OS Open

Ordnance Survey have this week released four new additions to their Open Data product suite. The four, which were announced earlier this month, are collectively branded as OS Open and include OS Open Map Local, which, like Vector Map District (VMD), is a vector dataset containing files for various feature types, such as building polygons and railway stations. The resolution of the buildings in particular is much greater than VMD – surprisingly good, in fact. I had expected the data to be similar in resolution to the (rasterised) OS StreetView but it turns out it’s even more detailed than that. The specimen resolution for OS Open Map Local is 1:10000, with suggested uses down to a scale of 1:3000, which is really quite zoomed in. Two new files in OS Open Map Local are “Important Buildings” (universities, hospitals etc) and “Functional Areas” which outline the land containing such important buildings.


Above: Comparing the building polygon detail in the older Vector Map District (top left), previously the largest scale vector building open data from Ordnance Survey, and the brand new OS Open Map Local (top right). The new data is clearly much higher resolution, however one anomaly is that roads going under buildings no longer break the buildings – note the wiggly road in the centre of the top left sample, Malet Place, which runs through the university and under a building, doesn’t appear in full on the right. Two other sources of large-scale building polygons are OS StreetView (bottom left), which is only available as a raster, and OpenStreetMap (bottom right). The OS data is Crown Copyright and Database right OS, 2015. The OSM data is Copyright OSM contributors, 2015.

The other three new products, under the OS Open banner, are OS Open Names, OS Open Rivers and OS Open Roads. The latter two are topological datasets – that is, they are connected node networks, which allow routing to be calculated. OS Open Names is a detailed gazetteer. These latter three products are great as an “official”, “complete” specialised dataset, but they have good equivalents on the OpenStreetMap project. OS Open Map Local is different – it offers spatial data that is generally much higher in accuracy than most building shapes already on OpenStreetMap, including inward facing walls of buildings which are not visible from the street – and so difficult for the amateur mapper to spot. As such, it is a compelling addition to the open data landscape of Great Britain.

The OS also confirmed last week the location for its new Innovation Hub. It is indeed a mile from King’s Cross – specifically, it’s in Clerkenwell, and the hub will be sharing space with the Future Cities Catapult. Conveniently the new space has a presentation space and the May Geomob will be taking place there.

Visit the new oobrien.com Shop
High quality lithographic prints of London data, designed by Oliver O'Brien

Taking the Scenic Route – Quantitatively?

A friend forwarded me this article which discusses this paper by researchers at the Yahoo Labs offices in Barcelona and the University of Turin. The basic idea is that they crowdsourced prettiness of places in central London, via either/or pairs photographs, to build up a field of attractiveness, then adjusted a router based on this map, to divert people along prettier, happier or quieter routes from A to B, comparing them with the shortest pedestrian routes. The data was augmented with Flickr photographs with associated locations and appropriate locations. and The article that featured this paper walked the routes and gives some commentary on the success.

Quantitatively building attractive routes is a great idea and one which is only possible with large amounts of user-submitted data – hence the photos. It reminds me of CycleStreets, whose journey planner, for cyclists, not only picks the quickest route, but adds in a quieter (and “best of both worlds”) alternative. Judging locations by their attractiveness also made me think of the (soon to be retired) ScenicOrNot project from MySociety which covered the whole of the UK, but at a much less fine-grained scale – and without the either/or normalisation.

In the particular example that the paper uses, the routes are calculated from Euston Square Station, which happens to be just around the corner from work here, to the Tate Modern gallery. It’s a little over 2 miles by the fastest route, and the alternatives calculated are only a little longer:
Above: Figure from http://dx.doi.org/10.1145/2631775.2631799

I really like the concept and hope it gets taken further – for more places and more cities. However, I would contend that local knowledge, for now, still wins the day. The scenic route misses out the Millennium Bridge which is surely one of the most scenic spots in all of London with its framed views to St Paul’s Cathedral and the Tate Modern itself. The quiet route does go this way, but the route is far from quiet when you consider the hordes of tourists normally near the cathedral and on the bridge. The pretty route goes down Kingsway which is a pretty ugly, heavily trafficked route, ignoring the nearby Lincoln Inns Fields, which is lovely. I think that the following, manually curated 3.0 mile route wins out as a much more beautiful route than the algorithmically calculated one:


Highlights include:

  • Walking through UCL’s Front Quad, through the university campus
  • Down Malet Street, past the imposing Senate House
  • Walking through the Great Hall of the British Museum
  • Bloomsbury Square garden and Lincoln’s Inn Fields
  • Chancery Lane
  • New Street Square (modern but attractive)
  • The statue of Hodge, Dr Johnson’s Cat
  • Wine Office Court, with the Ye Olde Cheshire Cheese Pub
  • Fleet Street and Ludgate Hill, with the famous view to St Pauls
  • The vista from St Paul’s Cathedral, across the Millennium Bridge to the Tate Modern.

Maps in this article are © Google Maps.

Visit the new oobrien.com Shop
High quality lithographic prints of London data, designed by Oliver O'Brien

Ordnance Survey Open Data – The Next Level of Detail

An encouraging announcement from BIS (the Department for Business, Innovation and Skills) a few days ago regarding future Open Data products from the Ordnance Survey (press release here) – two pieces of good news:

  • The OS will be launching a new, detailed set of vector data as Open Data at the end of this month. They are branding it as OS OpenMap, but it looks a lot like a vector version of OS StreetView, which is already available as a raster. The key additions will be “functional polygons” which show the boundaries of school and hospital sites, and more detailed building outlines. OS Vector Map District, which is part of the existing Open Data release, is already pretty good for building outlines – it forms the core part of DataShine and this print, to name just two pieces of my work that have used the footprints extensively. With OpenMap, potentially both of these could benefit, and we might even get attribute information about building types, which means I could filter out non-residential buildings in DataShine. What we do definitely get is the inclusion of unique building identifiers – potentially this could allow an crowd-sourced building classification exercise if the attribution information isn’t there. OpenMap also includes a detailed and topological (i.e. joined up under the bridges) water network, and an enhanced gazetteer, i.e. placename database.
  • The other announcement relates to the establishment of an innovation hub in London – an incubator for geo-related startups. The OS are being cagey about exactly where it will be, saying just that it will be on the outskirts of the Knowledge Quarter, which is defined as being within a mile of King’s Cross. UCL’s about a mile away. So maybe it will be very close to here? In any case, it will be somewhere near the edge of the green circle on the (Google) map below…

p.s. The Ordnance Survey have also recently rebranded themselves as just “OS”. Like University College London rebranding itself as “UCL” a few years ago, and ESRI calling itself Esri (and pronouncing it like a word), it will be interesting to see if it sticks. OS for me stands for “open source” and is also very close to OSM (OpenStreetMap), so possible confusion may follow. It does however mean a shorter attribution line for when I use OS data in my web maps.

Screen Shot 2015-03-04 at 17.47.52

London’s Knowledge Quarter

Manchester – Languages and Jobs

Many of my visualisations have focused on London – there is an advantage of being in the city and surrounded by the data, which means that London is often the “default” city that I map. However, I’ve created a couple of Manchester versions of my popular maps Ward Words and Ward Work. Logistics and time reasons mean that I present these as images rather than interactive websites, although I used the existing London-centric website as a platform to work with the Manchester data. A bonus is that, by presenting these as images, I can use LSOAs which are more detailed than wards – there are too many of them for my interactive version to be very useable but they work well within a standalone graphic.

I’m only showing the top* result, and the way the categories are grouped can therefore significantly influence what is shown. For example, if I grouped certain categories together, even ones which don’t appear on the map itself, then the grouped category would likely appear in many places because it would more likely be the top result. It would therefore easy to produce a version of this graphic that showed a very different emphasis. (*Strictly, second-top for the languages.)

The maps were created using open, aggregated data (QS204EW and QS606EW) from the ONS which is under the Open Government Licence, and the background map is from HERE maps. Enjoy!

1. Languages second-most commonly spoken in each LSOA in the Greater Manchester area (click for a larger version):
second_languages_manchester N.B. Where the second language is spoken by less than 2% of the population, I simply show it as a grey circle. LSOAs have a typical population of around 1500 so the smallest non-grey circles represent around 30 speakers of that language.

2. It’s important to remember that, except in a single area, English is not represented on the map at all. If you show the primary language (i.e. English) to the same scale, the map looks like this:

3. Here’s the equivalent of the first map, for (most of) London. Note I’ve changed the key colours here. I appreciate that it is difficult to use the key, as there are so many more languages shown, and the variation between the colours is slight – particularly as they are shown translucently on the map:

4. The most popular occupation by (home) LSOA (again, click for a larger version):
manchester_occupation_adornedI’ve used grey here for the “Sales Assistant” occupational group, as this is the dominant occupation in large urban areas.

5. By way of comparison, and at roughly the same scale, here is (all of) London:
My interactive (London only I’m afraid) version is here – change the metric on the top left for other datasets.

Seeing Red: 15 Ways the Boris Bikes of London Could be Better


A big announcement for the “Boris Bikes” today, aka Barclays Cycle Hire. London’s bikeshare system, the second largest in the western world after Paris’s Velib and nearly five years old, will be rebranded as Santander Cycles, and the bikes with have a new, bright red branding – Santander’s corporate colour, and conveniently also London’s most famous colour. As well as the Santander logo, it looks like the “Santa Bikes” will have outlines of London’s icons – the above publicity photo showing the Tower of London and the Orbit, while another includes the Shard and Tower Bridge. A nice touch to remind people these are London’s bikes.

velibIt’s great that London’s system can attract “big” sponsors – £7m a year with the new deal – but another document that I spotted today reveals (on the last page) that, despite the sponsorship, London’s system runs at a large operating loss – this is all the more puzzling because other big bikeshare systems can (almost) cover their operating costs – including Washington DC’s which is both similar to London’s in some ways (a good core density, same bike/dock equipment) and different (coverage into the suburbs, rider incentives); and Paris’s (right), which has a very different funding model, and its own set of advantages (coverage throughout the city) and disadvantages (little incentive to expand/intensify). What are they doing right that London is not?

In financial year 2013/4, London’s bikeshare had operating costs of £24.3m. Over this time period, the maximum number of bikes that were available to hire, according to TfL’s Open Data Portal was 9471, on 26 March 2014. This represents a cost of just over £2500 per bike, for that year alone. If you look at it another way, each bike is typically used three times a day or ~1000 times a year, so that’s about £2.50 a journey, of which, very roughly, the sponsor pays about £0.50, the taxpayer £1 and the user about £1. In those terms it does sound better value but it’s still a surprisingly expensive system.

As operating costs, these don’t include the costs of buying the bikes or building the docking stations. Much of the cost therefore is likely ocurring in two places:

  1. Repairing the bikes – London’s system is wildly* successful, so each bike sees a lot of use every day, and the wear and tear is likely to be considerable. This is not helped by the manufacturers of the bikes going bust a couple of years ago – so there are no “new” ones out there to replace the older ones – New York City, which uses the same bikes, is suffering similar problems. (* Update: To clarify, based on a comment from BorisWatch, this assertion is a qualitative one, based on seeing huge numbers of the bikes in use, in certain places at certain times of the day. Doubtless, some do remain dormant for days.)
  2. Rebalancing/redistribution activity, operating a fleet of vehicles that move bikes around.

I have no great issues with the costs of the bikes – they are a public service and the costs are likely a fraction of the costs of maintaining the other public assets of roads, buses, railway lines – but it is frustrating to see, in the document I referred to earlier, that the main beneficiaries are in fact tourists (the Hyde Park docking stations consistently being the most popular), commuters (the docking stations around Waterloo are always popular on weekdays), and those Londoners lucky enough to live in Zone 1 and certain targeted parts of Zone 2 (south-west and east). Wouldn’t be great if all Londoners benefited from the system?

Here’s 15 ways that London’s bikeshare could be made better for Londoners (and indeed for all) – and maybe cheaper to operate too:

  1. Scrap almost all rebalancing activity. It’s very expensive (trucks, drivers, petrol), and I’m not convinced it is actually helping the system – in fact it might be making it worse. Most cycling flows in London are uni-directional – in to the centre in the morning, back out in the evening – or random (tourist activity). Both of these kinds of flows will, across a day, balance out on their own. Rebalancing disrupts these flows, removing the bikes from where they are needed later in the day (or the following morning) to address a short-term perceived imbalance that might not be real on-the-ground. An empty docking station is not a problem if no one wants to start a journey there. Plus, when the bikes are in sitting in vans, inevitably clogged in traffic, they are of no use to anyone. Revealingly, the distribution drivers went on strike in London a few months ago and basically everything carried on as normal. Some “lightweight” rebalancing, using cycle couriers and trailer, could help with some specific small-scale “pinch points”, or responding to special events such as heavy rainfall or a sporting/music event. New York uses cyclists/trailers to help with the rebalancing.
  2. Have a “guaranteed valet” service instead, like in New York. This operates for a certain number of key docking stations at certain times of the day, and guarantees that someone can start or finish their journey there. London already has this, to a certain extent, at some stations near Waterloo, but it would be good to highlight this more and have it at other key destinations. This “static” supply/demand management would be a much better use of the time of redistribution drivers.
  3. rrrHave “rider rewards“, like in Washington DC. Incentivise users to redistribute the bikes themselves, by allowing a free subsequent day’s credit (or free 60-minute journey extension) for journeys that start at a full docking station and end at an empty one. This would need to be designed with care to ensure “over-rebalancing”, or malicious marking of bikes as broken, was minimised. Everyone values the system in different ways, so some people benefit from a more naturally balanced system and others benefit from lower costs using it.
  4. Have more flexible user rules. Paris’s Velib has an enhanced membership “Passion” that allows free single journeys of up to 45 minutes rather than every 30 minutes. London, like Paris, is a large city, and the current 30 minute cutoff seems short and arbitrary, when considering most bikes are used around three times a day. Increasing the window would therefore have little impact on the overall distribution of the system and might in fact benefit it – because the journeys from the terminal stations to the City or the West End, which are the most distinctive flows seen, are acheived comfortably in under half an hour. In London, you have to wait 5 minutes between hires, but most systems (Paris, Boston, New York) don’t have this “timeout” period. To stop people “guarding” recently returned bikes for additional use, an alternative could be make it a 10 minute timeout but tie it to the specific docking station (or indeed a specific bike) rather than system-wide. Then, if people are prepared to switch bikes or docking stations, they can continue on longer journeys for free.
  5. Adjust performance metrics. TfL (and the sponsors) measure performance of the system in certain ways, such as the time a docking station remains empty at certain times of the day. I’m not sure that these are helpful – surely the principle metric of value (along with customer service resolution) is the number of journeys per time period and/or number of distinct users per time period. If these numbers go down, over a long period, something’s wrong. The performance metrics, as they stand, are perhaps encouraging the unnecessary and possibly harmful rebalancing activity, increasing costs with no actual benefit to the system.
  6. lyonRemove the density rule (one docking station every ~300 metres) except in Zone 1. Having high density in the centre and low density in the suburbs works well for many systems – e.g. Bordeaux, Lyon (above) and Washington DC, because it allows the system to be accessible to a much larger population, without flooding huge areas with expensive stations/bikes. An extreme example, this docking station is several miles from its nearest neighbour, in a US city.
  7. Build a docking station outside EVERY tube station, train station and bus station inside the North/South Circular (roughly, Zones 1-3). Yes, no matter how hilly* the area is, or how little existing cycling culture it has – stop assuming how people use bikes or who uses them! Bikeshare is a “last mile” transport option and it should be thought of as part of someone’s journey across London, and as a life benefit, not as a tourist attraction. The system should also look expand into these areas iteratively rather than having a “big bang” expansion by phases. It’s crazy that most of Hackney and Islington doesn’t have the bikeshare, despite having a very high cycling population. Wouldn’t be great if people without their own bikes could be part of the “cycling cafe culture” strong in these places? For other places that have never had a cycling culture, the addition of a docking station in a prominent space might encourage some there to try cycling for the first time. (*This version of the bikes could be useful.)
  8. Annual membership (currently £90) should be split into peak and off-peak (no journey starts from 6am-10am) memberships, the former increased to £120 and the latter decreased back to £45. Unlike the buses and trains, which are always full peak and pretty busy off-peak too, there is a big peak/offpeak split in demand for the bikes. Commuters get a really good deal, as it stands. Sure, it costs more than buying a very cheap bike, but actually you aren’t buying the use of a bike – you are buying the free servicing of the bike for a year, and free distribution of “your” bike to another part of central London, if you are going out in the evening. Commuters that use the bikes day-in-day-out should pay more. Utility users who use the bike to get to the shops, are the sorts that should be targetted more, with off-peak membership.
  9. officialmapA better online map *cough* of availability. The official map still doesn’t have at-a-glance availability. “Rainbow-board” type indications of availability in certain key areas of London would also be very useful. Weekday use, in particular, follows distinct and regular patterns in places.
  10. Better indication of where the nearest bikes/docks are, if you are at a full/empty docking station, i.e. a map with route indication to several docking stations nearby with availability.
  11. Better static signage of your nearest docking station. I see very few street signs pointing to the local docking station, even though they are hard-built into the ground and so generally are pretty permanent features.
  12. Move more services online, have a smaller help centre. A better view of journeys done (a personal map of journeys would be nice) and the ability to question overpayments/charges online.
  13. hubwayEncourage innovative use of the bikeshare data, via online competitions – e.g. Boston’s Hubway data visualisation competitions have had lots of great entries. These get further groups interested in the system and ways to improve it, and can produce great visuals to allow the operator/owner to demonstrate the reach and power of the system.
  14. Allow use of the system with contactless payment cards, and so integration with travelcards, daily TfL transport price caps etc. The system can’t use Oyster cards because of the need to have an ability to take a “block payment” charge for non-return of the bikes. But with contactless payment, this could be achieved. The cost of upgrading the docking points to take cards would be high, but such docking points are available and in use in many of the newer US systems that use the same technology.
  15. Requirement that all new housing developments above a certain size, in say Zone 1-3 London, including a docking station with at least one docking point per 20 residents and one new bike per 40 residents, either on their site or within 300m of their development boundary. (Update: Euan Mills mentions this is already is the case, within the current area. To clarify, I would like to see this beyond the current area, allowing an organic growth outwards and linking with the sparser tube station sites of point 7.)

London has got much right – it “went big” which is expensive but the only way to have a genuinely successful system that sees tens of thousands of journeys on most days. It also used a high-quality, rugged system that can (now) cope with the usage – again, an expensive option but absolutely necessary for it to work in the long term. It has also made much data available on the system, allowing for interesting research and increasing transparency. But it could be so much better still.

Washington DC’s systems – same technology as London’s, not that much smaller, but profitable.


Various websites I’ve built, and mentioned here on oobrien.com from time to time, are down from Friday at 5pm until Monday noon (all times GMT), due to a major power upgrade for the building that the server is in.

This affects the following websites:

  • DataShine
  • CDRC
  • Bike Share Map
  • Tube Tongues
  • OpenOrienteeringMap (extremely degraded)
  • Some other smaller visualisations

However the following are hosted on different servers and so will remain up:

GeoComputation: A Practical Primer

geocomputationGeoComputation: A Practical Primer, edited by Profs Chris Brunsdon and Alex Singleton, has just been published by SAGE.

The book acts both as a reference guide to the field and as a guide to help you get to know aspects of it. Each chapter includes a worked example with step-by-step instructions.

Each chapter has a different author, and includes topics such as spatial data visualisation with R, agent-based modelling, kernel density estimation, spatial interaction models and the Python Spatial Analysis library, PySAL. With 18 chapters, the book runs to over 300 pages and so has the appropriate depth to cover a diverse, active and fast-evolving field.

I wrote a chapter in the book, on open source GIS. I focused particularly on QGIS, as well as mentioning PostGIS, Leaflet, OpenLayers (2) and other parts of the modern open source “geostack”. My worked example describes how to build a map, in QGIS, of London’s railway “not-spots” – places which are further than a mile from a railway station, using open data map files, mainly from the Ordnance Survey. With the guide, you can create a map like the one below:


That little spot on its own in central-ish London, by the way, is part of Burgess Park, near Peckham.

The book has only just been published and I was able to slip in brand new screenshots (and slightly updated instructions) just before publication, as QGIS 2.6 came out late last year. So, the book is right up to date, and as such now is a great time to get your copy!

It’s available now in paperback on Amazon: Geocomputation: A Practical Primer.

The first part of my chapter:


London Boroughs and Tube Lines


How many of London’s 32 boroughs (& the City of London) would you pass through on a single end-to-end journey on the tube?

It turns out that if you travel the length of the Piccadilly Line (Uxbridge branch), then, in a single journey, you’ll pass through 14 boroughs (and stop at least once in all of them but Barnet). That’s more of London than if you travel on any single Crossrail journey, once it opens in 2018.

Line Branch # Boroughs
with Stops
# Boroughs
Piccadilly to Uxbridge 13 14
Crossrail to Shenfield 10 13
Central to West Ruislip 11 12
Piccadilly to Heathrow 11 12
Central to Ealing Broadway 10 11
Northern High Barnet to Morden 10 10
District Upminster to Richmond/Ealing Broadway 10 10
Overground Richmond to Stratford 8 10
District Wimbledon to Barking 9 9
Hammersmith & City 9 9
Jubilee 9 9
Northern Edgware to Morden via Bank 9 9
Overground Clapham Junction to Stratford 8 9
Northern Edgware to Morden via Charing Cross 8 8
Bakerloo 5 8
Overground West Croydon to Highbury & Islington 7 7
Metropolitan 7 7
Circle 7 7
Victoria 6 7
Overground Clapham Junction to Highbury & Islington 6 7
Overground Gospel Oak to Barking 6 6
Overground Watford Junction to Euston 3 6
District Wimbledon to Edgware Road 5 5
DLR Bank to Lewisham/Woolwich Arsenal 4 4
Tramlink Wimbledon to New Addington 3 3
Waterloo & City 2 3
Cable Car 2 2

See for yourself at http://vis.oobrien.com/tube/#map.

Of course, if you are aiming to see a cross-section of London’s boroughs, in a rush, then the tube probably isn’t the best way, as you’ll be underground for quite a lot of the journey…

Quick-and-Dirty WordPress Site Cloning


Here is a guide to clone a WordPress(.org) blog, on the same server, 10 steps, on Linux, You’ll definitely need admin access to the blog itself, and probably to the database and server too, depending on your setup. I did this recently as I needed a copy of an existing production site, to hack on. If you don’t fancy doing it the quick-and-dirty way, there are, I’m sure, even quicker (and cleaner) ways, by installing plugins.

In the following instructions, substitute X and Y for your existing and new blog, respectively.

0. Do a backup of your current website, like you do normally for an upgrade or archiving, in case anything goes wrong. e.g. under Tools > Export in the WordPress admin interface.

1. Copy all the files:
cp -r /home/~username/www/blog_X /home/~username/www/blog_Y

2. Edit wp-config.php in your new blog directory:

$table_prefix = 'wp_X_';
$table_prefix = 'wp_Y_';

3. Copy all the database tables (prefixed with wp_X_). The new ones should have a prefix wp_Y_ instead. I used the Copy functionality under the Operations tab in phpMyAdmin (see screenshot below).

4. Edit wp_Y_options:
update wp_Y_options set option_name = 'wp_Y_user_role' where option_name = ' wp_X_user_role';

5. Edit wp_Y_options:
Edit the option_value for rows with option_name values of siteurl and home, pointing them to the new location – mine are the same but one might be different, e.g. if you have your WordPress core files in a subdirectory relative to the directory for the site entry-point on the web.

update wp_Y_options set option_value = 'http://your_server.com/~username/wp_Y' where option_name = 'siteurl';
update wp_Y_options set option_value = 'http://your_server.com/~username/wp_Y' where option_name = 'home';

There may be other rows referencing your old blog name, but these are probably from plugins and therefore probably don’t need to be changed.

6. Edit wp_Y_usermeta:
update wp_Y_usermeta set meta_key = replace(meta_key, 'wp_X', 'wp_Y');

(You can edit the affected rows manually, but I had a lot to do – there’s around 5 for each user.)

7. Drop force-upgrade.php in the same directory as wp-config.php and run it from your browser. This rebuilds caches/hashes stored in some of the tables. You can run it repeatedly if necessary, (e.g. if you missed a step above), it shouldn’t do any harm.

You can find force-upgrade.php here.

8. Delete force-upgrade.php. Leaving it is a security risk.

9. Log in to your blog in the new location, as normal. Usernames and passwords should be preserved.


Bad Maps

<rant> Three maps with glaring errors which I came across yesterday. I’m hesitant to criticise – many of my own maps have, I am sure, issues too (i.e. my Electric Tube map, on the right, is deliberately way off.) But I couldn’t resist calling out this trio which I spotted within a few hours of each other.

1. Global Metropolitan Urban Area Footprints


This is, in itself, a great concept. I particularly like that the creator has used the urban extent rather that administrative boundaries, which rarely follow the true urban extent of a city. The glaring error is scale. It looks like the creator traced the boundaries of each city’s urban extent in Google Maps (aerial view) or similar. All well and good, but a quirk of representing a 3D globe on a 2D “slippy” map means that the scale in Google Maps (and OpenStreetMap and other maps projected to “WebMercator”) varies with latitude, at a fixed zoom level. This hasn’t been accounted for in the graphic, with the result that all cities near the equator (i.e. most of the Asian and African ones) are shown on the map smaller relative to the others, while cities near the poles (e.g. London, Paris, Edmonton, Toronto) are shown misleadingly big. This is a problem because the whole point of the graphic is to compare footprints (and populations) of the major cities. In fact, many of those Chinese and African cities are quite a bit bigger relative to, for example, London, than the graphic suggests.

2. Where Do All The Jedi Live?


The map is in the Daily Mirror (and their online new media) so it doesn’t need to be a pinnacle of cartographic excellence – just a device to get a story across.However, Oxford and Mid Sussex – 40% of the datapoints – are shown in the wrong place – both are much closer to London than the map suggests. The author suggests they did this to make the text fit – but there better ways to accommodate text while having the centroid dots in the correct location. It might take a little longer but then it wouldn’t be – quite simply – wrong. I’m somewhat disappointed that the Mirror not only stoops to the level of Fox News in the accuracy of their mapping, but appears to have no problem with maintaining such an error, even when readers point it out. It’s sloppy journalism and a snub to the cartographic trade, that just relocating whole cities for artistic purposes is not an issue, particularly as so many people in the UK have relatively poor spatial literacy and so can be potentially easily manipulated.

3. A London map…


I’m not really sure where to begin here. I’m not sure if any of the features are in fact in the right place!