Re: [Talk-GB] Postboxes & Payphones

[Ed Avis]

Perhaps I should give a little bit of background, explaining why I have started
reconciling Dracos and OSM data, hoping to import it.

I've been on many mapping trips around London, to start with mostly filling in
missing street names.  As the street names have become more complete I have
focused more on POIs such as shops, named buildings and of course post boxes. 
If I spot a post box I will usually cross the road to take down its ref number.

However, I've sometimes noticed after a mapping expedition that all the
postboxes I went hunting for are already known in the Dracos data, but nobody
had bothered to push them upstream to OSM.  This is quite dispiriting, and while
mapping is a fun thing to do, volunteers are still a scarce resource.  While I'm
happy to resurvey and refine existing data (as I sometimes do for Naptan bus
stops), nobody wants to feel they are wasting their time.

Now, if the data is of poor quality, bad enough to be worse than nothing (n.b.
there are certainly mistakes in the existing surveyed OSM boxes) then of course
it shouldn't be imported.  But I don't think there is any great divide between
armchair data importers and rugged, rosy-cheeked outdoor mappers.  Often the
motivation for using more automated tools or external data is to help the
surveying work.  For example it is certainly easier to walk round and check
existing bus stop nodes, which are at least
*approximately* in the right place, than to find them all from scratch.

Anyway, all this is moot until we have a bit more information on the quality of
this particular data source.  Let me make some reports on that (hopefully this
weekend) so we can decide which parts of the data, if any, are worthy of being
included in OSM.  And if you know perl, feel free to try out the code from
 and see what it generates for your area. 
Any additional sanity checks or reconciliations that can be suggested are
welcome - I will certainly add some more myself.

--
Ed Avis 


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Andy Allan]

On Mon, Nov 9, 2009 at 4:06 PM, Andy Allan  wrote:
> I'd rather have a
> partially complete, high quality map than a bodge of crap data
> (NAPTAN, anyone?) that just encourages more people to add crap data
> and get the "community to fix it".

I've been mulling over this point that I made, and to save the ML
servers some traffic I put my thoughts here instead:

http://www.gravitystorm.co.uk/shine/archives/2009/11/10/the-pottery-club/

I hope that illustrates my point of view better, of course anyone can
feel free to disagree with me :-)

Cheers,
Andy

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Andy Allan]

On Mon, Nov 9, 2009 at 11:38 PM, Tom Taylor  wrote:
> On 9 Nov 2009, at 16:06, Andy Allan wrote:
>
>> Please don't. Imports are a real, real annoyance. There's no
>> indication that anyone using the dracos site has "located" these with
>> any accuracy or personal knowledge - I wouldn't be surprised to find
>> many just somewhere "near" the road in question. I'd rather have a
>> partially complete, high quality map than a bodge of crap data
>> (NAPTAN, anyone?) that just encourages more people to add crap data
>> and get the "community to fix it".
>
> There's no indication in areas with a single mapper that they haven't just
> made stuff up in OSM either, but we assume that eventually the correct data
> will out.

Sure, in theory. But I'm more interested in what actually happens than
a theoretical debate. Out of the 4,600 mappers here in the UK we know
that around a dozen have made stuff up. So for any individual mapper
we know that, with a very high probability,  the data is of a good
quality. Additionally, we know that out of the few dozen significant
imports to OSM, all bar one or two (AND) have been of a much lower
quality than we would expect from a volunteer.

Cheers,
Andy

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Andy Allan]

On Mon, Nov 9, 2009 at 11:49 PM, Frederik Ramm  wrote:
> Hi,
>
> Andy Allan wrote:
>>
>> automated imports of sub-standard data
>
> It is probably wishful thinking to hope that they will go away.

Not really. We could state that imports are not permitted, and enforce
it, if we want to. "Wishful thinking" is too close to "lost cause",
and we're not there yet.

> But one
> thing we could do to limit damage is to have something like "layering" in
> the OSM data base;

I disagree, but only with your suggestion not the intent behind it. If
we take a step back, then our editors (especially JOSM) can cope with
a variety of "aides" or "sources" for the mappers. We have GPX traces
(both local and central), raster aerial imagery (from multiple places,
not just Yahoo!), geo-tagged photos, raster maps (PD and/or OOC),
audio, video, scribbles on printed maps. The bit that is missing is
being able to use vector map sources (e.g. shapefiles, CSVs of
postboxes or WFS services) in a similar fashion - as assistance to a
mapper in the same way as everything else. So if I'm adding a
building, I can use aerial imagery to assist, if I'm adding postboxes,
I can use the dracos list to assist.

I think with Vector Sources support in editors we can massively reduce
the need for bulk imports.

Cheers,
Andy

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Chris Hill]

Ed Loach wrote:
>> As has been
>> mentioned before, the process of hunting down an elusive
>> postbox often has
>> the benefit of some other missing feature getting mapped as
>> well, just
>> because you happen to be in the neighbourhood.
>> 
>
> I'd agree with that (and can also think of two adjacent postboxes
> with different references, which I hadn't considered earlier). This
> morning I was out verifying bus stops before work (more accurately I
> was trying to get to Colchester for 06:30 when Wickes opened, so I
> could be home again before work, but stopped at many of the stops en
> route except for a few where they loomed out of the mist too late
> for me to stop safely with a vehicle close behind me) and found a
> whole new lane I'd missed every other time I'd driven past it, which
> is often. I think I also spotted some recently constructed
> residential cul-de-sacs that I don't remember being there last time
> I passed, so will check and perhaps map those when I go to Wickes
> next (when their forklift driver gets to work they'll move the stock
> from out back into store so I can actually buy it). So probably this
> evening, though might not as I may take a different route to
> Colchester and try and get the route of the 74 verified to the same
> level as the 76 route I followed this morning.
>
> But if I weren't trying to spot bus stops I wouldn't have spotted
> the lane I think I had previously dismissed as a driveway (which as
> it only goes to a farm is perhaps what it effectively is).
>
> Although perhaps bus stops is a bad example as that is a case where
> the data has been imported. 
>
> Ed
>
>   
I have hunted post boxes and bus stops.  I used the RM list to find post 
boxes.  It was a struggle to find some (two seem to have vanished off 
the face of the earth), but in the process I improved all sorts of 
things, POIs, speed limits, corrected some mis-spellings, found closed 
pubs etc.  Post boxes are scattered all over the place; the locations 
given on the RM list are often cryptic and sometimes just plain wrong.  
The Draco site allows anyone to position a box anywhere.  Just pick a 
box off the list and plonk it on the map, which is exactly what some 
people have done.  Some people have used the RM description and assumed 
the location, without checking.  That is why I think loading Draco's 
data is a bad idea. Using the original FoIA response list to find boxes 
on the ground is a completely different process.

I have used the NaPTAN import to find bus stops.  I treat it as a 
similar process, except that the NaPTAN stops were surveyed in the first 
place (yeah I know it doesn't always seem like it).  You can apply 
corrections to the imported data to improve them and by reporting any 
discrepancies to your local council there is a chance that they will 
improve NaPTAN too. 

I am skeptical about the idea that data imports improve OSM.  I think 
they can be valuable, but I think they should be imported for an area 
where someone local to that area can take responsibility for checking 
them, either before or after importing, so that their true value can be 
realised and we don't just import junk.

Cheers, Chris

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Ed Loach]

> As has been
> mentioned before, the process of hunting down an elusive
> postbox often has
> the benefit of some other missing feature getting mapped as
> well, just
> because you happen to be in the neighbourhood.

I'd agree with that (and can also think of two adjacent postboxes
with different references, which I hadn't considered earlier). This
morning I was out verifying bus stops before work (more accurately I
was trying to get to Colchester for 06:30 when Wickes opened, so I
could be home again before work, but stopped at many of the stops en
route except for a few where they loomed out of the mist too late
for me to stop safely with a vehicle close behind me) and found a
whole new lane I'd missed every other time I'd driven past it, which
is often. I think I also spotted some recently constructed
residential cul-de-sacs that I don't remember being there last time
I passed, so will check and perhaps map those when I go to Wickes
next (when their forklift driver gets to work they'll move the stock
from out back into store so I can actually buy it). So probably this
evening, though might not as I may take a different route to
Colchester and try and get the route of the 74 verified to the same
level as the 76 route I followed this morning.

But if I weren't trying to spot bus stops I wouldn't have spotted
the lane I think I had previously dismissed as a driveway (which as
it only goes to a farm is perhaps what it effectively is).

Although perhaps bus stops is a bad example as that is a case where
the data has been imported. 

Ed



___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Gregory Williams]

It's potentially better, but still wrong though. For a start I've come
across several cases of two postboxes with different refs that are located
side-by-side. It's a bit of a lottery whether it'll tie up the correct ref
here. Also, I've seen several instances of bad data in the Dracos set, e.g.
a postbox in the middle of a farmer's field, and another out at sea. Tying
up to a dodgy location in the first place will just result in dodgy data.

Personally I don't want to see the Dracos data imported into OSM either.
We've got plenty of people mapping on the ground now and we only need a
little patience to have mapped all of the boxes directly. As has been
mentioned before, the process of hunting down an elusive postbox often has
the benefit of some other missing feature getting mapped as well, just
because you happen to be in the neighbourhood.

Gregory

> -Original Message-
> From: talk-gb-boun...@openstreetmap.org [mailto:talk-gb-
> boun...@openstreetmap.org] On Behalf Of Ed Loach
> Sent: 10 November 2009 08:20
> To: talk-gb@openstreetmap.org
> Subject: Re: [Talk-GB] Postboxes & Payphones
> 
> I just noticed a changeset from user elbatrop which although it says
> it was to "Tie 10 Royal Mail references to known postboxes", it has
> 1709 nodes in the changeset:
> http://www.openstreetmap.org/browse/changeset/3079245
> 
> Assuming this is import related, then linking a ref to a previously
> mapped postbox is probably the least likely to cause issues.
> 
> Ed
> 
> 
> 
> ___
> Talk-GB mailing list
> Talk-GB@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-gb


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Ed Loach]

I just noticed a changeset from user elbatrop which although it says
it was to "Tie 10 Royal Mail references to known postboxes", it has
1709 nodes in the changeset:
http://www.openstreetmap.org/browse/changeset/3079245

Assuming this is import related, then linking a ref to a previously
mapped postbox is probably the least likely to cause issues.

Ed



___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Frederik Ramm]

Hi,

Andy Allan wrote:
> automated imports of sub-standard data 

It is probably wishful thinking to hope that they will go away. But one 
thing we could do to limit damage is to have something like "layering" 
in the OSM data base; not thematic layering like in traditional GIS, but 
source layering. Stuff that gets imported gets onto a layer of its own, 
and then it can be transferred to "OSM proper" by a click of a button 
somewhere in an editor.

Continuing that line of thought, every user could have his (or any 
number of) own layers where he could "work on stuff". - Of course, lots 
of technical issues especially if what you're importing is somehow 
connected to what's already there.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Tom Taylor]

On 9 Nov 2009, at 16:06, Andy Allan wrote:

> Please don't. Imports are a real, real annoyance. There's no
> indication that anyone using the dracos site has "located" these with
> any accuracy or personal knowledge - I wouldn't be surprised to find
> many just somewhere "near" the road in question. I'd rather have a
> partially complete, high quality map than a bodge of crap data
> (NAPTAN, anyone?) that just encourages more people to add crap data
> and get the "community to fix it".

There's no indication in areas with a single mapper that they haven't  
just made stuff up in OSM either, but we assume that eventually the  
correct data will out.

What Matthew Somerville (who runs Dracos.co.uk) has built is a tool  
designed for doing one thing, very easily. And if OSM is going to  
reach more people, then it needs to work out how to integrate with  
tools and services that meet a specific need, rather than just mapping  
in general.

Most people don't care about making a map, but they do know where  
their nearest postbox is, when asked that specific question.

I have no idea what the quality of the data is like in this case, but  
let's make sure that OSM can work with external services when they  
have a lot to offer.

Cheers,

Tom

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Jerry Clough - OSM]

Can I add my voice to those who would prefer that these are not imported? 

There are several reasons:

1. Data quality is not assured.
2. Sourcing. For instance I have located some postboxes on Dracos site, 
BUT have not placed them in OSM because I've used information which may not be 
compatible with OSM's terms. I would not like these to be imported after I'd 
made a decision not to add them directly. We have no knowledge of the person 
who added the data, how they came by the information and whether it uses 
sources such as the PO post office finder.
3. Spoiling the fun. Hunting down the odd poorly described postbox 
often results in much more information and better mapping of the adjacent area. 
Importing the references and locations removes this incentive for a fairly 
trivial gain.
4. Distorting mapping priorities. Imports of postboxes will give an 
erroneous impression that an area has been surveyed on the ground and lead to 
further neglect. Furthermore an import of postboxes may cause mappers in an 
area to re-survey them to ensure accuracy, thus diverting them from their own 
goals (this is more-or-less the same affect as vandalism).I am not opposed to 
imports provided that they add value to the mapping process, and are controlled 
or requested by active mappers in an area (e.g., NaPTAN). All imports require 
follow-up, both in tidying up the data, and checking its validity: only perform 
the import if everyone is happy with the implied commitment.

Jerry Clough
SK53





From: Ed Avis 
To: talk-gb@openstreetmap.org
Sent: Mon, 9 November, 2009 14:23:51
Subject: Re: [Talk-GB] Postboxes & Payphones

Mike  writes:

>I've been looking at the Dracos postbox list
>http://www.dracos.co.uk/play/locating-postboxes/

I have recently started work on importing this,
see <http://wiki.openstreetmap.org/wiki/Dracos_import>.
So far I have just done the E10 and E17 postcodes near where I live.
After a bit of consultation I will start importing the rest of the country.

>and the UK payphone DB: http://payphonedb.bobsbasement.co.uk/.

I didn't know about that, I might look once post boxes are done ;-p.

>Looking through
>the OSM data, there's a lot of postboxes and payphoens but very few have
>the appropriate tags to tie them into the other databases.  Would OSMers
>mind tagging any postboxes/phones that they come across to tie them in
>with these side projects?

Yes, many contributors do this for post boxes, in London at least.
My import job also tries to match up post boxes by location, and sets the
'ref' tag to record the match found.

-- 
Ed Avis 


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb



  ___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Ed Avis]

Chris Hill  writes:

>Please don't import the Draco database into OSM.  The quality is very 
>dubious in places.  The original FoIA data that it is based on is very, 
>very general.  They have to be surveyed on the ground to confirm that 
>they are even within 200m.

I acknowledge your and Andy Allan's concerns.  I will do some auditing on
the data (by comparing non-OSM-based Dracos data with what's currently in OSM)
and report back.

There are several different components to the Dracos data which can be
imported separately.  Where a Dracos user has simply added info such as
collection times to an existing postbox node that came from OSM, this is
more likely to be reliable.  If a postbox exists in the Dracos list but is
not in OSM at all, then at a minimum this indicates an area to be surveyed.
I do not propose to uncritically suck all the Dracos postbox data into OSM
but rather to reconcile the two with plenty of sanity checks.

In my test import of E10 and E17, almost all of the changes were adding
information to existing OSM postbox nodes, with only a handful of missing
postboxes to add (which I am happy to survey myself).  Where Dracos and OSM
disagree, the OSM data is left alone.

Anyway, I'll let you know the results of the audit; if the location data
really is as bad as you say, I won't enter it into OSM.

-- 
Ed Avis 


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Chris Hill]

Ed Avis wrote:
> Mike  writes:
>
>   
>> I've been looking at the Dracos postbox list
>> http://www.dracos.co.uk/play/locating-postboxes/
>> 
>
> I have recently started work on importing this,
> see .
> So far I have just done the E10 and E17 postcodes near where I live.
> After a bit of consultation I will start importing the rest of the country.
>
>   
Please don't import the Draco database into OSM.  The quality is very 
dubious in places.  The original FoIA data that it is based on is very, 
very general.  They have to be surveyed on the ground to confirm that 
they are even within 200m.

I want OSM to be the best map in the world - we'll never get that by 
importing other people's junk.

Cheers, Chris

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Andy Allan]

On Mon, Nov 9, 2009 at 2:23 PM, Ed Avis  wrote:
> Mike  writes:
>
>>I've been looking at the Dracos postbox list
>>http://www.dracos.co.uk/play/locating-postboxes/
>
> I have recently started work on importing this,
> see .
> So far I have just done the E10 and E17 postcodes near where I live.
> After a bit of consultation I will start importing the rest of the country.

Please don't. Imports are a real, real annoyance. There's no
indication that anyone using the dracos site has "located" these with
any accuracy or personal knowledge - I wouldn't be surprised to find
many just somewhere "near" the road in question. I'd rather have a
partially complete, high quality map than a bodge of crap data
(NAPTAN, anyone?) that just encourages more people to add crap data
and get the "community to fix it".

Let real mappers, in the real world, do the mapping. That's what we do
best. Leave the automated imports of sub-standard data to our
colleagues in the US.

Cheers,
Andy

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Ed Avis]

Mike  writes:

>I've been looking at the Dracos postbox list
>http://www.dracos.co.uk/play/locating-postboxes/

I have recently started work on importing this,
see .
So far I have just done the E10 and E17 postcodes near where I live.
After a bit of consultation I will start importing the rest of the country.

>and the UK payphone DB: http://payphonedb.bobsbasement.co.uk/.

I didn't know about that, I might look once post boxes are done ;-p.

>Looking through
>the OSM data, there's a lot of postboxes and payphoens but very few have
>the appropriate tags to tie them into the other databases.  Would OSMers
>mind tagging any postboxes/phones that they come across to tie them in
>with these side projects?

Yes, many contributors do this for post boxes, in London at least.
My import job also tries to match up post boxes by location, and sets the
'ref' tag to record the match found.

-- 
Ed Avis 


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb

Re: [Talk-GB] Postboxes & Payphones

[Shaun McDonald]

On 7 Nov 2009, at 01:36, Craig Wallace wrote:

> On 06/11/2009 17:48, Mike wrote:
>> Folks,
>>
>> I've been looking at the Dracos postbox list
>> http://www.dracos.co.uk/play/locating-postboxes/ and the UK  
>> payphone DB:
>> http://payphonedb.bobsbasement.co.uk/.  Both tie into the OSM  
>> (postboxes
>> keyed with a ref: tag and phone boxes with telephone:  Looking  
>> through
>> the OSM data, there's a lot of postboxes and payphoens but very few  
>> have
>> the appropriate tags to tie them into the other databases.  Would  
>> OSMers
>> mind tagging any postboxes/phones that they come across to tie them  
>> in
>> with these side projects?  Perhaps we could even have a virtual (ie  
>> UK
>> wide) mapping party to update the map with these?
>>
>> What do people think?
>>
> I agree with you on this, its useful to add refs to postboxes. I've  
> been
> working through them in my area, and ticking them off on the Dracos  
> list.

Every so often the Dracos site will import the OSM data, so you don't  
need to manually enter it there, it will automatically be picked up  
after a few weeks.

I know some mappers do go and specifically collect the refs of post  
boxes.

>
> I've not seen that payphone DB before, how does it work? Does it
> automatically get payphones from OSM?
> I've tried searching in my area, but it doesn't find any. Do they need
> particular tags to appear on that site? I have tagged a few with their
> phone number, but can't find them if searching for that. And it  
> doesn't
> find anything at all if searching in my county.

I don't think this site is integrated with OSM.

Shaun

>
> Craig
>
> ___
> Talk-GB mailing list
> Talk-GB@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-gb


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-gb