GoogleIO Map Design Talk Response

Below you'll see a session on map design (called 'master class in map styling') featured at Google IO this year discussing Google Maps.



I think they make some excellent points and explain how to put their ideas into practise code via js code examples in Google maps.  In the introductory section they discuss the map design process as being one where you;

• Think about what data to remove
• Refine the data that's left (e.g. adding selective emphasis)
• Test the map for multiple zooms

which I think is very sensible.  I've blogged about the value of removing data myself.

These are the points they make in the rest of the talk that I particularly liked;

• Lowest Zoom first design the lowest zoom first then edit other levels afterwards by stripping out data as you zoom out higher.
• Roads as Landmarks its handy to leave roads in for orientation (at 27:00).  
• Width Editing The ability to edit the width of elements is a new API feature and they show how it can be used effectively at 27:25.

However, I differ with their opinions on a couple of points:

• Ocean = White  At 25:54 they change the color of the ocean and harbour water from blue to white saying this has less visual impact.  A desaturated blue is better IMHO because it blends into the background better than white and people naturally associate blue with water.
• Parks = Gray  Similarly at 26:01 they change the color of parks from green to gray.  Again, people naturally associate green with a park and a desaturated green would fade into the background sufficiently while being easily interpreted IMHO.

Data Density and Keys What I think they could have usefully added to their talk beyond the 3 major points they use above is discussion about coping with data density, i.e. the difference between a map mashup showing a few points and a map with hundreds of points.  At 28:46 they show a map mashup with some circles representing who likes cats or dogs at certain locations in San Francisco and at 29:00 they go on to dismiss the idea of using animal icons such as the ones in the image below (showing eagles and wolves from this post discussing icons) as being 'cheesy'.

They are quite right that A cat/dog icon could be naff but map icons have an inherent advantage over simple circles: you can work out what they mean without looking at a key.  You need extra information to work out that their red circle symbol = number of dog lovers.

That's not to say that simple icons like circles shouldn't be used, in fact they're very useful when there are hundreds of data points to plot (see above showing the location of Boris bike stations, further detail here), lots of icons on a map leads to the 'flock of sheep' problem (see below, the same Boris bike example)

So its worth using simplified icons when you need to show many data points even though they need a key to provide interpretation.  

Animated Maps: The Way Forward?

I didn't make it to Google IO this year but have got a lot out of watching the Google maps talk videos.  One announcement they made was a new type of map element; 'symbols'. These enable developers to animate Google maps more.  Ed Parsons has bigged up the possibilities and I agree with him that animated maps have great potential and that there should be more use of animated maps.  At the moment they are nowhere near as common as simple map mashups.  I've been researching map animations recently so today I'll share some thoughts.

The potential of animation in Maps:   Any graphic or text needs to perform a number of functions:

1. Catch the eye of the user
2. Lead the eye around the graphic intuitively 
3. Communicate effectively
4. Produce a memorable and/or emotional response if possible

We can argue about how good this list is but I think its generally true (I've paraphrased it  from a book on graphic design 'White Space is not your Enemy').  

Movement always catches the eye (think blinking cursor) so map animations always do well at [1].  They can also be extremely effective at communicating effectively [3], the gap minder animation below is geographical information although strictly it isn't a map.  IMHO its very effective at communicating the story of how most countries have developed considerably in the 20th century and its also easy to pick out individual countries from the animation.

I also like this animation of the spread of post offices in the US (although its probably better watched at a higher resolution than I have here)

Downsides to Animation in Maps. However, there are a number of problems with animations.  

- Going, Going, Gone:  Unlike a static graphic, the information is only visible for a fixed amount of time so the user has to see the information, make sense of it and remember it before the animation moves on.  With a well designed graphic, such as Edward Tufte's diagram showing the development of storm clouds, all the information is available for review and elements are constantly visible.

one of the nice aspects of the post office map above is that it keeps the dots on screen after they've appeared which mitigates this problem somewhat.

- Pat Head, Rub Belly:  The kids game of trying to pat your head and rub your belly at the same time illustrates another problem known as split attention - we find it difficult to do or follow multiple things happening at once.  An example is the map based animation at 55 seconds in this clip showing rainfall and taxi location at once.

Solutions:  There are a few solutions to these problems:

- Simplify Icons/other symbology:  Animated maps need to be simplified as much as possible, an easy examples include making icons plain shapes such as rather than a more meaningful bike shaped icon.  The background map on which the data is being animated can also be simplified e.g. removing colored roads.  The post office map above is a good example of both of these things.

- Play Slowly:  In complex parts, slow down the speed the animation is playing at.  This needs to be done with clear signalling otherwise they could miss that it was happening. 

- Annotate:  If something interesting is happening on a screen full of changing data (the split attention problem) a labelled arrow drawing the user's eye to the correct location on screen can be very helpful.

Conclusion:  Map animations can be very powerful but they need to be implemented carefully with a user centered focus.  See this post for more detail and Mark Harrower's discussion

Color Blindness and Color Blends

I came across these color blends in a post on the web today (to save blushes, I'm not linking) from a mapping developer.  Here's what they look like to the majority of us

and here's how 6% of the male population see it (via the excellent color oracle freeware)

clearly the bottom two blends will be very difficult for color blind users to interpret.

Color blindness is a lot more complex than I've explained here but generally blends from yellow to blue are much better than reds to greens.  More info on color blindness in maps here.

If you're interested in the more general uses of color in maps, the Haklay text book has a good couple of chapters.

Usability Problems with Zeros on Cluster Maps

Please excuse the lack of post recently, I've been particularly busy with teaching over the last month.

Some Experimental Results: My PhD student Craig has some results from a usability experiment on the UK Crime Maps site.  It's too early in the analysis to discuss in detail but one aspect is coming across very clearly: users are getting confused by the representation of postcodes with no crimes on cluster maps (explanation of cluster maps).

Problem:  When you zoom in on an area in the crime map, you are presented with a view as on the left.  The number of crimes in the last month is shown for each postcode as a number in a blob, e.g. for the postcode mid left (above Thurleigh Rd) there were 4 crimes.  The problem with this visualisation is that the boundaries of the postcodes aren't visible.  If you look on the right I've illustrated roughly where two of the postcodes are with dotted red lines.  The postcode on the left has 1 crime, the one on the right (Devereux Rd) has none and so hasn't got a blob at all.

When users were asked to pick out the postcode with the least crime on the actual crime map (left) they failed to understand that no blob = no crime and identified the blob with the lowest number as the postcode with the lowest crime.

Solution:  On collated point maps zero crime postcodes should be represented by blobs with zeros in them.  

Basemaps and Learning Curves

I'm up against the wall in terms of time still so short blog post today on a couple of topics both via James Fee.

Subtle Basemaps:  one of the basic problems with mashups has been putting data over a map designed for route planning.  Recently with Google maps API v3 and tools like tile mill people have started de-emphacizing background maps so that the data stands out better against the background map.  I agree with James that the OSM bright minimal style is a interesting development in this line of work.

I'd love to see some user test results to see how people fared working with a mashup based on this base map.

Learning Curves:  Also via James I read a great quote about teaching OS mapping software to students.

"One of the trepidations I have with teaching mapping courses using open source is that it usually requires some modicum of programming which is always way beyond the scope of any beginner class about making maps.  In addition, open source tends to favor linux or unix based tool chains that require config/make/make install tap dance before starting anything. This is akin to telling folks that they will need to forge their tools before they can start building a dog house."

Amen to that (emphasis mine).  I'm teaching web cartography at the moment mostly using Google Earth because I want them to learn about color, symbology and chart junk, not some abstract javascript that they will forget a week after the course is over.

The rest of the post goes on to explain why Sophia thinks TileMill for windows is a game changer.

Grids are good

Excuse the use of map inserts rather than screenshots, I haven't got much time at the moment.

Look at an old Ordnance survey map of the uk and you'll see it's covered by a thin, blue lined grid.  On recent map systems such as Google Maps and OSM there is no grid.  My discussion today argues that we should bring the grids back.

Grids No Longer Necessary?  To locate yourself on an old map such as OS a grid allowed you to use grid references.  Web maps produce points at the right place without such a grid so because a grid adds to the visual clutter of a map, we're best rid of them?

Nope.

Landmarks and Scale:  A grid is helpful because it gives a sense of scale to the map.  If you're looking at part of the antarctic


View Larger Map

 you have no sense of the size of what you're looking at.  Fix a grid of 1km spacing over the top and you can immediately judge distances.

Point Landmarks.  To locate yourself on a map your brain looks for known features and fixes your position in relation to this point.  Here in London that means that Londoners often describe an area in relation to underground (=metro) stations because that's how most of us get around.  The problem with this is that in tests it's found that people tend to position points that they remember by a landmark too close to the landmark.  So if my house was 1 mile away from Clapham South underground station if I was tested I'd tend to draw it closer, say 3/4 of a mile.

Line Landmarks Another way we locate points is in relation to line features such as a major road or coastline.  This has problems too.  Consider the question, which is further west of these UK cities: Glasgow or Cardiff?


View Larger Map

 Glasgow is towards top of map in Scotland.  Cardiff is due west of London and Bristol, you may have to zoom out or in a little 

Without looking at a map most Brits think that Cardiff is further west.  That's because they have a mental map in which the East coast of the UK runs north south whereas it actually runs NNW-SSE and they warp space to accomodate this misconception.

Grid Solution Because a grid runs exactly north-south and east-west it gets over the Glasgow-Cardiff problem.  Also, it minimises the underground station problem as it represents a continuous set of landmarks rather than discrete points (although this point is worthy of more testing to prove this is the case).  So a grid can deliver both a sense of scale and a landmark system which allows people to fix positions on a map and remember their positions correctly.

When to use Grids: I suspect that its more useful to use a grid in an area without a strong set of landmarks (desert, tundra) rather than somewhere with good landmarks, if you put a grid over US cities with their right angled street systems I imagine users would ignore the map grid because the street system provides such a good system of landmarks already.

Google, Bing and OSM Maps: Train Station Symbolization

As part of my research for last weeks blog looking at how the three maps in the title handled symbolization with zooming, I had a look at some other locations in London.  The one that stuck out for me in terms of usability was Clapham Junction Train station at a low altitude zoom level.  This is a local station of mine and also happens to be the busiest station in the UK with over 100 trains an hour.  With so many passengers it makes sense that it should be mapped well.

The User's Needs when Zoomed In:  What is a general user looking for if they access a web map of this station?   If they are zoomed in close then chances are they are searching to answer these questions:

1. Where is a station entrance with temporary parking? - I want to drop off someone by car.*
2. Where is a station entrance handy for me?  - I'm getting to the station by bus/bike/foot?* 
3. Where do I find food/drink/toilets/cash machine?

If they are changing trains at Clapham they will probably follow the station signs rather than looking at a web map so 'where is the platform I need?' doesn't get onto the list. 

Entrances and Exits are Key: So from a user's point of view I think the primary information that a map of this scale should show are the station entrances and exits.  If the user question is [3] then they still want to know about exits/entrances since often services are ouside the station and they will have to get in and out.  Using the above list it seems to me that the map could also usefully show temporary parking, bus stops, cycle racks, food outlets, toilets and cash machines.  However, the symbols showing these services should be secondary in visual impact to the entrances.

Let's see how the map providers fare in meeting these user needs:

Entrance locations: Before we begin the review we need to understand a little about the entrances to the station.  On the Open Street Map below I have added the four entrances to the station as red squares.  Its also important to know that the Brighton Yard entrance (bottom) is relatively new compared to the others.


Open Street Map of the Station:

First up, Open Street Map is packed with detailed information.  All the entrances are marked except Brighton Yard, however, they are symbolized weakly as dotted red lines and are difficult to pick out.  Services are marked on the map at many locations although I note that food outlets on the platforms and the access bridge are missing (I've now added them myself, because that's the point of Open Street Map).  There is no opportunity to click symbols and access further information.

Whilst Open Street Map has done well in terms of having the information available, my problem with this map is that it shows too much.  Most of the information is there but by packing all the data onto the map it becomes visually complex and the entrances are not emphasized.  A particular problem is the rail tracks: all the lines and sidings are shown but the location of these has no use for the average user as they will be navigating within the station guided by signs within the station.  Also, the dashed line symbolisation just doesn't work when the lines are packed together.

I discussed the 'too much' information problem in a previous post.

Google's Map of the Station:

Google's map is far clearer when compared to the Open Street Map but important information is missing and the symbolisation is confusing.  It shows roads and bus stops clearly but other services aren't well marked.  Entrances to the station are missing.  The multiple rail lines have been reduced to two which show the general directions of the lines leading out from the station - much better symbolisation.

My main problem with this map is that rather than use a polygon and a single marker to show the station (see last weeks discussion), Google has a confusing group of 3 markers.  You would be mistaken for thinking the station was quite small and in three separate locations.  In fact it's one large station that covers most of the above map, as you can see from the Open Street Map.  Click the rail station symbol and you are not linked to anything to do with the station but see information about the nearest buses.  Very odd.


Bing Map of the Station:

Bing's map is similar to Google's in terms of clarity:  Few train lines are shown, roads are clear and easy to see.  Bing also uses a single marker at the main St John's hill entrance which is logical and an improvement on Google's map.  When clicked this station symbol links to a route calculator which is more sensible.  

However, the other entrances are not marked and no services are marked.  At this zoom level I would expect at least some services marked and the lack of entrances is a real issue.

Conclusion:  I've outlined why I think at this scale of map, the key information users need about Clapham Junction station is the location of the entrances.  The maps vary in how many entrances they mark but none of them show all of them with the required level of emphasis IMHO.

Looking at other data that could be usefully marked on the map; Open Street Map has the most detail but it is presented in a visually complex way.  Bing maps goes to the other extreme with a very clear map but showing little useful information.   Google's approach of mixing some service information on a relatively clear map represents the most sensible approach but they have used some very odd symbols in the process.

*Obviously 'entrance' could be replaced by 'exit' in these questions for journeys going in the other direction.  For clarity of discussion I just describe outgoing journeys in this post as return journeys are exactly the same from a map point of view.

Google, Bing and OSM Maps: Zooming Symbolization

How Symbolisation changes with Zooming Out: A couple of weeks ago I sang the praises of Google’s 3D building symbolisation. I said putting 3D building models into Google maps with a faded gray coloring could really help people using their maps to navigate around a city. Such models represent symbolisation: the simplification of reality (3D buildings) into simpler symbols that help understanding. This week I look at how the symbolisation of buildings changes as user’s zoom out from a close in view.  I'm assuming that the user is doing some sort of navigation task again and I compare

• Google maps
• Bing maps
• OpenStreetMap

Change Sucks: The first thing to point out is that evidence from my PhD student Craig’s work on the usability of maps shows that changing symbolisation as users zoom out makes it difficult for users to track what’s going on. So any sybmolisation change, say from a building polygon to a point symbol, has a mental  ‘cost’ to it, symbolisation should only change where there is good reason to do so.

Polygons to Points: So how should building symbolisation change as a user zooms out from an urban landscape? A good best practise is to have buildings represented as polygons at low altitude. As the user zooms out, the polygons become difficult to differentiate visually so it makes sense to change symbolisation to a labelled point. This is exemplified by Google’s rendering of the White house as in the screen video below. (apologies for the quality and varying sizes of the videos below, I'm experimenting with a new screen capture tool and haven't got it all worked out yet)




Points to note:

• The symbolisation starts out as a 3D gray model
• It switches to a polygon on zooming out
• At all times the building is represented by a labelled point. As the user zooms out the polygon disappears and we are left with just the labelled point.

Its pretty faultless, every change in symbolisation is helping the user and it's an intuiative system.

Parliament by Google: So how does the white house example compare with the UK seat of government, the houses of parliament?



Oh dear, nowhere near as good.  A clear glitch that all labels and symbols disappear at altitude.

Parliament by Bing:  How does that compare to Parliament by Bing Maps?  Bing uses a different approach, instead of having one base map that switches symbolisation as you zoom out they have incorporated an old static style road map as the base level which changes at altitude to what amounts to a completely different map.



As you can see, the symbolisation at the lowest level is better than Google's Houses of Parliament but less good (IMHO) than Google's White House.  As you zoom out, the symbolisation stays the same which is easy to follow as a user, but the view becomes very cluttered.  The switch to a completely different map system is clumsy at best.

OpenStreetMap:  In comparison OpenStreetMap does well at low altitude with good labeling (shown here as an image rathe than a video):

Their approach is to have labels without points which then disappear at altitude while the polygons persist.  I think a labelled point at high altitude is the better system.  This alternative technique isn't too bad visually as the polygons are a neutral color and, as we've discussed,  keeping symbols the same helps users understand the map.  However, if the polygons became paled out at high altitude (so they blended in with the background) then they would work better IMHO.  This is because the road network operates as a landmark system at altitude so needs to become visually dominant over the buildings. 


Conclusion:  I wouldn't want to generalize to judging the quality of these three mapping systems on the 'Houses of Parliament' test.  As we've seen with Google, the quality can vary from place to place.  Its also easy to pick holes with mapping systems in this way, there's an enormous amount of work in producing intelligent symbolisation for multiple zoom levels across the entire globe and the visibility of polygons, labels and points at different zoom levels has to be automated in some way.  However, IMHO the symbolization could be improved in all of these three map systems and the building is important: its the seat of government in a G20 country and a major landmark in a world class city.    

3D buildings as Landmarks in Mobile Maps

Maps and City Navigation: Today I’m going to be looking at ways of including buildings as landmarks for user’s trying to navigate an urban landscape.  How important are landmarks for navigation and recall?  its relevant that memory champions who can memorize the sequence of a deck of cards in a minute use spatial landmarks as a system to aid their recall:  The memory palace technique has the user walking through a building well known to them imagining vivid tableaus such as a moonwalking Einstein (the title of an excellent book on memory) happening in different rooms.  The technique taps into our spatial memory system and shows how powerful landmarks can be. 

Google’s 3D buildings as Landmarks:  As an example I really like google's 3D buildings in google maps (see above, bottom screen shot). In a city scape the buildings can carry important landmark information for users trying to orient themselves using a map on a mobile device. A 'you are here' marker from GPS readings can make such orientation irrelevant but cities can cause a GPS trouble because of the 'urban canyon' effect so landmarks are still important.

You could choose to put full color 3D buildings onto a mobile device map as landmarks but this has issues that are useful to consider: 

1. Device processing time: full color models would slow the device down, the data bandwidth needed is very demanding. 
2. Visual processing: full color 3D can be complex for the user to process cognitively, for instance the color and structure of the buildings would make it more difficult to see the road pattern below.
3. Bird’s eye View:  For way finding, the bird’s eye view of a building is often not its most salient feature visually.  It is often the view from the pavement (sidewalk for US readers) like the colorful facade of the building or the grand doorway that provides the memorable image a user can latch onto (see top part of image above which relates to the location shown on the map).  

Google’s gray, plain rendering of buildings in 3D is a good solution to the first two issues but fails on the third.


Interesting 2D Alternative: Another way of including buildings in maps so they can be used as landmarks is to use photos.  A photo of an interesting building from road level at a node (road junctions or other important route points) mimics streetview functionality but pulls out the best landmark features of the view at that node (see top part of the image above which is taken from the streetview image of the map below). When a user trying to navigate a city reached such a junction and had a choice of routes she could locate herself using the building photo.  IMHO this addresses the three problems listed above but unfortunately it introduces another issue:  what makes a good visual landmark requires human input whereas Google’s gray building layer can be produced automatically.

Interesting 3D Alternative: A second alternative is to move into sub 3D view as illustrated above. Selected buildings at nodes are shown, they are rendered in gray at an enlarged scale (from a post by Digital Urban )

If you follow the link to the post you can also see the image placed into a zoomable Google Maps interface but at a 45 degree angle, this has a lot of potential for navigation but it has issues:

• Plan View: A plan view is better for viewing roads
• Single Direction View: You can only see the building from one compass angle direction (North Eastwards in the above image).  Often from ground level it’s a street level feature like a large doorway that is most salient to the viewer.  Unlike features such as the Gherkin (the torpedo building in the above image) a model of a doorway doesn’t make any sense viewed from the wrong direction.  

Usability of 3D maps in Navigation:  Related to our discussion of 3D in navigation, the comments on this post of mine about a 3D map of Southampton University's campus are interesting:  they suggest that 3D may be causing usability problems for those who are not familiar with maps .

Conclusion: Google have to provide a base map for any journey that their journey planner may suggest to a user that works on a mobile device.  Overall, I think their current solution is arguably the best given the need to produce their maps automatically and addressing the three issues I list above.  Where it would be worth trying out the alternatives I've suggested in this post would be a map for a particular route e.g. a map informing people how to get from a particular train station to a museum.

Steve Jobs on Design

I often quote examples from Steve Job's career as examples of good design, and by design I mean usability and a more holistic sense of knowing everything about a product that is important and getting all those things right.  So I was pleased to find this quote from him via Stephen Fry:

“In most people’s vocabularies, design means veneer. It’s interior decorating. It’s the fabric of the curtains and the sofa. But to me, nothing could be further from the meaning of design. Design is the fundamental soul of a man-made creation that ends up expressing itself in successive outer layers of the product or service.” 

Steve Jobs in an Interview with Fortune Magazine, 2000

Point Clustering Usability Example

Intro: A number of times before (1, 2) I’ve discussed point clustering maps. I thought it would be useful to present an actual case study of a mobile app* that makes use of clustering to explain why I think the technique has problems from a usability point of view.  

Herd of Sheep Problem: Clustering maps have appeared as a solution to an old cartographic problem: If you have too many points on screen they overlap each other producing a map that looks like a herd of sheep has wandered over it (particularly if you use white markers as below).

Barclay's bike hire Stations map in Zoomed out view   

The Technique: The clustering solution involves sorting points into groups, each group then gets a maker (usually a circle) which is labelled with the number of points in the group.  This reduces the total number of markers rendered on screen.  As the user zooms in on an area big groups break up into smaller groups.  At a low altitude  the point icons themselves separate out from the groups.  

Boris Bike Example: My case study is the London Cycle App for the iPhone which shows in real time where Boris Bikes (aka 'Barclays Bikes') can be hired from.  Bikes can be picked up from a large number of docking stations and the app is very useful as bike stations often become empty as a typical day progresses.  The app needs to deal with the herd of sheep problem, the screenshot above is actually the bike stations it needs to visualise, it does this using group clustering.  We'll discuss the map's symbolization at high and medium zoom levels and compare that to the needs of an imaginary user who's using the app to find a bike to hire for a trip into town.

High Alitude Zoom:  At this viewpoint we see two cluster circles which show how many bikes are available for hire in two groups (repeated 3 times below).

High Altitude View with different possible group catchments.

The major usability issue here is that the catchment areas that the groups have been created from are not visible.  Any of the dotted catchment areas I've added in the three screenshots above are possible.  This is an issue for our user for two reasons:

• Where are the bikes?  It isn't possible for our user to know precisely where the 230 and 178 bikes are so how can she know where to head to?
• Density:  Knowing the total number of bikes in a group isn't as useful to her as knowing the density of bikes per square km.  For example, if the centre map catchments applied she'd be sensible to head for the 230 group SW of Islington but if the right hand map catchments apply the 178 group may well be a better bet as it has less bikes but in a much smaller area.

Overlap?  Not knowing the catchments of the cluster groups isn't a complete fail, its just unnecessarily vague. However, what is clearly a serious problem is the group circles overlapping.  That's plainly confusing for our imaginary user, it has no meaningful interpretation that I can think of.

Alternatives:  I think that both heat maps and grid density maps (below) are better solutions to the herd of sheep problem in this context as they communicate the data as density of bikes rather than total count.

Alternatives: heat map of house prices (bottom) Grid map of mobile web coverage (top) from this BBC site).  Boundary is of the whole bike scheme.

 (Note that I didn't have the time to create actual maps of the data we're discussing, these are maps of different data sets over London but they illustrate the techniques).  

High Zoom Best Alternative: No Data  My solution to the head of sheep problem is not heat or grid maps and is shown below:

its slightly counter intuitive but I've come to it by thinking about what the user actually needs.  She's probably getting into London by public transport and therefore knows the rough area that she wants to pick up a bike from.  My reasoning is that at this zoom level, the data is actually a visual distraction, what she really wants is to see landmarks so she can zoom in on her chosen area.  The bike availability is best kept off screen until she has zoomed in   You can see I've used three landmark layers above:

• Boundary: A polygon showing the boundary of the area covered by the Boris bike scheme 
• Big Train Stations: The location of a number of mainline train stations.  Not only are these well known landmarks in London but a large number of people wanting to pick up bikes will be arriving here (red pins).
• Bike Stations: The location of the bike stations themselves (blue circles)*

Medium Zoom: As the user zooms in from the High Altitude Zoom view, the two big clusters resolve into smaller groups (below left).

Medium zoom screenshot (left) and suggested alternative (right) created by screen capture from excellent oobr's bike share map

With each further progression down the zoom levels the number of clusters, their size and their location changes.  This change of symbolization proved highly confusing to users.  My PhD student Craig did some user tests on a similar visualization, his observations show that users expected that the circles would persist across zoom levels - some even said they wanted to use them as landmarks and were very put out when they disappeared from the screen!  

Medium Zoom, Improved View (above, right):  At this zoom level, by reducing circle size, its actually possible to present circles representing all the bike stations with color coding showing where bikes are available (red = lots available, blue = few, circle size = number of docking points at each bike station).  Our user can probably locate a station to head for without having to zoom in further, a good usability gain plus she avoids the problems of changing symbolization outlined above.  The alternative map isn't optimized for an iPhone screen but I think you can see my point.

Conclusion:  My suggestions for improvements to the bike app screen shots at high and medium zoom levels involve not using a clustering technique at all. They could be combined as an complete alternative to the app; at high zoom levels the user sees just landmarks, when she zooms on her chosen area the map view switches to the medium zoom visualization illustrating bike availability. This alternative app would;

• Avoid multiple confusing changes of symbolization with zoom (the most important advantage)
• Avoid cluster circle overlaps
• Avoid obscuring the view of the map by big cluster circles at high altitude zoom levels 
• Show landmarks to aid effective zooming in
• Show the user bike location data precisely - the user only sees bike availability at each station, there is no ambiguity created by grouping stations.  
• Present the user with bike availability at the highest zoom level meaning she doesn't need to zoom in unnecessarily.  

I don't think this technique of avoiding point clustering completely would work in all situations but in the case of the Boris Bike app, I think it's the best solution.

Further Testing:  My PhD student Craig is going to be comparing the usability of alternatives such as the heat and grid map to the cluster circles in the future.  So far preliminary results show there are serious issues with point clusters but also that there are issues with the alternatives as well.

*I've got the boundary of the Boris Bike scheme and the station points wrong in some screenshots but I haven't corrected it because I'm short on time at the moment and it doesn't impact the discussion.

Goldilocks Maps: Enough Info but no more

Introduction: There are Goldilocks planets (not to hot, not too cold, just right) and Goldilocks economies (moderate inflation, moderate growth). I want to propose that we aim for producing goldilocks web maps: Not too little information, not too much, but just enough. 

Finding Cafes Problem: A web map example is using satellite imagery or a road map as base data in a mash up.  Satellite imagery has lots of information people find interesting, like who has the biggest patio in their neighbourhood (below left). 

However, if you're on your iPhone trying to find somewhere to eat (cafes in the above map are red pins or dots) you don't want the visual complexity of seeing patios or even gardens.  You want to see a road map mash up (above right), where the only information is the cafe locations, roads and road names.

Related Web Map Design Issues: It's not just base data this applies to, there are a number of web map design topics where  we should be actively thinking 'have I got the information balance right here'.  Examples include:

• 'Graying down' the road map to lower its visual impact (a tool available in Google Maps API v3).
• Making visually simple icons, also discussed in why web 2.0 icons suck 
• Ways to cluster points making the map simpler

The Goldilocks web map concept is related to Edward Tufte's idea of 'chart junk': any pixels not directly showing data or helping show data (e.g. a key) are junk and need to be removed.  

’Mapping London's Subterranean Rivers’ from sandra crisp on Vimeo.

A lovely idea for a map but full of chart junk.

There's more of an argument for putting too much information into a map than allowing it to contain chart junk. Extra information, by definition maybe useful whereas junk is, well, junk.  For example, we could add other data layers to the cafe map by splitting the cafes into vegetarian, greasy spoon and coffee bars.  Its more complex to understand but will be important for vegetarians.

For most maps the concept of chart junk holds true but I recently posted ideas about when chart junk has a place by discussing the flashiness of Google Earth Tours.

London tube map:  This is the perfect example of a Goldilocks map, 

The Modern Tube map is shown on the top and the real station locations are on the bottom.  The Circle Line is in yellow and has been distorted from reality on the tube map to improve clarity.  Other distortions to improve clariry include making lines vertical, horizontal or at 45 degrees .

Harry Beck's genius was to realise that when navigating within the tube system the true locations of the tube stations is information that can be left out, users can navigate perfectly well given just lines and node information.  This means the actual locations and distances on the map can be distorted in order to improve the clarity of the map.  E.g. the area inside the circle line (yellow above) is expanded, it has a high concentration of stations and expanding them makes the station labels easier to read. The idea has been copied for transport systems the world over making it very familiar today but it isn't hard to imagine people's initial reactions:  A map that doesn't show locations properly?  What good is that? Dismissed by the authorities at the time when released it to the public it proved very successful and it is now, rightfully, a design classic.