[OSM-dev] The wiki defines the database (was: relations)
frederik at remote.org
Wed Nov 5 02:36:23 GMT 2008
Erik Johansson wrote:
> Yes that is very cumbersome but how often does this happen, and does
> it really warrant that flippant attitude? Having a better way to
> handle multiple meanings of tags might help.
The core of this "flippant attitude" is easily explained.
When OSM was started - that was before my time, so I'm just telling
other people's stories here - it was not the only collaborative mapping
Other, "competing" projects started out by first trying to set up a good
tagging scheme (an "ontology" as people say) for everything, and never
got far beyond that.
OpenStreetMap didn't bother, and just started mapping - differentiating,
initially, only between railway, waterway and highway and that was it.
Things evolved from there to where we are now; OSM has swept away
anything remotely comparable.
Like many computer people, my instinct is to do exactly what the failed
projects have done; it is what you are taught at uni or in the
workplace: Analyse problem, make data model, acquire data, process data.
OpenStreetMap managed to largely skip the initial phases, going against
perceived wisdom, and it worked out well.
Now, with the ever larger influx of new people to the project, this
"perceived wisdom", this "how things are usually done", comes in through
the back door. There's not a single day where you don't hear somebody
say "but we need a unified tagging scheme", "everybody needs to adhere
to the same standard", "it will never work otherwise", "the data will be
useless unless everybody means the same". (But "it will never work" is
something that has been said about OSM from day one.)
Things that are special about OSM, things that have been OSM's strengths
in the past, are often unreflectedly discounted as weaknesses by these
newcomers: "Any database must ... blah blah blah ... lest it is
There are two possibilities:
1. OpenStreetMap did the right thing initially, but what was the right
thing *then* is not the right thing *now* anymore; we really need strict
standards, a body to govern them, a dictionary of approved tags, and
editors that will only allow you to tag things differently if you press
"I am sure" three times. That is, as far as I can see, the model that
Google's Map Maker uses.
2. OpenStreetMap is really different from anything else, the usual rules
do not apply, and trying to apply perceived wisdom to OSM will break
what is precious about it. The people calling for standards, rules,
unified tagging and all that are just not flexible enough; they think
they know what works and what doesn't, and fail to see that OSM is a
different environment to which they cannot simply transport their
experiences from the workplace or from software projects or from Wikipedia.
I tend to assume that 2. is correct and I also tend to make fun of those
who, I like to think, cannot adapt their brains to something that works
differently. But it is very well possible that I am wrong, or that at
least situation 1. will be true at some time in the near future.
Frederik Ramm ## eMail frederik at remote.org ## N49°00'09" E008°23'33"
More information about the dev