[Talk-us] Removing tiger:* tags

Alan Mintz Alan_Mintz+OSM at Earthlink.Net
Fri Jul 30 19:24:15 BST 2010


At 2010-07-30 07:28, Anthony wrote:
>On Fri, Jul 30, 2010 at 12:24 AM, Alan Millar <amillar503 at gmail.com> wrote:
> > I haven't seen a conclusion on what people want to see in the naming
> > convention (see for example the thread at
> > http://lists.openstreetmap.org/pipermail/talk-us/2010-April/003138.html).
> >
> > Just because the conversation is ongoing, that doesn't mean you need to
> > delete the data in the meantime.
>
>No, I don't need to, and I generally only do so when I'm otherwise
>editing the ways anyway.
>   I've explained my reasons, and I haven't
>heard anything to change my mind about them.

[ This is long. Sorry. ]

I really don't understand your argument. It's the nature of OSM that many 
people will contribute many types of data, much of which will not be cared 
about or understood by the majority of consumers. What's wrong with that, 
and why do you think removing it because you don't understand or like it is 
acceptable behavior in a crowd-sourced environment?

The only reason that makes sense might be "it's wrong". In the case of 
tiger:*, it's not wrong. It's in its own namespace because it indicates the 
value as it was in another database at the time of import. Not that I 
believe we need to justify it, but the three (at least) of us arguing to 
keep the tags in this thread, each for reasons that we've described, should 
be sufficient to prove that someone needs the data, and you really have no 
right to stomp on our work, or data that we need for our work. Also, we're 
not alone - many people recognized the need to fix the way names are 
stored. Having to go back to history will be adding an order of magnitude 
to the complexity of that.

Have a look at tagwatch and you'll see that tiger:* is just one of many 
such import namespaces, most of which you are not likely to care about, 
whether they are doc'd or not.

There's another, very important use for the "tiger:reviewed" tag. It was 
designed to let you know what ways need to be satellite- or GPS-aligned, 
since the original data was very poorly aligned. Having these render 
differently in JOSM is an important workflow tool. After I'm done aligning, 
I remove that tag, as documented in the wiki. When I've surveyed it in real 
life, I add source and source_ref tags to cite my source. BTW, someone 
started stomping on those as well because they saw no need for my picture 
#s[0], but after discussing it, was convinced to leave them alone.

Someone asked a ways back whether the tiger:* tags could be combined into a 
single value, which leads me to think that there is a hidden reason that at 
least two people don't want these. Does it have something to do with the 
editing tools being used? In JOSM, the tags appear in alpha order, which 
ends up placing them almost always below any of the commonly edited tags. 
Is the real problem that other editors aren't doing this, resulting in 
clutter in the editing process? Can't we just solve this in editors, maybe 
by placing the common import namespaces last in sort order?

FWIW, the only time I intentionally *remove* data is when I'm certain (or 
as close as possible to certain) that it is wrong, almost always replacing 
it with my own correct data. I believe this is one of the fundamental 
principles of the community, and would hope that others adhere to it. One 
recent exception is that, over a large chunk of southern California, a user 
had entered maxspeed values that were incorrectly converted from mph to kph 
using a wrong, and sometimes unpredictable, factor. I've moved the ones 
that I know to be wrong (because they are not integral multiples of 5 mph, 
are inconsistent with the road type, and were edited by this user) to 
bad_maxspeed=*. When adding the correct maxspeed from my own survey, I then 
remove the bad_maxspeed tag. Unfortunately, some remain with maxspeed=40, 
for which it is not possible to determine accuracy in all cases, but that's 
not a reason to remove them until I have proof. BTW, the same user also 
can't spell (English/American/Spanish names mostly), and I've spent a fair 
amount of time having to research and correct those, too. I don't wipe them 
out just because I think they're likely wrong, though, until I research them.

Notes:
[0] Those source_ref=AM909_* values in source_ref are links to the pics I 
have of the names of the streets. There are other source_ref:*=* values for 
other attributes that are proved by pics. At some point, they could be 
links to an online repository of these pics, given interest, and some 
post-processing to remove faces and license plates. Right now, they allow 
me to look back at partial surveys of attributes (like speed limits) and 
combine them with newer surveying to form a complete picture, and are an 
important part of my workflow.

--
Alan Mintz <Alan_Mintz+OSM at Earthlink.net>




More information about the Talk-us mailing list