[Talk-us] Excellent progress, u.s.

Toby Murray toby.murray at gmail.com
Tue Apr 17 09:56:39 BST 2012

On Tue, Apr 17, 2012 at 2:22 AM, Alan Mintz
<Alan_Mintz+OSM at earthlink.net> wrote:
> At 2012-04-16 20:41, Toby Murray wrote:
>> On Mon, Apr 16, 2012 at 8:37 PM, Nathan Edgars II <neroute2 at gmail.com>
>> wrote:
>> > On 4/16/2012 9:18 PM, Alan Mintz wrote:
>> >> At 2012-04-16 14:06, Nathan Edgars II wrote:
>> >>> Or you can simply add odbl=clean if there's nothing ungood about the
>> >>> object (e.g. it was split from a TIGER way and the splitting is
>> >>> something you would have done anyway).
>> >> Is this really sufficient? Can someone from the redaction squad
>> >> comment?
>> >> Can I protect/"bless" a way or node and prevent its redaction simply by
>> >> (in good faith) adding this tag?
>> > We have no idea what rules the OSMF will use.
>> Well I won't claim that communication has been great but this
>> statement is a little over dramatic.
>> First of all: odbl=clean *will* be honored.
>> ...
> On nodes as well as ways? As I wrote earlier, if I have tagged a way with a
> source that includes imagery, and removed the tiger:reviewed=no tag, it
> means I have aligned it to that imagery, including leaving nodes that are in
> the correct place alone (sometimes). Can I bless the nodes in the same way?

Yes. odbl=clean immediately removes any object from further processing
by the bot. See comments on the first function:

>> Also there is this:
>> http://wiki.openstreetmap.org/wiki/Open_Data_License/What_is_clean%3F
> A nice empty page. Tough to argue with :)

Works for me. You can search for "what is clean?" and it should be the
first result.

>> And of course the code is available for anyone to view... although I'm
>> not going to claim that this is really good documentation on the
>> matter:
>> https://github.com/zerebubuth/openstreetmap-license-change
> Nor can you reasonably expect people to use this as a guideline. And I'm a
> programmer.


>> There has been talk of the "v0 rule" which I believe is being
>> implemented in the code. This means that the act of creating an object
>> by a decliner doesn't automatically make it dirty. So if a way was
>> created by a decliner with the tag name=Fred and then someone else
>> added the tag highway=footway then after the bot gets done with it,
>> the way will still exist but only have the highway=footway tag. If an
>> accepting user changes the value of the name=* tag then it will be
>> clean... except, see the next paragraph. However if all of the way's
>> nodes are dirty and get removed then the way itself will have to go
>> too since you can't have a zero-node way.
> I contend, though, that you should not have to change a node to make it
> clean. If one has tagged a source with an imagery (or GPS) value, they are
> saying that they vouch for the position of the way, including its nodes.
> Same applies to removing tiger:reviewed=no (or gnis:reviewed=no). The user
> is specifically claiming to have reviewed the position and tagging and
> approved it. Should that not be sufficient?

I don't disagree with your points although things get complicated in
practice. I've seen people mass removing tiger:reviewed tags on any
way that they happened to load into JOSM while mapping when they
obviously didn't even look at it. Also, what if a user only
reviewed/improved the geometry of a dirty way in one spot but the way
is several miles long? This may not be the case for things you have
touched but there are a lot of people who have done a lot of edits
that are less rigorous.

>> Unfortunately neither badmap nor OSMI fully implement all of these
>> rules so yes there is still far too much uncertainty. But there are
>> some facts to be had.
> Why, then, is it acceptable for us to be sitting here with a dagger hanging
> over our heads, uncertain as to when and how it will fall? Shouldn't all of
> this be nailed down, followed by a reasonable notice period? Why is there a
> deadline other than "we need to get it done for the long-term benefit of
> OSM?"

Won't disagree with this either. Ideally the bot code would have been
developed over the past year instead of the past month and then been
available to make tools like OSMI and badmap that use the actual code
to show what will happen. But that's now how things happened. I'm not
trying to lay blame. I've been mostly a spectator to the process
myself so I'm certainly not going to throw stones.


More information about the Talk-us mailing list