[OSM-talk] Cross-renderer tag support, now with OSMdoc!

[John Smith]

2009/12/20 Ulf Lamping <ulf.lamping at googlemail.com>:
> It's up to you to write a tool like this. I would probably love it.

Steve was asking for suggestions to improve his, why do I need to do
yet another tool simply for the sake of it?

> But I want to avoid the problems that arises when someone stands up and
> judges whats an old dying trend and what's a new trend. If Steve would
> randomly remove stuff that he thinks is old and irrelevant - would make the
> table probably almost useless for me.

We all make judgements about tags to use all the time, these decisions
are updated as things are updated on wiki pages, as discussions occur
on mailing lists and even face to face mapping parties, knowing old
trends doesn't mean we should highlight them just because it was a
popular way to do things in the past, doesn't mean this is a sane way
to continue into the future, there is a good reason things get
depreciated.

> The current table gives an overview of what the current situation is. What
> you are talking about is a trend analysis of what the situation might be in
> some months from now. That is an interesting (but different) topic.

No, I'm talking what has happened recently, I'm not talking about
speculating into the future, I'm talking about setting a cut off point
that doesn't make sense to keep looking at old data because of many
reasons already pointed out in this email.

> If I want to make a map, then I'm probably interested that abutters is used
> n-times, so I can judge myself if I want to add it or not. I'm not arguing

Just knowing how many times it is used is a pointless exercise,
someone might have bulk imported large chunks of data in the past
which has no bearing on current best practises. It is foolish to look
at raw numbers and not interupt what the bigger context is.

> that having an indicator that it's usage numbers are e.g. going up or down
> would be a big plus to make good educated decisions here.

How can you tell that unless you do trend analysis based on
differences over time, which doesn't seem to be currently employed.

> But that *is* the current situation. The current situation *is* an average
> of old and new stuff.

No, that is an overview, not the current situation, the current
situation is based on the current trend, not what happened 5 years
ago, or what happened due to a massive bulk import. Those are
exceptions that will skew trends.

> So how do the mappers out there know? From the voices inside their heads?
> ;-)

There is a lot of people that use information from the wiki -- not
just the map features page, map data in their area and suggestions
from mailing lists, and just what makes sense based on other tags.