[OSM-talk] Suggestion more complete mapping verifactiion
Andy_J_Robinson at blueyonder.co.uk
Tue Sep 18 00:19:29 BST 2007
John Baker [mailto:rovastar at gmail.com] wrote:
>Sent: 17 September 2007 11:59 PM
>To: Steve Coast
>Cc: Andy Robinson; talk at openstreetmap.org
>Subject: Re: [OSM-talk] Suggestion more complete mapping verifactiion
>On 17/09/2007, Steve Coast <steve at asklater.com> wrote:
> On 17 Sep 2007, at 22:26, John Baker wrote:
> | d) If someone posts to say Heddon Street in London W1 is not in
> database here is a link:
> | http://www.streetmap.co.uk/newmap.srf?
> +W1&searchp= newsearch.srf&mapp=newmap.srf
> | Is it then OK for someone to walk down Heddon Street with a GPS
> map this road in OSM?
> You have to prove they went down the road and didn't copy it, and it
> wasn't a purposful easter egg!
>So for every road mapped in in OSM you have to prove you went down that
>road and the GPS data is good enough proof not enough for you for it to be
>valid. I understand the termonology of Easter Eggs and I have mentioned
>them in numoureos email including that one.
>> 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.
>Of course it was considered, at least by me, but you weigh that up
>against getting nuked from orbit by people with space weapons
>(lawyers) like RM.
>But we're talking past each other, you haven't really answered my
>points about 192, or Dair's about accuracy of 'official' lists, and
>you don't want to prove your point by doing it.. so it's all a bit
>I am sorry I didn't realise you were awaiting a reply.
>I am unsure what points you want me to answer.
>The 192 case
>a) it was proven for 192 and there was no case to answer for.
>b) I am presuming they correct the information based on RM data? From the
>case it was not clear. I am not, or ever have, proposed doing this.
>c) we are taking a much smaller subset of the data. No postcodes, building
>names, house numbers. Although you might think (and maybe we are) breaking
>licensing on this it is a lesser offence.
>d) they publish full searchable content online. We publish the content
>online that we check by hand and not as much.
>I mean I could just type all the streets in the index of a city AtoZ in to
>check. Would you be more safe with that?
>To be honest Dair mail I didn't even think he understood what I was trying
>to do and didn't have time to explain as I have mentioned it before.
>I am fully aware that they might not be perfect but what harm will it do
>(all legalities aside) if you compare them with the OSM data?
>We have an area that we think is completed in OSM and the only way we can
>know this is well actually we cannot the only suggestion there was if it
>has a load of roads mapped on it we think it might be mapped? That is a
>carzy situation to be be in really. A basic benchmark would be to compare
>it to *something*.
>If we do say central london to an "official" source and there are say 100
>roads in the "official" list that do not match up to ours. Potential
>omissions, mistakes. Then someone/many people walk the streets and check
>these questionable area and in the end we find that 10 of those roads the
>"offical" source has "wrong" maybe some are Easter Egss, some simple
>mistakes, or a street was recently renamed.
>We will update/correct our OSM data and now it is more complete. Do you
>In fact we have better street name data then the "official" source like
>Royal Mail or OS. We could even let them know that we think there are
>mistakes in their maps.
>It doesn't matter if the offical maps are "wrong" but using their data our
>can be more complete.
>How else do you check?
>Even if you believe that the current central London OSM street name data
>(and that is all I am concerned with atm) is better than the OS/RM street
>index data. It can do no harm (all legalities aside) to check this. OSM can
>only improve. Who is going to check those areas again are we simply going
>to leave it just to chance? This would give us pointer that an area/street
>needs to looked at.
>Does this explain more?
>BTW looking again on google
>Maybe this would help for london at least.
Well, the T&C for London Online don't exactly give me confidence that I can
copy and use their street name listing so therefore I won't.
I'm still a little confused as to why you think a "complete" list of
streetnames is so imperative. I'm sure the OSM database will have as
complete a list of street names as any other database with time. It's a wiki
so it's pretty easy to fill in gaps, even if it takes a few iterations to
get it perfect. If someone finds a street is missing it gets added. Why does
it need verification?
I accept that it's perhaps not useful always to suggest a location has been
fully mapped and is hence "complete", especially if it's not had some sort
of quality control check, but marking places as complete is a useful measure
of progress and permits energy to be focused elsewhere. At some point, as
areas get mapped out, the OSM userbase will fall back to checking and
infilling and generally improving the data they have already uploaded, and
that will include fixing errors and missing data.
OSM is still in its infancy with a lot of data to add even for the most
heavily mapped areas such as the UK. Give places like London another year
and I think the situation will look a lot different and the number of
streetnames in the database for a given city will probably more closely
reflect the actual number of streetnames to be found when you visit them.
The best way for us all to help this process along is to find a blank area,
map it and get those streetnames into the database. If you don't have a GPS
then stick to somewhere with Yahoo imagery. That's all there is to it.
Andy_J_Robinson at blueyonder.co.uk
More information about the talk