[Talk-us] MassGIS building conversion

[Frederik Ramm]

Jeff,

On 12/11/12 08:48, Jeff Meyer wrote:
> Frederik - one question I have about changeset tags and things like
> correspondence tables - how easy will it be for people to discover that
> this data even exists? How will they know to look for it? Will they need
> to be software developers? For example, will a GIS expert from the state
> of Massachusetts who comes across OSM know to look for this information?

Changeset tags are like the small print in a contract, or the tiny 
instruction leaflet you find in a box of pills - they are there for 
someone who looks for them, without getting in the way of someone who 
isn't interested.

If the GIS expert from the state of Massachusetts just looks at the map 
on www.openstreetmap.org then he will see no source tags, no object 
history, no changeset tags, and surely no correspondence tables, and 
that is good. If he chooses to dig deeper then he can see more:

* one click activates the data layer
* second click selects object and shows immediate object tags
* third click shows object history and therefore a potential "created by 
soandso_import_account"
* fourth click ("Details") shows relvant changeset IDs and changeset 
comments
* fifth click (on changeset ID) shows additional changeset tags

Ideally the additional changeset tags, if not already the changeset 
comment, would inform about a web site where you can find details about 
the import - what data was taken from where, how it was converted, who 
the people were that did it and how to contact them, plus any log 
files/correspondence tables generated.

I think this is near perfect; one could argue about the exact UI flow, 
maybe eliminate the fourth click and display more in the data browser 
directly or throw people into iD once that's ready. But something like a 
correspondence table isn't useful to you if you aren't a programmer or 
at least a technical person, therefore it shouldn't be shoved down the 
throat of anyone who dares to look.

In case you are wondering whether you should go for source tags on 
objects, or source tags on changesets - something that I haven't talked 
about yet - my advice would be to go for source tags on changesets.

A source tag on an object is like an adjective; a source tag on a 
changeset is like an adverb. The adjective describes the object; the 
adverb describes the process. The object - e.g. a closed way 
representing a house - should only have tags that apply to the 
real-world house. How many floors, what kind of roof, when built, what 
housenumber and so on. The "source" tag, on the other hand, logically 
applies to the process of importing and not to the house. It must go on 
the changeset.

The actual data source for an object in OSM can then be found by looking 
at the source tags of all changesets that affect the object - it might 
be imported from a MassGIS dataset but it might later be modified 
according to Bing imagery, and if you place a source=MassGIS on the 
house then whoever modifies it according to Bing is extremely unlikely 
to remove or amend the source tag (massive proof of this can be found in 
the OSM database). In fact, if I were a MassGIS official I would be very 
unhappy to find objects that still carry a "source=MassGIS" tag even if 
they have been modified later, possibly to a lesser standard than my 
department would apply. (If I were a European GIS official I would be 
very likely to prohbit the use of my data for OSM, for fear of being 
made responsible for whatever quality-reducing modifications OSMers 
might make to "my" data. Sad but true!)

Bye
Frederik

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