[Tagging] Fast food vs. restaurant vs. cafe

[Roy Wallace]

On Wed, May 5, 2010 at 6:30 PM, Ulf Lamping <ulf.lamping at googlemail.com> wrote:
>
>> http://img94.imageshack.us/img94/1179/amenity.gif
>
> You are asking for black and white definitions/decisions where there's
> lot's of room for grey.

There's only "room for grey" (w.r.t. the OSM definitions) if we want
there to be.

> What about a place that serves limited breakfast in the morning, would
> classify as a cafe throughout the day, serves full meals only at noon
> and becomes a bar selling cocktails at night?

You can still design a flowchart to cope with this as desired, with a
bit of tweaking (e.g. "Can you expect a waiter to deliver your order
to your table *at lunch and/or dinner time*?").

> What you just can't do is find a precise definition that is valid
> throughout the world and will be doubtless in all possible situations.

I am talking about a 'stipulative' definition
(http://en.wikipedia.org/wiki/Definition). This definition cannot
shown to be "valid" or "invalid" - it is simply either "useful" or
"useless". To me, a useful definition for X is one that allows you to
state whether something IS X OR IS NOT X, verifiably. That is, if I
ask OSM to "find me a cafe", it would at least be nice to know how OSM
defines a "cafe", even if (especially if!) this is different to my own
subjective understanding of what a cafe is (this is incredibly obvious
to me).

I think I do understand your point, though, that you think it better
to keep using these tags in a fuzzy, subjective, variable way
throughout the world. To avoid going around in circles, maybe we can
agree to disagree on that.

> BTW: The flowchart is using highly subjective language
> "heavily-advertised pseudo-food" which is *very* certainly not a good
> way to find a concensus. Why does it try to offence junk food fans? Oh,
> and the definition of "pseudo food" will very certainly differ between
> people from the western world and people in africa starving right now.

Hehe, this was a bit tongue-in-cheek, incorporating Greg Troxel's
thoughts. Like I said, I'm happy to spend more time on the flowchart,
and welcome suggestions to improve it - but only if there is demand
for verifiable definitions. Otherwise, I won't bother.