Boundary-drawing policy issues

[Deleted]

However, if you run the algorithm enough time, eventually amalgamating the best parts of each result should give you something workable

[islington]

The advantage of an algorithm is that it would come up with possible solutions that no human being would ever have thought of.
The problem, of course, is that the reason no human being would have thought of them is, almost always, that they are terrible.
But once in a blue moon, maybe, the algorithm might come up with something of real merit; so perhaps it's worth using one provided a human being takes the final decision.

I'm reminded of a story about early efforts to get computers to interpret the English language. The programmers reasoned that vocabulary is a straightforward matter, and that the rules of grammar, although complex, are essentially mechanical in nature: therefore it should be perfectly possible, if laborious, to teach a computer English.

What they hadn't reckoned on is the potential ambiguity of even apparently simple English sentences. Human users of English are generally not aware of this because we bring to any newly-encountered English sentence a wealth of experience about the way the language is used, and the kind of information or ideas someone is likely to want to convey; and without even realizing it we apply this experience, unconsciously reject perverse or absurd interpretations (even if they are perfectly sound grammatically), and we almost always interpret the sentence in the manner its author intended.

The story goes that a computer was presented with the innocuous and apparently unambiguous sentence, 'Time flies like an arrow.'

It came up with the following interpretations of it -

- as an imperative: Measure the speed of flies like you would measure the speed of an arrow.
- as a different imperative: Measure the speed of flies using methods that an arrow would use.
- as yet a third imperative: Measure the speed of flies with qualities resembling those of arrows.
- as a declarative: Time moves in the way an arrow would. (Actually the intended meaning.)
- as a different declarative: Certain flying insects, called "time flies", enjoy an arrow.

[Adrian]

The Electoral Calculus algorithm is a useful tool for making statistical projections, but it is not a useful tool for the actual creation of new boundaries.

[carlton43]

Nov 1, 2020 18:58:11 GMT islington said:

The advantage of an algorithm is that it would come up with possible solutions that no human being would ever have thought of.
The problem, of course, is that the reason no human being would have thought of them is, almost always, that they are terrible.
But once in a blue moon, maybe, the algorithm might come up with something of real merit; so perhaps it's worth using one provided a human being takes the final decision.

I'm reminded of a story about early efforts to get computers to interpret the English language. The programmers reasoned that vocabulary is a straightforward matter, and that the rules of grammar, although complex, are essentially mechanical in nature: therefore it should be perfectly possible, if laborious, to teach a computer English.

What they hadn't reckoned on is the potential ambiguity of even apparently simple English sentences. Human users of English are generally not aware of this because we bring to any newly-encountered English sentence a wealth of experience about the way the language is used, and the kind of information or ideas someone is likely to want to convey; and without even realizing it we apply this experience, unconsciously reject perverse or absurd interpretations (even if they are perfectly sound grammatically), and we almost always interpret the sentence in the manner its author intended.

The story goes that a computer was presented with the innocuous and apparently unambiguous sentence, 'Time flies like an arrow.'

It came up with the following interpretations of it -

- as an imperative: Measure the speed of flies like you would measure the speed of an arrow.
- as a different imperative: Measure the speed of flies using methods that an arrow would use.
- as yet a third imperative: Measure the speed of flies with qualities resembling those of arrows.
- as a declarative: Time moves in the way an arrow would. (Actually the intended meaning.)
- as a different declarative: Certain flying insects, called "time flies", enjoy an arrow.

Ambiguity is the core of humanity and the very essence of all art.

[emidsanorak]

The Boundary Commissions will take into account:

"(a) special geographical considerations, including in particular the size, shape and accessibility of a constituency;
(b) local government boundaries as they exist on the most recent ordinary council-election day before the review date*;
(c) boundaries of existing constituencies;
(d) any local ties that would be broken by changes in constituencies;
(e) the inconveniences attendant on such changes."

* changed in another clause to allow local government boundary changes which are prospective to be used.

A lot of the schemes posted on this website don't. In particular, many unnecessarily divide towns and villages.

For example, I posted a ten seat Leicestershire which does not divide any settlement outside the city. The other Leicestershire schemes don't. The villages of Enderby, Ibstock and Sileby are particularly prone to vicious attacks from multiple posters.

I would suggest that: if you change the colours and borders settings on Boundary Assistant, changing the opacity to 55%, changing the current constituencies and local authorities to 2 pixels; if you click on satellite background, Primary Authority and Westminster Constituency, you can see much more clearly how to avoid butchering towns and villages and to keep to the other guidelines

[East Anglian Lefty]

One thing worth considering with regard to cross-county constituencies is that they're likely to be more transitory than other constituencies, because a) nobody really likes the idea, b) we can't guarantee the rules allowing them will be around forever and c) they're created because a county is on its way to gaining/losing a seat but hasn't got all the way there yet.

So as a general principle, when creating a cross-county seat it's probably helpful to think about how it can be undone in a review or two's time when long-term population trends have run a little further. This isn't that helpful for mets, where ward sizes mean it's primarily a numbers game, but it might be relevant in counties like Suffolk and Herts - to work out which of several plausible opportunities for cross-county seats is best, perhaps it's sensible to draw the county with an extra seat first and work out how a cross-county seat might be a stepping stone to that end point.

[J.G.Harston]

Jan 20, 2021 8:34:53 GMT East Anglian Lefty said:

One thing worth considering with regard to cross-county constituencies is that they're likely to be more transitory than other constituencies, because a) nobody really likes the idea, b) we can't guarantee the rules allowing them will be around forever and c) they're created because a county is on its way to gaining/losing a seat but hasn't got all the way there yet.

So as a general principle, when creating a cross-county seat it's probably helpful to think about how it can be undone in a review or two's time when long-term population trends have run a little further. This isn't that helpful for mets, where ward sizes mean it's primarily a numbers game, but it might be relevant in counties like Suffolk and Herts - to work out which of several plausible opportunities for cross-county seats is best, perhaps it's sensible to draw the county with an extra seat first and work out how a cross-county seat might be a stepping stone to that end point.

A good point, but it has to be explained clearly and carefully. I made this point at the Sheffield ward review initial discussions - look ahead to the future to see where potential future growth is likely to be and that can guide if you draw a line this side of a street or that side of a street - and the wrong end of the stick was immediately grasped - so, we don't have to ensure electoral equality?
Sigh! No, you *do* have to ensure electoral equality, but in ensuring that electoral equality you can - in your internal mental assessments - look ahead to where that electoral equality will eventually be broken and chose to look for lines that ensure electoral equality today but are more amenable to adjusting in the future when the electoral equality is broken.
So we don't have to ensure electoral equality?
ARGGH!!

[Wisconsin]

Jan 19, 2021 12:31:32 GMT emidsanorak said:

The Boundary Commissions will take into account:

"(a) special geographical considerations, including in particular the size, shape and accessibility of a constituency;
(b) local government boundaries as they exist on the most recent ordinary council-election day before the review date*;
(c) boundaries of existing constituencies;
(d) any local ties that would be broken by changes in constituencies;
(e) the inconveniences attendant on such changes."

* changed in another clause to allow local government boundary changes which are prospective to be used.

There’s a sixth factor for England. And note the statutory language “may take into again if and to such extent as they think fit” not “will take into account”

Here’s the statute:

5 (1) A Boundary Commission may take into account, if and to such extent as they think fit—

(a) special geographical considerations, including in particular the size, shape and accessibility of a constituency
(b) local government boundaries which exist, or are prospective, on the review date;
(c) boundaries of existing constituencies;
(d) any local ties that would be broken by changes in constituencies;
(e) the inconveniences attendant on such changes.

(2) The Boundary Commission for England may take into account, if and to such extent as they think fit, boundaries of the English regions specified in sub-paragraph (2A) as they exist on the most recent ordinary council-election day before the review date

[islington]

Jan 20, 2021 8:34:53 GMT East Anglian Lefty said:

One thing worth considering with regard to cross-county constituencies is that they're likely to be more transitory than other constituencies, because a) nobody really likes the idea, b) we can't guarantee the rules allowing them will be around forever and c) they're created because a county is on its way to gaining/losing a seat but hasn't got all the way there yet.

So as a general principle, when creating a cross-county seat it's probably helpful to think about how it can be undone in a review or two's time when long-term population trends have run a little further. This isn't that helpful for mets, where ward sizes mean it's primarily a numbers game, but it might be relevant in counties like Suffolk and Herts - to work out which of several plausible opportunities for cross-county seats is best, perhaps it's sensible to draw the county with an extra seat first and work out how a cross-county seat might be a stepping stone to that end point.

There's a real policy dilemma about cross-county seats.

Historically, as has been pointed out in another thread, for over a century it has been entirely unknown, except where Rutland is involved, for any seat in England, at the time it is drawn up, to comprise parts of more than one county ("county" as per local government at the time, not historic county, and also ignoring the fact that UAs may be technically defined as counties).

So long-established practice, to say nothing of the very valid point made by EAL, suggests that cross-county seats should be avoided as much as possible.

On the other hand, the BCE, during both the recent unimplemented reviews, showed a surprising (to me) willingness to cross county boundaries, including in cases where there were obvious single-county alternatives. Moreover, we see in the present review a tension that is particularly acute in Hampshire between (a) respecting existing seats in most of the county, but crossing the boundary with either Berks or Surrey; and (b) drawing 18 seats wholly within Hants but at the price of effectively ripping up the existing map across great swathes of the county.

I can see a powerful case for either approach. My first instinct is to adhere to the county boundary if at all possible, but the extent of the disruption to the current map certainly give me pause.

I've also been thinking about other cases involving at least one two-tier county where the numbers push us towards cross-county solutions. On the face of it the numbers look bad in Devon but we seem to have arrived at a satisfactory treatment; Cumbria obviously has to pair with Lancs but despite bad numbers there's no need for a further crossing into Gtr Manchester or Merseyside (although some plans do this); N Yorks is mathematically required to pair with something, although details vary between plans; Lincs-Rutland seems obvious; Herts seems to pair well with the Beds UAs; Norfolk and Suffolk are a reasonably logical combination; the Sussexes even more so (this hardly feels like a cross-county issue at all).

Where a crossing is necessary, it seems to work best when the cross-county seat can be focused either on a single town near the border such as Hitchin, or on significant towns on either side such as Thetford and Newmarket. I was much impressed by the superiority of the latter combination compared with the alternative of a cross-county seat involving vast swathes of rural southeast Norfolk and sortheast Suffolk with no proper focus and not even a reasonable name.

These are just my thoughts: I'm not in the business of trying to second-guess what the BCE will do but if I had to put money on it, I'd expect a fair number of cross-county proposals.

[Wisconsin]

I think the job is hard enough already for them to also have to consider factors not set out in the legislation.

If in doubt, the statutory factors should clearly trump respecting archaic county boundaries.

[YL]

County boundaries are in the legislation.

(What I'm going to say here is always "in general" rather than absolute. I think making absolute rules in this risks creating bad results, so there's always the potential for exceptions.)

I do tend to try to respect county boundaries where reasonably possible, and would go slightly further to include other major local government boundaries, i.e. unitaries (most of which are actually legally counties anyway) and Met boroughs, but at a slightly lower level of protection. (Many of them are too small to actually be treated on their own, of course.) I have various reasons for this; one is that I think that most of these have quite a strong identity and people know which one they live in, which makes a good basis for constituencies. Another reason is that I think the regions are really too large to be sensible review areas for the consultation: I don't like the situation where decisions made about boundaries in Sheffield have knock on effects in Great Ayton (a real example from the zombie reviews) and think it helps to break things down into manageable subregions, and thinking in terms of counties and Met boroughs helps a lot. (Yes, there can be choices made about what the subregions should be, so it's not simple.)

Related to that is my view that it should be OK to split the odd ward when the only reason why a major boundary might have to be crossed is awkward ward sizes on one side. Wards in some areas (and with more unitaries the number may be increasing) are really too big to be used as building blocks, and in these areas the wards are often just convenient agglomerations of neighbouring areas which may well be broken up at the next ward boundary review; I think the case for keeping them together is quite a bit weaker than for trying to avoid crossing the boundary. (In general, of course. Of course there's a case that some Met boroughs in particular are also just agglomerations of towns with a weakish overall identity.)

However, I think when a county's electorate is such that the seats consistently need to be at one end of the range (e.g. Hampshire, Devon), it's worth looking at the situation and deciding whether the county can actually be done reasonably sensibly, with regard to the other criteria and just generally ensuring sensible seats, or not. I think that Devon can, but I'm not convinced in the case of Hampshire. Perhaps a split ward in Southampton would help a bit...

[Sibboleth]

Quite radical changes in re-warding is now common practice, so it isn't as if wards have a fundamental long-term integrity to them that ought to be respected: they are convenient and nothing more. So when it is convenient to split wards (as with a 5% quota it often will be) then there should be no default prejudice against doing so.

[Khunanup]

Constituencies should be based on communities and arbitrary administrative boundaries (whether that is at ward, local authority or ceremonial county level) should be a secondary consideration.

The fact that it isn't a primary consideration is down to it being easier for those who are tasked to come up with the boundaries rather than any desire to put more work in and create coherent seats.

[hullenedge]

[Peter Wilkinson]

Feb 18, 2021 13:29:30 GMT hullenedge said:

Of course, that map only indicates some of the seats that will need to be redrawn.

Firstly, when a seat needs to be redrawn because it is too large or too large small, then the surplus or required voters need to go to or come from some other seat - and seats that are too large do not come in convenient contiguous pairs with seats that are correspondingly too small. So this, of course, very often means that seats that are within quota have to be redrawn to accommodate changes to neighbouring large or small seats, and frequently then need to have this compensated by changes to other seats that may themselves also be in quota. For instance, only three of Hertfordshire's eleven seats are not within quota (all three being too large), but any valid revised plan for Hertfordshire seems to require redrawing most of the other eight seats.

Secondly, ward boundaries have often changed since the last completed parliamentary boundary review - and in quite a number of cases in ways which mean that current constituency boundaries no longer wholly follow current ward, or even (where known) polling district, boundaries. Where this is the case, it may be in principle be possible to retain the previous boundary, but in practice, potential ongoing administrative problems - or just the difficulty of obtaining the figures to demonstrate that the seats concerned are still within quota - will require the seats to be redrawn to follow current ward (or polling district) boundaries.

And beyond these considerations, there are other possible, if probably less likely, factors to take into account - making sure, for instance, that each nation and region (or even the UK as a whole) contains the right number of seats.

[hullenedge]

Feb 18, 2021 16:10:29 GMT Peter Wilkinson said:

Feb 18, 2021 13:29:30 GMT hullenedge said:

Of course, that map only indicates some of the seats that will need to be redrawn.

Firstly, when a seat needs to be redrawn because it is too large or too large small, then the surplus or required voters need to go to or come from some other seat - and seats that are too large do not come in convenient contiguous pairs with seats that are correspondingly too small. So this, of course, very often means that seats that are within quota have to be redrawn to accommodate changes to neighbouring large or small seats, and frequently then need to have this compensated by changes to other seats that may themselves also be in quota. For instance, only three of Hertfordshire's eleven seats are not within quota (all three being too large), but any valid revised plan for Hertfordshire seems to require redrawing most of the other eight seats.

Secondly, ward boundaries have often changed since the last completed parliamentary boundary review - and in quite a number of cases in ways which mean that current constituency boundaries no longer wholly follow current ward, or even (where known) polling district, boundaries. Where this is the case, it may be in principle be possible to retain the previous boundary, but in practice, potential ongoing administrative problems - or just the difficulty of obtaining the figures to demonstrate that the seats concerned are still within quota - will require the seats to be redrawn to follow current ward (or polling district) boundaries.

And beyond these considerations, there are other possible, if probably less likely, factors to take into account - making sure, for instance, that each nation and region (or even the UK as a whole) contains the right number of seats.

Yes, the chap that drew this map is, if I'm correct, a member of this forum. Hope that he posts more material.

[YL]

Feb 18, 2021 16:10:29 GMT Peter Wilkinson said:

Secondly, ward boundaries have often changed since the last completed parliamentary boundary review - and in quite a number of cases in ways which mean that current constituency boundaries no longer wholly follow current ward, or even (where known) polling district, boundaries. Where this is the case, it may be in principle be possible to retain the previous boundary, but in practice, potential ongoing administrative problems - or just the difficulty of obtaining the figures to demonstrate that the seats concerned are still within quota - will require the seats to be redrawn to follow current ward (or polling district) boundaries.

It is certainly true that such constituencies are almost certain to be redrawn, but surely current polling districts won't be split between constituencies, as they might still be needed if a by-election (or a snap General Election) were to be called? In Sheffield, there are several small polling districts which are areas which were left in a different constituency from most of their ward after the last ward review. One of these actually just consists of part of a block of flats which was built across an old ward boundary which is still a constituency boundary. So a few of the flats in the development are in Sheffield Hallam while the rest are in Sheffield Central; the whole development is now in Walkley ward, but the Hallam part is a separate polling district.

Sheffield Hallam is indeed an example of an in quota constituency which needs to be redrawn for this reason; it'd be absurd to retain those flats in Hallam so that the constituency can be left unchanged!

[J.G.Harston]

Feb 19, 2021 11:20:58 GMT YL said:

Feb 18, 2021 16:10:29 GMT Peter Wilkinson said:

Secondly, ward boundaries have often changed since the last completed parliamentary boundary review - and in quite a number of cases in ways which mean that current constituency boundaries no longer wholly follow current ward, or even (where known) polling district, boundaries. Where this is the case, it may be in principle be possible to retain the previous boundary, but in practice, potential ongoing administrative problems - or just the difficulty of obtaining the figures to demonstrate that the seats concerned are still within quota - will require the seats to be redrawn to follow current ward (or polling district) boundaries.

It is certainly true that such constituencies are almost certain to be redrawn, but surely current polling districts won't be split between constituencies, 

Why not? Polling districts are not electoral divisions, they are arbitary conveniences for the gathering of ballots, and can be redrawn at any time. They are drawn *after* electoral divisions have been drawn for the convencience of sub-division organisation. No divisions should be bound by polling district boundaries, that gives the mess we had in Sheffield in 1983-2004.

Edit: re-reading a couple of times clarifies that YL's saying: surely current PDs aren't split across current constituencies, I read it as: surely current PDs won't be split across new constituencies. YL is correct, it's impossible for a polling district to cover more than one electoral division of any sort, as it is a *subdivision* of an electoral area for collecting ballots, so all those ballots must be from the same electoral area, so a PD must be wholly within the smallest electoral area that it is within. Of course, multiple PD may vote in the same *station*, but that is a different matter, eg South Road No. 1 and No. 2 both vote at Walkley Library for local elections, but for a Parliamentary election No.2 has to vote in the Hallam station as it's in Hallam.

[islington]

The issue of how far it is acceptable to propose seats that straddle substantial bodies of water has come up a couple of times in the Scotland thread, but in principle it could arise in any part of the UK so I'm addressing it here.

In defence of these seats the plea is often entered that there is a bridge. My question is whether this is enough.

Take the case illustrated. It certainly passes the 'bridge' test (there are three) but would this seat (electorate 72463) be acceptable if someone were to suggest it? I think not; because there's a difference between a river and an estuary (or inlet). A river is generally acceptable to cross, definitely if there's a bridge within the seat and perhaps on occasion even if there isn't. But I'm much more dubious about an estuary or inlet, even if bridged.

As to where one draws the line between a river and an estuary (or inlet), we might need to ask the advice of a geographer. I'd say it's something to do with (a) how wide it is; (b) whether it is tidal; and (c) more nebulously, something to do with how closely the line of each bank tracks the line of the other. I'm not sure any of these factors is decisive in itself - it's a matter of how they combine in the particular case. 

[YL]

Apr 14, 2021 12:29:04 GMT islington said:

The issue of how far it is acceptable to propose seats that straddle substantial bodies of water has come up a couple of times in the Scotland thread, but in principle it could arise in any part of the UK so I'm addressing it here.

In defence of these seats the plea is often entered that there is a bridge. My question is whether this is enough.

Take the case illustrated. It certainly passes the 'bridge' test (there are three) but would this seat (electorate 72463) be acceptable if someone were to suggest it? I think not; because there's a difference between a river and an estuary (or inlet). A river is generally acceptable to cross, definitely if there's a bridge within the seat and perhaps on occasion even if there isn't. But I'm much more dubious about an estuary or inlet, even if bridged.

As to where one draws the line between a river and an estuary (or inlet), we might need to ask the advice of a geographer. I'd say it's something to do with (a) how wide it is; (b) whether it is tidal; and (c) more nebulously, something to do with how closely the line of each bank tracks the line of the other. I'm not sure any of these factors is decisive in itself - it's a matter of how they combine in the particular case. 

In their provisional proposals for the Fifth Review, the BCE proposed a "Wallasey & Kirkdale" constituency which crossed the Mersey estuary, and which was connected by one of the Mersey tunnels, so it was quite comparable to your example. It was pretty much universally panned and was abandoned in the final recommendations, which gave Wirral MB four seats of its own (two of which were, and still are, rather undersized). So I think such arrangements are in theory allowable but I also think they are unlikely to be acceptable in practice. There was a reason for the choice of thread for my most recent map.

(Wallasey & Kirkdale is not to be confused with the infamous Mersey Banks of the first zombie review. That was even worse, and didn't include a crossing of the estuary.)