Keeping up-to-speed

Scotland was an early leader in broadband access, with a programme targeting universal access to 512 kb/s broadband completed in 2009. This was largely successful. In practice, however, some consumers only achieved somewhat lower speeds, and over 2000 have only satellite services, with latencies of around one second making them unsuitable for many interactive applications. More important, the infrastructures established to deliver these services will not scale to keep pace with the growth of broadband speeds worldwide.
The importance of competitive data communications has long been recognised by the european universal service obligation (USO).

The connection provided shall be capable of supporting voice, facsimile and data communications at data rates that are sufficient to permit functional Internet access, taking into account prevailing technologies used by the majority of subscribers and technological feasibility.

Median broadband speeds in the UK are already (in 2010) over 5 Mb/s. As the usage of broadband interaction increases, median speeds will follow Nielsen’s Law, which predicts continued growth over the next decade, by a factor of 8 every 5 years. So we anticipate median speeds in the UK of 40 Mb/s in 2015 and 320 Mb/s in 2020.
A dynamic definition of universal service is implied. The minimum connection rate for universal access must track advances in the information society to ensure continuing digital inclusion. Our recommendation is that Scotland should plan to deliver minimum speeds that are somewhat less than half of these median figures:

2010	2015	2020
2 Mb/s	16 Mb/s	128 Mb/s

These targets should be subject to rolling review. The current Ofcom USO dates from 2005 and cites a figure of 28.8 kb/s. Just updating this by applying Neilsen’s Law would give a figure of 256 kb/s in 2010 and 2 Mb/s in 2015 (which matches the latest Westminster government policy). We recommend that this should be reviewed to take into account prevailing technologies, with a universal service obligation to provide access to the internet at speeds that are not less than one quarter of the median speed delivered to UK subscribers.
These are targets for minimum download speeds for universal access. We expect median speeds to be 2-4 times as fast. In our next post we analyse the infrastructure implications of these targets.
Download speed is not everything – low latency and upload speeds are also important. However, we will find that the infrastructure required to achieve these download speed targets will also support low latencies and symmetric services.
Our aspirations should be compared with those of our international competitors. For example, Finland has a commitment to deliver 100 Mb/s to all by 2015. This places Finland five years ahead of broadband rollout curve. However, current Westminster policy is to lag a further five years behind even our more modest goal.

Functional Internet Access: what speed do you need?

The importance of data communications has long been recognised by the european universal service obligation.

Article 4.2 The connection provided shall be capable of supporting voice, facsimile and data communications at data rates that are sufficient to permit functional Internet access, taking into account prevailing technologies used by the majority of subscribers and technological feasibility.

A dynamic definition of universal service is implied: the minimum connection rate for universal access must take into account advances in prevailing technologies. Universal functional Internet access (FIA) is necessary to ensure continuing digital inclusion in the information society.
Median broadband speeds in the UK are already (in 2010) over 5 Mb/s. As the usage of broadband interaction increases, we expect speeds to continue to grow. Prevailing technologies will follow Nielsen’s Law, over the next decade at least, with speeds doubling every 20 months–increasing by a factor of 8 every 5 years. So we anticipate median speeds of 40 Mb/s in 2015 and 320 Mb/s in 2020.
Ofcom guidelines on FIA, that users should be able to expect connection speeds of at least 28.8 kb/s, were issued in 2003. The most recent review (2005) considered that, "the benchmark minimum speed should remain at 28.8 kbit/s for the time being."
A speed of 28.8 kb/s no longer provides functional internet access, on any common-sense understanding of the term. For example, even an optimised page such as the www.tesco.com home page weighs in at 54 KB, and would take around 15 seconds to load. The www.amazon.co.uk home page totals 740KB, and would take 3m25s to load.
Just updating the 28.8 kb/s figure from 2003 by applying Neilsen’s Law would give figures of 512 kb/s in 2010 and 4 Mb/s in 2015.
The UK government has recently announced an unambitious target of 2 Mb/s for 2015. This will leave us some 10 years behind Finland (100 Mb/s in 2015) and 20 years behind Singapore (where they are currently installing 1 Gb/s connections to the home).

Dispersion of Scotland's Population

In his comments on our interim report, Alan Arnott noted our statement that,

“Over 10% of the population of Scotland lives in population densities greater than 10,000 people per km², while the least accessible 15% live at densities of less than 1000 people per km²."

He commented that,

the UK has an overall population density of around 350 people per km² with Scotland around 65/km² so the above figures seem somewhat inflated.

We based our analysis on the Census 2001 data for "output areas" (OA). Our interim report attempted to summarise the contrast between the dense cities and sparse hinterland. The figures we gave are correct, but rounded.

In fact over 11.0% (but less than 11.1%) live at densities over 10,000 p/km². The least accessible 14.6% live at densities < 1,000 p/km². The least accessible 15% live at densities < 1,070 p/km². Instead of rounding the percentage, we should have rounded the population count and said, "while the least accessible 15% live at densities of less than 1100 people per km²".

However, the comment prompted us to return to make a finer comparison with England, with a surprising result.
The population densities averaged over each country are Scotland 65 p/km²; England 377 p/km². One might naturally expect this to imply that, in general, people in Scotland live at much lower population densities than people in England. This is not the case.

At a local level, 90% of the population of Scotland lives at population densities at least as high as those in England!

There are 46,604 output areas in Scotland, and 165,665 in England. The "Key Statistics" table gives area (A) and population (P) for each output area.
We calculate population density for each output area (as the quotient P/A), and list the output areas in order of decreasing population density. The cumulative percentage of population included in initial segments of this list is used as abscissa (x-axis) in the graphs below (click on the graphs for larger images).
Our interim report included this plot: It shows, for each x, the percentage of the area of Scotland occupied by the most densely housed x% of the population, with the area shown on a log scale.
To compare the population densities in Scotland and England, we plot two new graphs (below) of population density. In each case we use a log scale, and units of people per hectare (for p/km², multiply by 100). The data for England (green) are compared with those for Scotland (blue).
The upper graph plots the population density for each output area against the cumulative centile within which its population falls. For the first 90 centiles of population from each country, those in Scotland live at marginal densities at least as high those in England. Scotland has many small communities, often packed more densely than their counterparts in England. So "last mile" solutions that work in England should be applicable also to communities in Scotland. The lower graph plots the "cumulative density", computed as the quotient, cumulative population/cumulative area. This shows how the gap between the overall figures of 65/km² for Scotland and 377 p/km² for England as a whole arises because the last 10% of the population is much more sparsely distributed in Scotland than in England.
What these graphs do not show is that many of Scotland's communities are widely separated. The "middle mile" connections required to bring broadband to these communities are a particular concern.