[Tagging] compressed_air=yes / no - adding to amenity=fuel

[Matija Nalis]

On Tue, 14 Dec 2021 10:53:56 +0100, Martin Koppenhoefer <dieterdreist at gmail.com> wrote:
>> no, as =separate should be treated as pointless synonym of =yes
>
> if you want to count all x in an area you would omit the x=separate and
> count the x=yes, that's the idea. It will reduce the error margin from
> double counting if we start doing it, at the cost of having to deal with
> separate as a yes value for certain queries.

I agree.

Also, the counts are going to be wrong anyway (and not only because the 
map can never be "completed"). 
For example, I map mostly for users, so I'd try to add "amenity=atm" when I
see it.

If I have a lot of free time waiting there (which only sometimes happens),
I'll also try to map separate operators (for example 3 ATMs for 3 different
operators, with separate "operator=*" tags), as it might be quite relevant
for the users (mostly due to rampaging fees down here when using ATMs of 
different bank then your own).

However, AFAIR never yet had I been so bored / blessed with so much free
time, to map "capacity=4" when there are 4 ATMs of same operator (much less
micromap separate position for each ATM of same operator), as I perceive
value of that information MUCH lower. I guess I'm not the only one behaving
that way. And this does severely affect such attempts at counting.


>> If you count both atm=yes and amenity=atm then you WILL double count.
>
> Not combining both queries, you would miss _a lot_ of ATMs, I do not
> consider this an option. Errors can of course be reduced through filtering

I agree wholeheartedly. The availability (or not) of ATM or compressed air
(where one might expect it, like bank/fuel station) is *much more* useful
information than the mere count of those amenities in some area (especially,
as explained above, that count will be quite wrong anyway - one could try to
help minimize that counting error by using "*=separate" or relations, but it
is IMHO of secondary importance only).

>> Pretending that =yes is correct only when object is not mapped separately
>> is not viable, ignores existing mapping and is not helpful.
>
> it is not about pretending that "yes" is incorrect, it is an alternative
> yes with added semantics.

agreed, "separate" carries additional information so is preferred (if such
information is known), but that does not make "yes" incorrect. I.e. it is
much better to map "atm=yes" or "atm=no" missing such information on bank,
if one does not know if it mapped separately, then to forego adding that
information because some wrong count might be even slightly more wrong.

Same with "compressed_air" or "toilets" on fuel station, etc.

-- 
Opinions above are GNU-copylefted.