[OSM-talk] [Talk-GB] "Unsurfaced road" and "Byway"?

[Ulf Lamping]

Sorry, if my response may sound ironic, cynic or even aggressive, but 
your post made me really upset - because I think it's just wrong!

Andy Allan schrieb:
> I must have missed the point where the discussion / voting system
> became the *only* way for a tag to appear on the Map Features page. I
> think widespread use, flawed or not, historical or not, voted on or
> not, is also sufficient.
>
> In fact, if I were to make any comment on the voting system, I would
> say that nothing should be proposed for voting *until* there is a
> sufficient use of the tag. I see a awful lot of tags being proposed
> through someone trying to scratch an itch - which is why you're
> allowed to use whatever tag you feel like, without having to get
> things voted on Map Features first. 
So you'll end up with (yes, lot's of these variants were discussed on 
the proposal - and people are very "inventive" to find other variations):
shop=doityourself, shop=diy, shop=DIY, shop=do_it_yourself, 
shop=do-it-yourself, amenity=doityourself, amenity=diy, shop=hardware & 
type=diy, shop=Baumarkt, ...

Do you really think that it's better than a single proposed: 
shop=doityourself "right from the start"?!?
> And in another thread there was
> someone having problems with the length of Map Features page on the
> wiki - an indication, perhaps, that including so many sparsely used
> tags (whilst ignoring those in common use because of "procedural
> irregularities") is making the signal to noise ratio on that page
> worse and worse.
>   
... and the exact opposite to the millions of questions on the lists 
asking about how to tag this and that.
> So having nothing other than a cursory flick through my mailing list
> archives for some approved tags and comparing them with tags that I
> know haven't gone through the voting process:
>
> shop = outdoor ('approved') - 10
> crossing = toucan (not approved) - 49
> railway = subway_entrance ('approved') - 224
> shop = doityourself ('approved') - 22
> ncn_milepost = * (not approved) - 113
> amenity = car sharing (approved) - 23
> rcn = * or rcn_ref = * or route = rcn (none approved) - 2109
>   
Well, first of all, the rcn are three tags and not only one. You might 
know that these are the same, a newbie (or me) or a computer program 
will - unnecessarily - have a much harder life with it.

Interestingly, you took approved tags that were approved less than a 
month or so - do you think a freshly approved tag will appear a thousand 
times a day after it was appoved?
Interestingly, you took "not approved" tags that I don't really 
understand. These tags might make perfect sense to you - for me they 
have no real meaning (I'm not in the UK). What is crossing=toucan, 
ncn_milepost, rcn, ... - a voting process might have found a more 
international recognizable tag than what you are now referring to.
> and, of course, the original discussion was about:
>
> highway = unsurfaced (not approved) - 1938
>   
What's unsurfaced? Is this a magical highway that has no surface? Ok, 
more seriously: Am I allowed / is it possible to ride a bike on it? A car?

In addition we also have tracktype=grade1-5, we have surface=unpaved - 
did I mentioned that all this was *not* approved - and still confuses me?
> And more importantly, I know that every one of the not-approved tags
> is not only in use in the database, but even being rendered too (no
> prizes for guessing where!)
>   
So all is well now? "I got it rendered and I don't mind if anyone else 
has a problem with it - or don't understand it" - you mean this kind of 
solution?
> So lets all be a little less bureaucratic when trying to exclude
> things from "Map Features", or alternatively return the page to its
> original use (a useful guide to what tags we use) and take the
> bureaucracy elsewhere.
>   
This sounds pretty well, until you think about actually gonna *use* the 
data. It's probably no fun to write a renderer for the 20000 possible 
variations of 200 tag's (not to mention that it's ugly, hard to 
understand, error prone, and slower to work with) - as a result for not 
doing some proposal work "at the beginning".


There are lot's of tags that seems to be a "no brainer" at first sight, 
but are still not well thought out if you look deeper at it. In my 
experience only 10-20% of the proposals are really "no brainers", often 
someone comes up with a good point that should be taken into account and 
in fact makes the proposal much better - right from the start.

And IMHO it's a much better way to think about possible problems first, 
solve them (where possible) and then use a tag - in comparison to first 
get a plethora of similiar tags and try to sort it out later. People get 
pretty unflexible once they have used a tag for a while so fixing this 
pletora later is difficult or even impossible in OSM ...


Regards, ULFL

P.S: Yes, the proposal process can be annoying sometimes. But I think it 
actually speeds up OSM things, as a lot of the disputes can be quickly 
resolved in the proposal stage.