“We let algorithms redesign Scotland’s local government boundaries.” Here’s what happened

Glen Coe. The Highlands council areas are sparsely populated, wherever you draw the boundaries. Image: Getty.

There are certain things a wise politician would never tinker with: big issues such as council tax reform, which are placed in the “too difficult” box and remain unaddressed and unresolved for decades. Another such item is the redrawing of local government boundaries.

It’s not exactly a glamorous topic and, as BBC Scotland Political Editor Brian Taylor noted in 2012, “it is a courageous politician indeed who tampers with cooncil boundaries”/

Over the past few years in Scotland, there have been calls to reform council boundaries, and give local authorities a greater say in decisions made at Holyrood. But so far, nothing has been done and the current set of 32 Scottish council areas remain.

Ripe for reform

Despite the political risk associated with boundary reform, other governments across the world have grasped this particular nettle, because it can lead to more efficient governance and cut costs.

Recent regional rejigs in Denmark and France show that it can be done. And we believe an algorithmic approach can at the very least contribute to the debate. In France, the so-called “big bang des régions” of 2016 resulted in a total of 13 new regions, from a previous set of 22.

So, we thought it would be useful to make some suggestions about what a revised Scottish council geography could look like. To do this, we used commuting data and an algorithm called Combo, which groups together places with the strongest ties, using a set of rules. And instead of the current set of 32 Scottish council areas, we ended up with 17.

Combo shows how commuting patterns can help define areas. Image: Alasdair Rae and Ruth Hamilton.

The full set of results and the detailed method can be found in our recently published paper. However, we should emphasise that these are not proposals for new council areas. Instead, we hope our results offer decision-makers some useful evidence on where boundaries could be drawn, if they were based on travel to work patterns.

Of course, commuting is only one part of the story. But it’s a very important part, particularly in places such as Glasgow, where the local economy is highly dependent on commuters living in other areas.

Overall, we think the results are largely plausible. But they also show that, when it comes to drawing boundaries on maps, algorithms definitely shouldn’t have the last word – not least because they don’t have what we might call common sense.

Scotland’s council areas, if they were based on commuting and drawn by algorithm. Image: Alasdair Rae and Ruth Hamilton.

As you can see from the map, there are now 17 areas instead of the current set of 32 – two fewer than those proposed by think tank Reform Scotland in 2012. The most populous of the new areas is Greater Glasgow, with close to 1.2m people. This would make it the largest local authority in the UK.

This is followed by Edinburgh & the Lothians with 835,000. Lanarkshire is next, with 575,000, then Grampian with 476,000 people. These areas are considerably larger than their current equivalents.

Others, like Highland, Moray and Inverclyde stay the same, indicating that some administrative boundaries match existing travel to work patterns quite well. Or, to put it more simply, in some cases political and economic geographies look very similar.

Is or ought?

The table below shows the relationship between our new Combo areas and existing council areas. In the final column, you can see what percentage of the population of each new area comes from existing council areas. For example, 50.8 per cent of our new Greater Glasgow area is from the current City of Glasgow and 57.1 per cent of the new Edinburgh & the Lothians area is from the current City of Edinburgh.

Combo areas compared to current council areas. Image: Alasdair Rae and Ruth Hamilton.

On some levels, these new algorithmic subdivisions make a lot of sense – but others clearly require a human touch. For one thing, it doesn’t look quite right based on our understanding of how places function and how local services are delivered. The fact that 2 per cent of the population of our new Dundee, Perth and Angus area is across the Tay in Fife strongly suggests human intervention is needed.

The second issue relates to Scottish Enlightenment philosopher David Hume’s “is-ought” question. This is because algorithms are very good at telling us what is, but they cannot tell us what ought to be.


For example, the algorithm suggests that Inverclyde is an entirely separate area. But given how close it is to our Greater Glasgow area, from a human and common sense perspective, it appears that it ought to be part of Greater Glasgow. Indeed, such a geography would match the current NHS Greater Glasgow & Clyde region.

We need to recognise that algorithms are effectively blind to questions of history and identity. For example, our new boundaries know nothing of the Kingdom of Fife, or the Pictish Kingdom that preceded it. Yet such history often matters deeply to people and places.

This is where human decision-making, nuance and common sense come in, and why – when it comes to “too difficult” issues such as local boundary reform – humans should have the final say. But in getting to that point, we believe algorithms can certainly lend a helping hand.

The Conversation

Alasdair Rae, Professor in Urban Studies and Planning, University of Sheffield and Ruth Hamilton, Teaching Associate, University of Sheffield.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

 
 
 
 

Never mind Brexit: TfL just released new tube map showing an interchange at Camden Town!!!

Mmmmm tube-y goodness. Image: TfL.

Crossrail has just been given a £1bn bail out. This, according to the Financial TImes’s Jim Pickard, is on top of the £600m bailout in July and £300m loan in October.

That, even with the pound crashing as it is right now, is quite a lot of money. It’s bad, especially at a time when there is still seemingly not a penny available to make sure trains can actually run in the north.

But the world is quite depressing enough today, so let’s focus on something happier. On Saturday night – obviously peak time for cartographic news – Transport for London emailed me to let me know it would be updating the tube map, to show more street-level interchanges:

Connections between several pairs of stations that are near to each other, but have traditionally not been shown as interchanges, now appear on the map for the first time. These include:

  • Camden Road and Camden Town
  • Euston and Euston Square
  • Finchley Road and Finchley Road & Frognal
  • Kenton and Northwick Park
  • New Cross and New Cross Gate
  • Seven Sisters and South Tottenham
  • Swiss Cottage and South Hampstead

The stations shown meet a set of criteria that has been used to help determine which should be included. This criteria includes stations less than a 700m or a 10 minute walk apart, where there is an easy, well-lit, signposted walking route and where making the change opens up additional travel options.

The results are, well, this:

In addition, interchanges between stations have traditionally appeared on the Tube map as two solid lines, irrespective of whether they are internal or external (which means customers need to leave the station and then re-enter for the station or stop they need). This approach has now been updated and shows a clear distinction between the two types, with external interchanges now being depicted by a dashed line, linking the two stations or stops.

And lo, it came to pass:

I have slightly mixed feelings about this, in all honesty. On the positive side: I think generally showing useful street-level interchanges as A Good Thing. I’ve thought for years that Camden Road/Camden Town in particular was one worth highlighting, as it opens up a huge number of north-east travel options (Finchley to Hackney, say), and apps like CityMapper tell you to use it already.


And yet, now they’ve actually done it, I’m suddenly not sure. That interchange is pretty useful if you’re an able bodied person who doesn’t mind navigating crowds or crossing roads – but the map gives you no indication that it’s a harder interchange than, say, Wanstead Park to Forest Gate.

The new map also doesn’t tell you how far you’re going to be walking at street level. I can see the argument that a 400m walk shouldn’t disqualify something as an interchange – you can end up walking that far inside certain stations (Green Park, Bank/Monument), and the map shows them as interchanges. But the new version makes no effort to distinguish between 100m walks (West Hampstead) and 700m ones (Northwick Park-Kenton), which it probably should.

I’m also slightly baffled by some of the specific choices. Is Finchley Road-Finchley Road & Frognal really a useful interchange, when there’s an easier and more direct version, one stop up the line? No hang on West Hampstead isn’t on the Metropolitan line isn’t it? So that’s what it’s about.

Okay, a better one: if you’re switching from District to Central lines in the City, you’re generally better off alighting at Cannon Street, rather than Monument, for Bank – honestly, it’s a 90 second walk to the new entrance on Walbrook. Yet that one isn’t there. What gives?

The complete new tube map. The full version is on TfL’s website, here.

On balance, showing more possible interchanges on the map is a positive change. But it doesn’t negate the need for a fundamental rethink of how the tube map looks and what it is for. And it’s not, I fear, enough to distract from the Crossrail problem.

Jonn Elledge is the editor of CityMetric. He is on Twitter as @jonnelledge and on Facebook as JonnElledgeWrites.