Please use the comment form under each blog entry to give us feedback items on blog items.

If your comment is not related to a blog entry, please use the general feedback form.

CycleStreets blog

News from CycleStreets

Archive for April, 2010

Cycling Scotland

Wednesday, April 28th, 2010

We're really pleased to see that Cycling Scotland are now linking to Scotland CycleStreets from their front page!

It was the Scottish Government’s Sustainable Transport section who funded the initial development of Edinburgh CycleStreets (in the form of a small seed grant arranged by Chris Hill of Changing Pace).

We'd like to take this opportunity to thank Cycling Scotland and Changing Pace again for their support.

We are particularly proud to be the first (and, as far as we are aware, the only) cycle journey planning system to take hills into account – something we know our Scottish friends are particularly pleased to see about!

Update: We're informed that Dundee has a route planner that includes hills – see the comments!

Front page of Cycling Scotland's website

Current priorities

Monday, April 12th, 2010

At present, we've a couple of areas of work that we've been prioritising, to address known weaknesses with the system.

Firstly, we've been working to reduce the 'wigglyness' of some routes. The problem is that the journey planner engine does not yet take account of turn delays. Once this is finally rolled out, routes will be straighter and easier to remember.

Secondly, and following our recent Developer Day, it was very clear that we need to rewrite the core network traversal algorithm in a more efficient programming language, and we've started some exploratory work on this. The effect of this will ultimately be to speed up the route planning considerably, and to enable us to add long-awaited features like draggable way-points. It will also mean that we can tackle more quickly problems where the weighting of the data is wrong – e.g. how the engine sometimes gives busy roads when that is potentially avoidable. These are all things which we will address before moving out of 'beta'.

Thirdly, we're keen to address user interface issues, particularly increasing the map size and de-cluttering some screens such as the itinerary result page. If you are a user interface designer who could help us with this, please do get in touch!

Lastly, we're working hard to get more funding. The project so far has been achieved on under £12,000, and our two main developers have worked entirely unpaid so far. Our main target is to get 1 full-time plus two half-time funded developer positions for 12-18 months, so that the project can then be self-sustaining. This will enable us to enable us to ramp up routing quality much more quickly – including adding more attributes in – and to implement new, innovative features faster.

An interesting Freedom of Information request

Thursday, April 8th, 2010

Cambridge resident and democracy activist, Richard Taylor has written today about a Freedom of Information request he made recently on the subject of the government’s efforts to create a Cycle Journey Planner, available through the Transport Direct portal. For the avoidance of doubt, Richard is not involved with CycleStreets, though our database shows he has a user account.

We cannot let the outcome of his FoI request go unmentioned, as it overlaps with some of our own interests in open data and cycling community issues, and his blog post makes comparisons with our own system.

Key findings

In his blog posting, he outlines the key findings:

“The response to my [Richard Taylor's] freedom of information request revealed that the “Find a cycle route” feature on the Transport Direct website will have cost us [taxpayers] £2,383,739 (£2.4 million pounds) by the end the current financial year, and there are plans to spend a further £400,000 on adapting what has been produced to provide route planner for a Cycling for Schools programme. The figures given in the response can be used to calculate that each cycle route planned using the government website has cost the taxpayer about £57.”

and in a comment restates the table of figures for the cycling part of Transport Direct:

Financial Year Design Feasibility Software Development Software Licences
Data management Data collection Project management
07/08 £91,595 £16,156 - - -
08/09 - £334,115 £62,930 £13,850 -
09/10 - £17,608 £79,323 £184,671 £309,000 £133,000
10/11 - - £82,491 £293,000 £548,000 £115,000
Total £91,595 £367,879 £224,744 £594,521 £857,000 £248,000

Figures are Ex. VAT
Source (PDF).

The relevant documents are worth reading. They outline the request that Richard made, followed by the response from the Department for Transport. It is a credit to the DfT that these detailed documents seem to have been issued in response to the Request without sign of any resistance.

Interestingly, Richard Taylor has also found another figure covering Transport Direct more widely, when he writes:

“The £2.4 million spent on the government cycle route planner is just part of a much bigger “Transport Direct” project which cost £55 million between 2003 and 2007.”


As we wrote following Google’s announcement of cycle routing in some US cities last month, dealing with the question of competition between different systems:

“The simple answer is that more availability of bike routing is a good thing [for cycling]. There is undoubtedly space for CycleStreets, Google and others to co-exist: each will have their own benefits and niches, and competition means that everyone ends up with a better product.”

We hold the same view with respect to any system, including government-based ones. We have no complaint about the existence of competition.

We also support the principle of multi-modal journey planning. The idea that someone asking for directions could see cycling amongst options as part of their overall journey, e.g. from a house in Cambridge to a London train station, that includes cycling to the station, is a good thing for making cycling the mainstream activity it is and should more widely be. Multi-modal planning could of course be achieved through an API approach also (though we have never approached Transport Direct on that suggestion).

The issues as we see them

The issues raised in the FoI request by Richard Taylor, as we interpret it, are however different ones to that of competition.

1. The question of spending public funds on systems which compete with commercial and community-based systems that are achieved at much smaller cost to the taxpayer.

We are taxpayers ourselves, and we have a valid interest in ensuring that our taxation is spent efficiently. It is perfectly valid for government to be running services of various kinds if they are useful and provide value for money. Our only comment here is to present the facts and leave others to make their own judgements.

Figures for the government system are stated above in the quotation from Richard Taylor’s blog above, and total £2,383,739 by the end of the 2010/11 financial year. In terms of usage, according to their own papers, “Usage of the [Transport Direct] planner remains very low” (Programme paper, 12/Feb/2010), at 23,608 journey plans.

Our own project has been achieved at a cost of under £12,000; however, time spent has been completely unpaid (which is not sustainable; we would like to be able to pay 1.5-2.5 salaries for the coming 12-18 month period and are seeking funding to enable CycleStreets (which is set up as a non-profit-making entity) then to be self-sustaining). Some 76,107 journeys have been planned in what is still our beta phase – during which we are deliberately not promoting the system heavily while known problems are ironed out. An additional 47,000 journeys were planned on the previous Cambridge-only system.

2. The involvement of communities within government-led initiatives.

In our view, the success of any cycle routing system will in large part be dependent on the level of involvement it harnesses from the cycling community. The latter is something we are certainly working hard on, and it is an integral part of our system. Much of the data in OpenStreetMap data has been surveyed by cyclists, and cyclist involvement remains key throughout many aspects of our project, though we know that we always have more to do. CycleStreets would have been impossible without cyclist involvement.

It was originally intended that the cycling community would be one of three sources of data for the Transport Direct project (the other two being Ordnance Survey and newly-commissioned surveying to fill in the gaps in OS data). We know this because were early advisors to the Transport Direct project at a much earlier stage of their project, and attended a meeting where they decided this. We were invited to attend because, having created one of the very first ever online cycle journey planning systems in the world, in the form of a precursor Cambridge-only system, we were recognised as having expertise. Sadly, it is clear from one of the documents released in the FoI publication that the cyclist involvement aspect has been “deferred indefinitely” (Programme minutes, 17/Feb/2010).

3. The Open Data debate: availability of data which is collected at taxpayer expense.

If we have any fundamental issues with regards to the Transport Direct project, it is this issue.

As is clear from our own views on freeing of Ordnance Survey data, we are strong advocates of open data approaches. As individuals, we believe in the fundamental principle that data collected at taxpayer expense should be freely available in return. Doing so means that anyone who wishes to use such data can operate on a level playing field, creating competition and innovation.

This is an issue which we raised last summer with a contact we have at Cycling England (who to some extent are constrained by national policies on Ordnance Survey data, something which is at long last starting to change).

Government spending tax money to create a dataset usable only with a license payment, but which is competing with an open dataset is a somewhat unsatisfactory state of affairs in public policy terms. In this regard, we do feel it is unfortunate that Cycling England authorised and paid for the surveying of (for instance) Cambridge cycle-related data surely knowing that such surveying had already been done to an exceptionally high standard by OpenStreetMap volunteers, although we acknowledge this is a necessary corollary of competition by a system based on Ordnance Survey data rather than open/varied data sources.

Open data is clearly not well understood within government, but government will have to change. (We refer to a non-partisan meaning here!)

Concluding, these three themes do not change the fact that CycleStreets has to compete on its own merits, as it should. We are proud of our own product. We do not claim it is perfect, and the deficiencies we know about (principally routing which sometimes is either over-wiggly, over-busy, or slow) are things we are working hard to solve, despite meagre resources to do so.

Cycling England’s other initiatives

Both of the principal developers of CycleStreets are fans and active advocates of the work that Cycling England more generally is doing, as we know through our roles elsewhere as campaigners that it is achieving results.

Cycling England’s flagship Cycling Demonstration Town and Bikeability programmes are excellent. By raising standards and aspirations, these have achieved enormous benefits for cycling, at relatively low cost compared to many other government programmes. It is extremely important that these successful initiatives continue to be funded – and indeed that funding for them is increased - following the general election.

We should acknowledge that Cycling England is amongst a range of bodies to whom we are finalising approaches for core funding for CycleStreets, so that our project can then run on a self-financing, stable footing. Our remits of promoting cycling are clearly well-aligned. Naturally we will report back on any successful funding outcomes, as we have done with our other smaller grant applications.

“How do I add a bike shop to the map?”

Monday, April 5th, 2010

(… or indeed, any other point of interest!)

We’ve just had a feedback posting asking how a bike shop can be added to the map. Anyone can add data to OpenStreetMap. Having just replied to the user, I thought it would be useful to turn the reply into a blog entry.

If you have an iPhone, the simplest way to add locations like bike shops is to use the MapZen POI Collector app.

Alternatively, you can have a go adding it yourself via a web browser! Just follow these steps:

1. Go to the OpenStreetMap website.

2. Create a free account (top-right).

3. Scroll the map to the place where the bike shop is, and click on ‘Edit’ near the top. Use ‘Edit with save’, which means you won’t accidentally cause a problem.

4. Double-click the point where the shop is, and a green dot will show.

5. Click on the + symbol in the bottom right

6. Type in ‘shop’ and ‘bicycle’ in the bottom-left boxes.

7. Then click on ‘save’ and it will be done.

It will then be shown on the map within CycleStreets in a week or so, when OpenCycleMap (whose map pictures we use) imports the new locations.

A fuller guide to editing, including videos, is available.

Ordnance Survey data freed (partially)

Monday, April 5th, 2010

Easter came early this year!

In the last few days there has been a blaze of activity around the long-awaited changes by the Government to the way the Ordnance Survey’s data is controlled. A significant number of its excellent datasets have been freed up, which we think is excellent news for the UK geographical community.

The principle at stake is whether collected at public expense should be available to the public (including businesses) free of charge. This is a principle increasingly being accepted in the UK, through the launch of the excellent and London Datastore portals, and throughout the US where public data has long been freely available.

This change – strongly welcomed by many in the web mapping community – follows years of pressure by the Guardian’s Free Our Data campaign, and by many others who, like us agree that public data should be public.

The government announced the outcome of its consultation, to which we also responded, and the day after, some of the Ordnance Survey’s datasets were freed up, so they can now be freely downloaded. (Other more detailed datasets, often for more specialised uses, remain restricted.)

Creating value through open data

Although Ordnance Survey data collection/management is partly also paid for by license fees (many of which come from Local Authorities, i.e. from the government anyway!), ‘the Cambridge Study’ effectively argued that freeing up such data and paying the remaining costs through taxation would lead to increased taxation from businesses who are able to create innovative uses of this data. The quality of Ordnance Survey’s data has a reputation as being amongst the best in the world, and it is important that government funding remains so that quality remains high.

Open data facilitating community-based innovation

Furthermore, it has also been argued by some (ourselves included) that making data open means that private/community-sector groups can more cheaply and quickly create products that traditionally the government would spend millions creating in a ‘top-down’ fashion. As community-based social entrepreneurs, we believe that community-led projects are more likely to work most effectively.

An example of this is our own project, CycleStreets, which has been achieved at a fraction of the cost of the Government’s Transport Direct project that is competing with us and whose costs, usage levels, and restrictive data access terms have just been revealed as a result of a Freedom of Information request.

‘Derived data’ issues

One issue which many respondents to the government’s consultation raised is the problem of ‘derived data’. This is the way that the Ordnance Survey’s license believes that placing your own data over one of their non-free datasets, e.g. overlaying a set of crime points, gives them rights over your data. (This is a different issue to wholescale tracing of non-free maps, which is indeed copying.) The Ordnance Survey’s rules effectively create lock-in, because having overlaid data over a dataset, it then becomes non-portable to use elsewhere.

This kind of restriction is unacceptable in a world where web ‘mash-ups’ are widespread and where there is ever-increasing competition. Issues relating to derived data have yet to be tackled, but the consultation response acknowledges that this needs to be tackled.

What happens now?

We moved our own postcode finder to the OS Code-Point database within hours of it being released, and we believe we are the first public user of OS Free data (albeit only for postcodes).

Some of the new OS street data may well make its way into OpenStreetMap (whose data CycleStreets uses). The new data will surely help OSM achieve completion of UK street data more quickly, as there are still areas of the UK where OSM data is not sufficiently complete, though that is changing rapidly. CycleStreets currently suffers considerably from the lack of even basic data in some areas still.

We have no plans to move to Ordnance Survey mapping data, because the OpenStreetMap data is often richer than the data that has been released, and OSM has the benefits of being community-based. Crucial amongst this is ability for the cycling community to update and augment the data themselves.


We think there are several bodies who have been instrumental in getting the OS’ data release terms changed:

Ramping up planning speed

Friday, April 2nd, 2010

We’ve rolled out an improvement to the journey planning engine which should mean it returns the route solutions about a third more quickly than before.

It’s still not as fast as we would like, but we have plans to make a major change this coming month or so which we believe will make a much bigger difference. With people starting to use our API, we’re even more aware of the need to maximise the planner’s performance, both in terms of speed and route quality.

The current speed-up is due to a set of changes we’ve made to the ‘scoping’ of the map data. When start/end points are determined, an elliptical scoping is done to obtain the likely area in which a set of routes will be found. We’ve implemented a more intelligent way of dealing with the scoping, resulting in a notable reduction of the amount of data taken in – leading to faster route-planning and less memory usage.

Our next improvement will be rolling out the wigglyness reduction work we’ve been doing, which is – for the second time(!) – nearing completion. We’ll blog about that soon.

Here is a planned route (number 121095), with the red (fastest) route highlighted*. Following this is a diagram showing the elliptical scope used – basically what’s within the green oval is the data available to the router.

*Yes, we know we need to make these maps bigger! – User interface designer needed – can you help?

The planned route:    and …

… the elliptical scope used for it: