OpenStreetMap

How about mapping all buildings in London before the Olympics?

Posted by Kachkaev on 3 March 2012 in English (English)

A lot of areas in London still look very poorly mapped. A few examples:

That’s between Buckingham Palace and Piccadilly Circus, just at one of the Olympic venues:
click to se this map on osm.org

That’s around Harrods:
click to se this map on osm.org

Here is a residential area east from Olympic park, a place where a lot of visitors will probably rent property.
click to se this map on osm.org

I can give tones of such examples, but you get the point anyway. If you look at how buildings are mapped in London in general, you’ll see that even zone 1 is far from being well covered:
click to se this map on ITO website

And this is in spite of the fact that Bing imagery is just fascinating almost everywhere in London!

What I would like to suggest is to run a global online mapping party and trace as much buildings as we all can before Summer, thus giving future Olympics visitors a more complete map than we have at the moment.

I know that many contributors are against such massive “armchair mappings”. I respect this view and the reasoning, but let me go through the arguments I remember and discuss them:

  1. Having all building outlines does not mean that the map is complete; we are still missing addresses, POIs and heights.
  2. Because imagery is usually old, there is a chance to miss a building or map a demolished one.
  3. Such online mappings harm local communities.

Ok, the first one: Indeed, it’s sometimes hard to distinguish between a remotely traced area and a properly surveyed one when you’re looking at a slippy map thinking of the location for the next mapping party. But we’ve got other tools, don’t we? Just with three basic steps in JOSM I can determine what areas are not surveyed: 1) search for “building=* AND -“addr:housenumber" -"addr:housename"” 2) delete the selection 3) look for empty areas. Pretty simple, isn’t it? Now guess, how many buildings with names/numbers are there in Central London now? Less than 20%. No one seems to feel any worries about that so far.

Now about the reliability of the imagery we have. Bing hi-res (20) is relatively old (Aug 2007) but zoom level 19 and less are Oct 2011! And London is not Dubai, where the landscape changes every year :) In my personal experience I remember only two cases of outdated Bing in London: a demolished building near Manchester Square and Angel centre that has been removed and rebuild on the same place, slightly changing its shape – all now seen on new imagery. The rest of the buildings I’ve surveyed are exactly the same as on Bing. They’ve been there for more than a hundred years and will stay the same for another hundred or more. So isn’t it worth adding much more data with just a few mistakes than having almost no newly added data at all?

3. If we have pre-traced buildings, we’ll be able make our mapping parties more interesting and productive. We’ll be able to cover bigger areas, as it used to be years ago when mostly street names and some POIs were collected. I don’t really mind spending an extra hour on walking papers and photos of opening hours signs, but the thing that I’ll have to spend extra 2 hours on tracing the architecture in that additional “cake slice” stops me every time – that’s too much. I usually pre-trace the outlines the day before anyway just with Bing, and they change very rarely after the party, being honest. Supplementing and polishing in this case is easier than doing everything from scratch — it’s not the same as hacking someone else’s code :) London is really huge and even if we have all buildings mapped, there will be still hell a lot of work for everyone to do on mapping parties and between them. Apart from that, Russian experience in doing these online mapping parties shows that such things bring new people to the community afterwards.

Here is another argument in support of the idea: Not all uses of OSM involve addresses, POIs etc. There are many services that just want our data for the background, for instance, foursquare or TFL’s online countdown service (zoom in to see buildings) — that’s millions of users daily, and they want a fancy map now, not in 5 years. Why to make those users think that a cited “OSM and contributors (whatever that means)” has an very incomplete map “with even no outline of my office”?

Thus, I have a strong feeling that such experiment won’t harm the map or community and instead add a huge value both to data users and contributors in London. I do believe that OpenStreetMap can and must be very popular among the Olympics visitors, especially those that will use offline apps on their smartphones. And 6-8 local mapping parties we’ll have before July is far not enough, we do need help from overseas.

This online mapping party can last for weeks starting soon, before it gets warm enough to go out and map, and we’ll invite everyone to take part. Maybe it can even be a project of the week, month or whatever. We can also organize some local offline workshops for newbies or something like this to support the London 2012 mapping challenge and bring new people in.

I’m not going to take any further actions like creating mapcraft cakes and wiki pages with instructions as I am not Londoner and don’t feel like can launch such stuff without everyone else’s approval. This note is just a call for an open discussion of the problem. If other OSM contributors in London agree with this idea, we’ll think of what we should do on one of our pub meet-ups and see how to organize the process.

Please leave your thoughts in the comments.

Alex

Location: Westminster, Whitehall, City of Westminster, London, Greater London, England, WC2E, United Kingdom

Comment from Milliams on 3 March 2012 at 15:39

I think that this is a great goal to aim for. However, I think that the default Mapnik style will have to tome down the building colours a bit so that London doesn't look like one big purpley blob.

Hide this comment

Comment from Former OSM contributor on 3 March 2012 at 17:10

Alex

This looks like a project that would require a vast amount time and energy just when the UK map is about to lose a swathe of data due to the licence changes. On top of that, according to ITO analysis, there are over 100,000 roads still missing from the UK map even before that change!

To fulfil your wish, maybe we could convince some of our fellow overseas mappers to chip in and lend a hand with your Olympic dream?

Hide this comment

Comment from chriscf on 3 March 2012 at 17:43

Something else to remember is that the Bing imagery may be out of date, so without some local knowledge, you may be tracing buildings that just aren't there or missing some that are.

Hide this comment

Comment from Kachkaev on 3 March 2012 at 18:37

chriscf, I’ve mentioned this aspect in the post, if you haven’t noticed. Bing in London is dated Oct 2011, and is just a few months old.

Hide this comment

Comment from Wynndale on 4 March 2012 at 11:16

Thank you for pointing out that the lower levels are more up to date. I have got into the habit of using zoom level 21 in London because of the better alignment.

Hide this comment

Comment from Harry Wood on 5 March 2012 at 16:20

Thanks for a well thought out posting on this topic Alex. We had a long chat over burritos about it too, and I talked about the pros and cons. I know you're aware of all ths issues. I'm in two minds about it.

Armchair mapping can be quite "interfering" and discouraging to real mapping (See Armchair Mapping wiki page I recently wrote with some attempt to be balanced)

But we want better, and particularly more even/consistent, building coverage. Tracing from bing is the best (really the only) way to get that data initially. Ideally this would be followed up with a re-check on the ground. Thats what I've done in small patches during mapping parties throughout the summer, and I encouraged others to do the same. I've been quite cautious while proposing a focus on building outlines at mapping parties, and taken the time to explain the plan and try to get buy-in. But it seems like there's not much interest in that style of mapping from other Londoners, with a few exceptions. We've made slow progress particularly in spreading eastwards from Soho towards Mayfair and the Buckingham Palace area you show there.

I sense there may be even less enthusiasm, or maybe outspoken objection to a pure armchair mapping approach. from some. This is why I've stopped short of suggesting this myself. Even now I'm tempted to sit on the fence.

...but I think on balance I would support the plan for a bit of sketching prior to the olympics. In laying out the details though I think we can take a cautious approach. Firstly I think it's important that we do it from-the-centre-working-outwards, always aiming for even consistent coverage. We don't want somebody deciding to work from West London moving Eastwards, but then getting bored. Even just filling in an area which is unconnected to the central area, creates a messy looking result. Secondly we should agree an approach (OpenStreetBugs? fixme nodes?) for labelling spots which need follow up on-the-ground, because I need a way of choosing where we need to go mapping this summer! But I think this will help to highlight that this is not an exercise in dumping low quality map data into an uncaring database. Local mappers will be seeing the data arrive, and this is a positive thing. Armchair mappers should be adding data with their support, and work together to make a better map. We should be quite clear that flagging bugs is an expected part of the armchair mapping activity, and figure out examples of the kind of imagery interpretation which requires follow up. I'd be more comfortable if we also had better approaches for flagging imagery offsets, and imagery which is out of date, but that needs technical solutions.

Hide this comment

Comment from Kachkaev on 5 March 2012 at 16:49

Harry, I agree with the “from-the-centre-working-outwards” approach.

To achieve that, we could try do the mapping in several phases. For instance, we can create a cake with 40-60 slices covering travel zone 1 only + wiki page with instructions and see how the progress goes. If there are enough volunteers to cover that, we'll switch to phase 2 and let people do buildings in Canary Wharf and around the Olympic Park, and so on. Comments to cake slices on mapcraft can help us to tell remote mappers about recent changes we know about to protect people from mistakes. But the approach in general can be close to crisis mappings organised by HOT. I expect enough volunteers if we do all announcements well as the challenge is quite important for OpenStreetMap both in terms of usefulness and popularity. Olympics is a good chance to tell about ourselves to thousands of new people!

OpenStreetBugs is a handy tool here, but to my mind we should pay more attention not on that, but on the actual outline accuracy level in the area. I've looked recently at the buildings in Bloomsbury and places around, and to my unfortunate found that the quality of outlines is very bad in many places, as some things there were mapped before Bing became available (even by us on mapping parties). We simply can't remap everything ourselves by Olympics.

Finding a location for the next mapping party will be always easy - filtering stuff in JOSM is enough to find places with no POIs and addresses. Having pre-traced buildings is the only chance for us to cover 2-3 times more that we usually do.

I would not worry too much about the imagery offsets, as fixing that is very easy. Giving the list with numbers and a small reminder on a wiki page is enough to my mind.

Hide this comment

Comment from Rovastar on 5 March 2012 at 18:08

I am up for some of this hopefully I will be back based on London in a few weeks for a new contract. I was thinking I would do some more effort in London before/for the Olympics.

Oh about "armchair" mapping in general:

Harry, that wiki on Armchair mapping is not balanced at all. It is very anti armchair mapping. IMHO written by an old school mapper that doesn't embrace the modern toools we now have available.

A balanced approach would be adding "advantages" rather than just "disadvantages".
(Here are a few advantages of the top of my head

a) quicker to get coverage. I can get many, many times more coverage mapped doing armchair versus normal/old school mapping.
b) Some things are difficult to do without armchair mapping/aerial imagery. Building outlines, etc.
c) more accurate quality “form” of the roads, etc- e.g. You can get much better curves of roads via armchair mapping. have spent ages changes traffic islands, major roads etc to this from obviously inaccurate. No one maps on foot, car, etc by going down the white lines in the middle of road. With armchair mapping you can do this with ease. (you can always realign, if needed, afterwards)
d) The entry level to participate is lower, you don't need expensive equipment, it is less elitist.
e) most things road, etc don’t change for decades. If roughly know an area and you are mapping/tweaking major roads that you know were there x months ago and have not heard they have changed you don’t need to physically walk/drive down the road to map it as some normal/old school mappers would insist on.

(maybe I should add some disadvantages to normal/old school mapping too)

It works best in a combination of some knowledge of the area rather than totally blind – which is how I try and do my mapping.

I have a write a more balanced view on the wiki later.

Hide this comment

Comment from Former OSM contributor on 6 March 2012 at 19:09

Harry, I've just read the condescending wiki regarding 'armchair mapping'. As probably 80% of what I contributed has been done this way I think from your apparent standpoint that I have been wasting my time. Please feel free to revert all of my work. See you all, I'm off for good!

Hide this comment

Comment from Harry Wood on 13 March 2012 at 16:03

"apparent standpoint"? I've tried to reflect a balance of viewpoints and lay out some guidelines to prevent damage being done, and any future antagonism this can cause. If you find the armchair mapping guidelines to be so shocking that you're going to quit OSM, then well that seems like a ludicrous standpoint to me. If you'd prefer to have a level-headed discussion about the guidelines, head over to Talk:Armchair_mapping.

Here Alex is proposing the idea of encouraging/inviting a form of armchair mapping within London data. When people appear so shocked to be asked to do it carefully... that makes me quite uneasy about the idea.

Hide this comment

Comment from Rovastar on 14 March 2012 at 12:47

You might have thought you were being balanced Harry but to be fair it wasn't very balanced - it was from an anti Armchair mapper viewpoint. And overall it still is despite my changes.

I know many people do dislike armchair mapping but an anti armchair mapping standpoint is not helpful overall to the community. Armchair mappers are a large part of the community who are seen as a second class citzen but the anti-armchair mapping movement.

Hide this comment

Leave a comment

Parsed with Markdown

  • Headings

    # Heading
    ## Subheading

  • Unordered list

    * First item
    * Second item

  • Ordered list

    1. First item
    2. Second item

  • Link

    [Text](URL)
  • Image

    ![Alt text](URL)

Login to leave a comment