In the first part I looked the importance of using empirical data to support a local narrative when developing a broadband policy by looking at some of the pitfalls of opinion polls on the state of a communities infrastructure. In this second part I start to look at the kind of data and mapping exercises that can help target interventions.
The first exercise has to be to understand the state of now – what does the broadband landscape look like today? For me there is only one resource for this kind of information – Samknows.
Their data is all from primary resources – BT, Virgin, and the unbundlers – and is carefully cross-referenced against their own checks. Their data forms the backbone of the price comparison sites and their services are the basis of Ofcom broadband quality studies.
Over the last year or so I have done a number of projects with Samknows, mapping and analyzing their data. Each time it has thrown up new local insights and shone new light on the national picture. And each time it has helped to create a middle ground on which a less emotive debate can develop; when faced with the data its no longer really possible to simply blame BT or the regional government, but equally it provides policy makers with an image of the broadband landscape around which its possible to plan.
Such a rich image is only possible with very high quality data. There have been many attempts to model broadband speeds based on paper models of ADSL performance curves from manufacturers and GIS tools that calculate the radial, as the crow flies distance from the BT exchange.
Some of these have tried to build in factors for guessing the true cable length, quality and so on – but at the end of the day they are just increasingly smart guesses – and this becomes very clear when you try to map the result. Estimating broadband speeds based on the radial distance from the exchange will create nice, uniform shapes on a map from which one glance tell you its not a reality.
Mapping quality empirical data provides a more organic image of broadband performance which starts to mould itself to the geography and topography of the area. From this its possible to build a narrative which links the cold data with the tales of broadband woe from communities – and to slay a few myths along the way.
This can perhaps be best demonstrated by a piece of work I recently completed for Oxfordshire County Council. This is my home county and I wanted to see if I could create a map able to highlight some of the broadband stories I knew to be true, and from that to seek out new detail which may have lain slumbering for the reasons highlighted in part 1.
So as well as the nice clean maps which blanket fill a postcode polygon with traffic light colours to represent poor, mean and good broadband speeds , I ran the data through a series of GIS modules to create a contoured heat map of the county – while its harder to say precisely what the speed is at a given location, it does provide a much richer analogue from which the broadband landscape can be described.
Its clear that the underlying data is not based on a simple model and is far from producing the neat conical contours of a radial guesstimate. A quick glance shows how broadband speeds are affected by the contours of hills and valleys, and by man-made features like railways.
From the generalities I wanted to test some established broadband legends. There were long rumours that broadband in parts of central Oxford were slow, and the reasons given seemed perfectly plausible but unproven. The story was always that BT had centralised its various telephone exchanges in the city, which increased the cable lengths, and that in this part of the city the cables had to additionally skirt around the old Morris car plant making them too long to support a good broadband service, even though some of the homes and businesses affected my be just a few hundred metres from the newer telephone exchange.
A glance at the new map clearly shows a glacial valley of poorer broadband to the North East of Oxford’s broadband summit. So while the now BMW car plant is much more compact, the data appears to support tales of the ghost of the city’s industrial past still be haunting of one of the world’s most important knowledge centres.
Once I’d calibrated my eye using a few known problem spots, it became easier to start hunting new broadband stories. Slightly to the north of the city is Oxford Airport. Following the northern perimeter fence the map predicts the existence of an ox-bow lake of poor broadband coverage – perfectly obvious when you think about it. Would you really feel happy allowing Openreach engineers exercising the code powers to dig a trench across the runway?
The granularity of Samknows data again helps to highlights what is a small, localised broadband issue which would be completely missed by lesser datasets.
So now when I work with the good people at Samknows, I still produce the coloured political maps of broadband speed using traffic light colours to highlight potential problem areas but now I include the contoured explorers map which helps to link the data with the local legends.
In the final part of this series, I’ll start to look at some of the other data models which help to add further colour to impact of poor broadband.
