Personal Information Stack
On Saturday morning I awoke in a hotel in Ascot to the worst hangover and heaviest snow of recent memory.
By 1pm leaving the hotel became an inevitability so we checked the train times from Ascot. With conflicting information, stale websites saying one thing and local people saying another we were torn between driving and taking the train. I had a hired car and a determination to get back to Oxford, but also a determination not to get stuck on a dual carriageway somewhere with frostbitten toes.
A34 Snow Chaos by Ash Matadeen
The traffic layer of Google Maps decided it for me. The dark red sections told me something was seriously up with the roads around Oxford so I didn’t want to risk driving onto dual carriageways and getting stuck (which is exactly what happened to many on the A34 for about 8 hours).
People on my Twitter stream said it was chucking down with snow in Oxford, how many centimetres/inches, how close to snowpocalypse we were, marks out of ten. The travel news, having been reconstituted into 140-character updates and localised by county/motorway was appearing in my Twitter stream and told me that things were descending into chaos on the A34, M40 and local roads around Oxford. As if to verify this, Dale Lane’s bLADE traffic app on my Android phone calculated my route and showed it beset on all sides by cold water in varying forms.
National Rail Enquiries’ phone service just about managed to get us enough information to discover we could catch a train from Reading, assuming the roads were clear enough to get there from Ascot, and off we went.
We parked at Reading and got to Oxford on what appears to have been the last train into town before they all started to get cancelled. I felt really well informed the whole time, but my interface to this information was rarely via the official outlets. The times I did put the radio on all I got was reports about football and moaning about gritters and quickly turned it off. So it was a surprise to me to discover I’d been using my own personal information stack.
Personal Snow/Road Information Stack - Conceptual Model
[Before I explain this it’s worth pointing out that it really represents the mobile-accessible information. There are many flash-based bloaty frontends I could use to show this information, but I’m on the move, I don’t want to need a laptop on my knees to show some godawful flash mapping zoom thing when it could have been funnelled through one mobile-friendly user-interface had someone produced the information in a re-usable format. Rant over.]
At the top is the Google Android operating system. This could be your iPhone or any other smartphone. The only requirement is the device’s ability to work out where it is. Some of the software is specific to Android, but likely to be available for your iPhone or whatever.
On the next level, I use Twitter to see people I follow, who are often in the Oxford area, so it becomes implicitly filtered by home location. Also, doing a full search for #uksnow tweets give me an idea of where snow is occuring in postcode clusters. I was able to filter that by my current location to get an idea of my progress from Ascot to Reading.
The excellent bLADE UK Traffic app from Dale Lane allows me to create a route from any location (including ‘here’) and then checks traffic along that route by extracting it from the BBC Travel News feed.
As you can see, the Highways Agency only really succeed as backend information providers, providing a really poor link to their data on the front end. What they are critical to, however, is the data collection & verification. They have an operations centre which gathers information from many sources, including sensor data, camera information, and of course everybody’s favourite fake police cars: the Highways Agency Traffic Officers (HATOs). They provide feeds of varying levels of complexity, and it’s possible the local radio phone-in reports get fed into their system. As the tax-funded central operation they should be the authoritative source. But that’s a bit 1990s.
Google Maps telling me I should avoid the A417 near Hot Air Balloon Roundabout tonight
The final section, almost autonomous from the rest (except bLADE’s use of Gmaps routing API), is Google Maps (Mobile) Traffic Layer. Google collect traffic flow data, but are notoriously cagey about giving away operational details. There are three likely sources. First there’s Google Latitude, in which you personally volunteer your location data to Google so you can share it. Then there’s the Android Location API used in apps for things like “find me the weather”, “tweet with location”, which uses GPS, mobile cell ID or wifi to calculate its position. Finally, there’s the mobile phone providers (O2, Orange, Vodafone, T-Mobile) who aggregate the cell-switching information of phones travelling between cells along a road and sell that information to people - such as Google. All of these methods can be used to infer movement, and locked to a common path (a primary road) they can be assumed to be traffic flow. Unless they run parallel with Eurostar, that is.
Coast-Bound Eurostar in the Snow by jf01350 (CC-by-nc-nd licence)
At first I felt publishing my methods was just showing off, but it’s helped me articulate the levels at which I use this information and how it could be made easier. What this boils down to is a stack of information flowing from people/sensors up to my device, being aggregated in some cases, and filtered by location where possible. Surprisingly it’s nearly all crowdsourced, except for the core Highways Agency information.
My conclusion is that crowdsourced manually recorded data (eg. Twitter, Radio phone-in) is a wonderful hyper-local data source, giving you glimpses into the real situation on the ground, but not easily verifiable and very rarely updated when the incident clears. Meanwhile, the Highways Agency collect volumes of information but must try a bit harder to reach the top of this stack. They have to make themselves more readily re-usable by people who are crying out for an authoritative data-source for their applications & websites beyond a few snippets of incident reports.
I’m sure the future of us paying Trafficmaster to build flow sensors, or relying on Google Phone concentrations to give us symptoms of congestion is not one that the Highways Agency had in mind when they dug up half the network to install sensors and matrix signs. They need to produce feeds of their sensors for free or low-cost public consumption. What do they have to lose?
I feel a Freedom of Information Request coming on…
!speedcam by Ade Bradshaw (CC by-nc-sa licence)
