[Tagging] Structured vs. duck tagging tendency

Tom Pfeifer t.pfeifer at computer.org
Tue Jan 5 11:32:04 UTC 2016

Triggered by the discussion about leisure=aquatics vs leisure=sports_centre+subtag,
I would like to discuss the tendency of structured vs. duck tagging.

Thanks to Martin who pointed me to Richard's coining of the latter term:
https://wiki.openstreetmap.org/wiki/Duck_tagging, and I know there is a lot
of controversy particularly around the 'path' issue, which I do not intend to
warm up here.

There is however some observation that a quite a number of mappers prefer, and move to,
more structured tagging, once there has been an appropriate scheme designed, because
in the long run it helps to tag more precisely.

There is a nice analysis about tag migration by Matias Dahl:

When you scroll down, you find the table which ranks the "leave_ratio and transition_count".

It turns out that that transition of amenity=nursing_home to amenity=social_facility leads
the table with 13758 transitions (plus 2417 to community_centre) which is 84%+15%= 99% of
those who migrated away from nursing_home.

Thus with the appearance of the structured scheme for
   amenity=social_facility (41138)
   + social_facility=group_home (21057)
   + social_facility:for=senior (5977)
the former duck-tagged
   amenity=nursing_home (still 11580)
could be described much clearer, and in addition the group homes can be much better
distinguished e.g. from social_facility=assisted_living (9598) which previously had no
proper tag.

My conclusion is that a majority of mappers prefers a better structure and move
away from tag fragmentation when they have a chance. This then helps the data
consumers which do not need to run after each new duck tag to be implemented to
cover a certain class of features.


More information about the Tagging mailing list