
community using the imported data to improve the map. When there's
nobody to check whether the imported data is any good at all, and nobody
who cares - "this is government data, how wrong can it be, and anyway,
if it's wrong I'm sure someone will come along and fix it!!!" - then
imports are a liabilty.

> The question about keeping the data up-to-date is a general challenge
> in OSM, not only for data imported.

True but imports often drastically worsen the ratio of contributors to
data. If two mappers in a city cannot keep 1.000 objects current, that
is not a license to dump 100.000 more objects on them ("they couldn't
keep it up to date before either").

> When a country or a
> municipality is publishing and opening its address database, it's also
> not falling from the sky. This data is also the result of many
> contributors, skilled workers, collecting and keeping the data
> up-to-date (at least, trying to). 

This is not universally true, and has to be assessed in every individual

> If addresses are released in open licences by public authorities, it
> would be foolish to not take this opportunity to considerably speed-up
> the task.

I repeat - if there is no community in OSM that actually cares, then it
would be foolish to dump that data into OSM - it can easily be mixed in
if desired.

If there are local mappers doing addresses, and asking for help by
importing some data to speed up their task, then that might be another
matter, because then there's hope that we have a chance at
cross-checking the data and fixing problems by people who live there.

"By people who live there" is one of the key strenghts of OpenStreetMap,
and I don't want to see that squandered by people importing data in
areas they have never even been to!


Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"

More information about the Imports mailing list