[OSM-talk] Suggestion more complete mapping verifactiion

[D Tucny]

On 18/09/2007, John Baker <rovastar at gmail.com> wrote:
>
> I started this post because I looked at the map of central London that was
> meant to be complete and decided to some roads where I worked many years ago
>
>
> Heddon Street just off regent street, Central London, W1 was one.
>
> http://www.streetmap.co.uk/newmap.srf?x=529233&y=180753&z=1&sv=heddon+street&st=1&tl=Heddon+Street,+W1&searchp=newsearch.srf&mapp=newmap.srf
>
> I though it would be a good test case a small street and it is/was a
> cul-de-sac road with a walkway (no cars can get through) at the north end
> not a straight through crescent-like road like streetmap, a to z, etc
> display it as. (also then I noticed that Bean Street in OSM is really Beak
> Street)
>
> Now the map of that area appears all complete but this road was completely
> missing. Before I saw that I had confidence that the data was correct after
> I didn't. Of course I can (and will when I am next that way) map it myself
> but I have no GPS, etc. I know I can still add stuff but I started thinking
> with my IT head on about an easier way to sort this as I have little
> confidence in an area of the map that many believed complete and unlikely to
> be checked again. I wondered how many other areas like this are there. I
> presume someone carefully checked the area like many of you do (street
> walking/aerial photos) when mapping an area and simply a street got missed
> and it was/is not tagged as missed.
>

I've been down Heddon Street, not with a GPS, but, I've been there, eaten
there, drank there etc... from what I remember it's quite a small 'alley'
with reasonably tall buildings on each side with little natural light making
it's way in... We don't seem to have any public GPS traces for it looking at
the potlatch, don't know what other people have, I'd guess this would be one
street that would be very difficult to get a decent GPS trace off, looking
at the yahoo imagery, it seems to be pretty low res over this area... So...
This one will be challenge to get right... a number of GPS traces from some
very sensitive GPS devices might help, as might some higher res yahoo
imagery, but short of that, more traditional surveying tools may be
needed...

I think that in cases like this, i.e. you looked, you know some things are
wrong and/or missed, but you don't have the tools or are not in the right
place to be able to resolve the issue, then some easy way of reporting the
inaccuracy could be useful... Something along the lines of the tomtom error
reporting whereby you point at the place and give a text description of the
issue, e.g. Heddon St missing, off Regent St between New Burlington St and
Vigo St... As long as this was typed in by people, and didn't lead directly
to changes to topology, shape or names, i.e. someone took that information
as a hint for where to go resurvey an area, then I think that any data that
came from it is clean and safe, probably even if as a user, you noticed and
reported this inaccuracy with the aid of 3rd party sources... Though how we
publish these reports is a tricky one... We want as many people as possible
to see what errors have been reported so they can fix them, but, if these
reports have been based on copyrighted data, we shouldn't be publishing
them...


Obviously over time the wiki streetmap will get better but I thought we
> could simply compare map data with other (official) sources to get a more
> complete map.
>

As far as I'm aware, it's perfectly OK to for example, look at google maps,
look at OSM, and if you see differences, add them to whatever route you are
next planning on taking armed with GPS, camera, notebook etc... As long as
the actual data entered into OSM is from having visited and taken pictures
of signage, the data is clean, i.e. nothing has come from google maps at
all, except information used to plan a route which took you to areas that
weren't in OSM where information was independently gathered... But... OSM
can't publish other people's information without a license to do so, so we
couldn't for example capture bits of multimap's maps, modify them to
highlight areas that have been missed then publish them on OSM sites, while
the data that is entered into the database might well be safe, having been
captured correctly, the issue would be the fact that we'd misused multimap's
maps in our planning... equally, if we did something like produce and
publish some form of 'diff' of maps between OSM and multimap's maps, that
would also be dodgy...


I know nothing about mapping and came across this project a week ago by
> chance and I do much a fair few open source/free projects on the web and
> thought I would help out.
>
> There are 2 issues I can see here:
>
> a) If you could check 100% legally the streets against a known source of
> streets would you find this helpful? I cannot see how you would not find
> this helpful but many replies here seem to imply this.
>
> I only know about the UK and know little about mapping/street indexes
> hence the focus on them here. But surely this would be useful where there is
> a complete or even decent legal resource of street names. You can use it in
> conjunction with any other system you could check multiple databases against
> the OSM data.
>

I'm sure there are plenty of uses for a free to use, legal, complete street
name list... The mind boggles... But... For the UK, there isn't one that
we're aware of...

I would have thought that some countries out there would have more relaxed
> laws about the data and this could be very possible.
>

The US springs to mind... They have lots of streetnames in their TIGER
dataset, even more useful, it also has lots more information, as such, it
can be used on it's own to make map data... The import of this is happening
now...

To be honest, I seem to be finding hostility towards this idea in general
> and I don't understand way. It seem that some like to make things difficult
> for themselves and want to do *everything* by hand and don't want help.
>

I think the issue is, that people have been through these thought processes
before, they've carefully weighed up the pros and cons, the risks vs the
benefits, and they've not found any suitable such datasource for use by the
project as a whole... Sure, for you personally, you can buy a license to the
street name data, you can compare that to the OSM data and produce reports
to your hearts content, you can use those reports to make your own lists of
what you think it missing, then you can go to those places, capture the
information cleanly and correct it, but, you would need to be able to prove
that anything that you did publish, and anything that you added to OSM was
from personally having collected the data and not derived from the data you
purchased a license to... OSM would want this to show the data is clean, you
would want this so that you didn't personally get hit... If you were to take
the PAF, and the OSM data and produce a list of street name missing from OSM
that were in the PAF, then publish them along with their approximate
locations on a website, as this information was what was missing from OSM,
it would be entirely PAF sourced, so, I'd guess RM would be knocking at your
door shortly after they noticed what you'd done...

It is just a more pro-active way of helping the project progress. If German
> has relexed laws about this an a list of streets is know for say Berlin and
> if Berlin was nearly finished then we could compare the two. I would have
> thought it would really help out say a small group of dedicated mappers that
> have nearly completed a city, we missed a few streets there rather than
> walking the streets/looking again at the aerial maps of the area that they
> think they have mapped perfectly.
>

Yep, great, that could be very useful... But, the problem is that the key
word there is 'If'... We don't have that data... If all map data worldwide
was PD, it would make it even easier, but then, we probably wouldn't be
trying to create a map, we'd probably be working on doing some other cool
stuff with all the free data around...

Once you have confidence that all the streets are mapped then you can help
> promote OSM to local councils, business, etc, etc
>

 That's happening now in a number of places... Some people, businesses etc
like the fact that even if the data isn't 100% 'complete' that it's free and
complete-enough for their purposes... If they find issues, they can fix it
themselves, or report it to others who can investigate, especially in areas
that have active mappers... Though as I said above, making it easier for
'users' to report issues could be useful, though it would need someone to
come up with a way of doing that, just talking about it isn't going to make
it happen...

In the wiki of the UK there are many cities that state they are nearly
> complete (or complete!!) I imagine some of these are done by one or a few
> mapper wouldn't it be nice to check their data. I dont mean to damage egos
> by saying that sorry you missed x, y, z road.
>
> I find it surprising that you appear not to want to cross reference the
> data with another source - or in fact as many sources as possible. All
> legally of course.
>

Yes, checking the data is very useful... I don't know of any one who's egos
would be damaged if you pointed out x, y and z roads were missing... In
fact, in most cases I'd guess those involved would feel pretty good about
that fact that considering the vast amount of time they had put into doing
the work that someone other than them had looked close enough at their work
to find such issues... Legally is the key point here... You saying 'Bean St'
is actually 'Beak St' as an example... OK, in this case you do really know
the road, but, lets say you'd used your PAF to find this error and didn't
know the road... What do you do? Compare it to google maps? also shows Beak
St, compare it to Multimap, also shows Beak St, so, just update OSM? NO...
Because then, you've taken the name data from PAF, google and multimap, the
data is dirty... It's not legal... You could go visit it yourself, take the
photo of a road sign that says it's Beak St, then go home and update it,
that's fine... You could let some individuals know that you believe that
Bean St is actually spelled Beak St having looked at a number of non-free
data sources, they could go take some photos and update OSM, that's fine.
You could post a page on the wiki with this and all other names you found to
be wrong or missing, not fine, again, publishing data sourced from non-free
data sources is bad... so, yes, checking is important and people do want to
do this, but it must also be legal...

b) The legality of the UK data.
>
> I am no expert on this but have dealt a fair bit with legal software
> licenses and tbh I could not see too much wrong with the RM licenses. I can
> understand not wanting to put the project in jeopardy although first you are
> likely to get a C&D/takedown notice before any action is taken and open
> source project have a lot more leeway to as anyone could just restart the
> project again with the data in it. And once started again the companies
> would have to prove that data came from there source with what I am saying
> there is no way they can.
>
> The data would have been inserted in to the database manually from
> manually looking at street signs. I don't care how big your lawyer wallet is
> I cannot see anyone holding it up for a new project especially based on the
> same data.
>

So, we shouldn't worry about legal issues as we can just fork the project
incase if legal issues?

I am at a loss once it is in the database who can tell. No Easter eggs will
> be there as people have gone by hand to check it like what happened to the
> AA.
> It is like me comparing the data with the paper AtoZ by hand either
> looking at maps and saying that this street doesn't appear on our map.
>
> To be honest any info of street names would be useful even if they are not
> complete indexes and I cannot believe that there is no source for some info.
> I just don't want to do all that work myself.
>
> I think in general the project should be looking at linking up with more
> sources for verification. It seems a shame that no-one considers this.
>

OK, back around again, yes, having as many sources of verification as
possible = good...  So what data sources can we use?
PAF? No, we can't publish contents of this, while the data in the database
might be clean, what about the pages of lists of PAF contents that show
people what was missing? They'd be dirty... No go... Can you use the PAF
yourself to check the 'completeness' level of an area according to street
names? Probably... Can you publish this completeness level? Probably... Can
you publish the list of names missing? Probably not... Can you use the PAF
to make a map overlay showing missing information? Probably... Can you
publish this for the world to see? Probably not...

What else do we have?

It's a difficult one, it would be useful if we had lots of data to check
against, it would be useful if we had a list of streets that were freely
available with a compatible license... right now, I'm not aware that such a
thing exists... We can build one... And we are... with the added value of
having topology and geography linked to those names... And it's all
available to be used with a much more open license than other sources with
no cost attached... How cool is that?


If you do find a free, open source of information, please mention it, others
are looking for stuff all the time, but the world and the internet are big
places... In the meantime, feel free to continue coming up with ideas to
help the project, and remember, even without a GPS there is lots you can do
to help and if you happen to be passing Beak St or Heddon St any time soon,
snapping a few photos of the street signs and surround areas could well be
useful :)

d
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20070918/b725dfd0/attachment.html>