|
|
Post by Peter Wilkinson on Aug 10, 2020 8:46:57 GMT
The law is about to give the PBCs power to "take account" of "local government boundaries which ... are prospective, on the review date" - which means new ward boundaries which are unimplemented but have been authorised in secondary legislation. I would expect the guidance to be quite clear that where ward boundary changes have been authorised in a statutory instrument, the PBCE will use them and not the current ward boundaries. I can see a problem with that - this will in effect require, for instance, London boroughs to produce accurate March 2020 electorate figures for wards which they will not be using until May 2022, and for which they would otherwise not have to produce electoral registers using those boundaries before December 2021. This should not be impossible - but I get the impression that many local authorities resource their electoral registration departments at a low enough level that calculating the requisite March 2020 figures will turn out to be quite beyond their competence and resourcing. Of course, it should be possible in principle to do this not just at ward level but down to polling district level, but because of an extra consideration, I regard this as, in general, beyond the bounds of practical possibility. The consideration is that polling districts have to respect all electoral boundaries - including ward boundaries - with the consequence that any changes in ward boundaries will usually require changes to at least two polling districts. But changes to polling districts do not take place within boundary reviews - they are the result of a separate review by the relevant local authority, to be undertaken after, for instance, a ward review has been completed and before the new ward boundaries are to be used in an electoral register. For this reason, I would guess that most London boroughs will not only not yet have started polling district reviews that they need to conduct before the May 2022 elections (and preparation of the December 2021 register) but, given the distractions of Covid-19, will currently have no plan to start on these before the summer, or even the autumn, of 2021. I am fairly sure that local authorities could be required, where necessary, to bring these polling district reviews forward to within the next six months or so. But I very much doubt that the PBCE will insert anything into its guidance, without government instruction, that would activate this requirement. In brief - I would expect PBCE guidance only to commit to using new ward boundaries if the government is publicly prepared to back up the guidance by requiring local authorities to produce March 2020 electorate figures for the new wards (though I think it more likely than not that the government will do so). And I would further expect the PBCE only to be willing even to contemplate any ward-splitting if, as and when polling district electorate information for the whole of England is available (and I think it unlikely that the government intends to actively expedite this). |
|
|
|
Post by kevinlarkin on Aug 10, 2020 12:18:20 GMT
The law is about to give the PBCs power to "take account" of "local government boundaries which ... are prospective, on the review date" - which means new ward boundaries which are unimplemented but have been authorised in secondary legislation. I would expect the guidance to be quite clear that where ward boundary changes have been authorised in a statutory instrument, the PBCE will use them and not the current ward boundaries. One thing about this is unclear to me. If we get to 1st December and there are ward boundary changes for which a draft Statutory Instrument has been laid before Parliament but the negative resolution procedure (requiring 40 sitting days) is not complete, will the parliamentary boundary commissions be required to use the new ward boundaries or the old? |
|
|
|
Post by Wisconsin on Aug 10, 2020 12:39:02 GMT
The law is about to give the PBCs power to "take account" of "local government boundaries which ... are prospective, on the review date" - which means new ward boundaries which are unimplemented but have been authorised in secondary legislation. I would expect the guidance to be quite clear that where ward boundary changes have been authorised in a statutory instrument, the PBCE will use them and not the current ward boundaries. One thing about this is unclear to me. If we get to 1st December and there are ward boundary changes for which a draft Statutory Instrument has been laid before Parliament but the negative resolution procedure (requiring 40 sitting days) is not complete, will the parliamentary boundary commissions be required to use the new ward boundaries or the old? Interesting. The question is whether a draft statutory instrument is “an means an instrument made under primary legislation”. |
|
|
|
Post by andrewteale on Aug 10, 2020 12:55:52 GMT
The law is about to give the PBCs power to "take account" of "local government boundaries which ... are prospective, on the review date" - which means new ward boundaries which are unimplemented but have been authorised in secondary legislation. I would expect the guidance to be quite clear that where ward boundary changes have been authorised in a statutory instrument, the PBCE will use them and not the current ward boundaries. I can see a problem with that - this will in effect require, for instance, London boroughs to produce accurate March 2020 electorate figures for wards which they will not be using until May 2022, and for which they would otherwise not have to produce electoral registers using those boundaries before December 2021. This should not be impossible - but I get the impression that many local authorities resource their electoral registration departments at a low enough level that calculating the requisite March 2020 figures will turn out to be quite beyond their competence and resourcing. Of course, it should be possible in principle to do this not just at ward level but down to polling district level, but because of an extra consideration, I regard this as, in general, beyond the bounds of practical possibility. The consideration is that polling districts have to respect all electoral boundaries - including ward boundaries - with the consequence that any changes in ward boundaries will usually require changes to at least two polling districts. But changes to polling districts do not take place within boundary reviews - they are the result of a separate review by the relevant local authority, to be undertaken after, for instance, a ward review has been completed and before the new ward boundaries are to be used in an electoral register. For this reason, I would guess that most London boroughs will not only not yet have started polling district reviews that they need to conduct before the May 2022 elections (and preparation of the December 2021 register) but, given the distractions of Covid-19, will currently have no plan to start on these before the summer, or even the autumn, of 2021. I am fairly sure that local authorities could be required, where necessary, to bring these polling district reviews forward to within the next six months or so. But I very much doubt that the PBCE will insert anything into its guidance, without government instruction, that would activate this requirement. In brief - I would expect PBCE guidance only to commit to using new ward boundaries if the government is publicly prepared to back up the guidance by requiring local authorities to produce March 2020 electorate figures for the new wards (though I think it more likely than not that the government will do so). And I would further expect the PBCE only to be willing even to contemplate any ward-splitting if, as and when polling district electorate information for the whole of England is available (and I think it unlikely that the government intends to actively expedite this). The 2000-06 boundary review in England was drawn up entirely based on the new wards from the LGBCE's review process, most of which weren't introduced until 2002-04. The EROs were expected to produce 2000 electorate figures retrofitted to the new boundaries. |
|
Adrian
Co-operative Party
Posts: 1,742 
|
Post by Adrian on Aug 10, 2020 13:20:28 GMT
I've just been looking at my presentation to the Chester enquiry in 2011 (which was favourably received by Nick Winterton, inter alia) and found this: The boundary paradox The law says that if you use boundaries you have to use the 2010 boundaries, but the law doesn’t say you have to use those boundaries. ukelect.files.wordpress.com/2011/10/chester-presentation-a-bailey.ppt |
|
|
|
Post by Wisconsin on Aug 10, 2020 13:29:35 GMT
|
|
|
|
Post by emidsanorak on Oct 29, 2020 13:32:10 GMT
The Electoral Calculus car crash algorithm prompts me to try to suggest what a sensible algorithm would look like. The legislation says:
"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 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 this time to 2 March 2020
This means that we should be looking at issues around council boundaries, existing constituency boundaries, contiguity, accessibility, keeping settlements whole, and how it looks on the map.
Regions should be divided into sub-regions in order to minimise the number of seats that cross first tier LAs (counties, UAs, MBs, London boroughs). Sub-regions should be divided into sub-sub-regions in order to minimise crossing District boundaries.
For example, I, and many others, propose a sub-region of 8 seats for Lincolnshire and Rutland. I then go on to propose three sub-sub-regions: one seat in West Lindsey; three seats in Boston, East Lindsey and South Holland; and four seats in Lincoln, North Kesteven, South Kesteven and Rutland. This means that three of the proposed constituencies are contained within one LA and the other five have elements of two. This means that, within the sub-region, there are only 13 LA components of seats.
A sensible algorithm would aim to minimise the total number of these components in each region and sub-region.
Existing constituencies within quota might have a presumption of remaining unchanged, but this presumption is not a prescription. As can be seen from the example above, I propose to change the Gainsborough seat, but this is to make it coterminous with the West Lindsey District. Where local ward boundaries have been changed which necessitate a change in constituency boundaries, the presumption might not apply except where, such as in the case of Lincoln, no electors have actually been moved.
A sensible algorithm would seek to minimise the number of electors moved from one seat to another.
We should perhaps tell Electoral Calculus that, just because wards are next to each other on a map, it doesn’t mean that you can drive, cycle or walk between them. Mersey Banks is not acceptable. One way of dealing with this might be to mark each neighbouring ward only as being adjacent where there is a road that links them. This doesn’t deal with wards with detached parts such as Milton & Waterbeach in South Cambridgeshire or wards which are otherwise not internally accessible such as Gowy Rural in Cheshire West and Chester UA or South Hanningfield Stock & Margaretting in Chelmsford District.
A sensible algorithm would seek to flag up and avoid issues of inaccessibility.
We need to be able to define what a settlement is. Cities and towns which are bigger than quota have to be divided. Within such settlements there are smaller settlements which should not be divided unless they already are. The way the guidance in the legislation is written, it is clear that, where there is an existing LA or constituency boundary, this is not so much of a problem. In Derby, the Chaddesden, Derwent and Oakwood wards make up the settlement of Chaddesden. Chaddesden and Derwent wards are in Derby North while Oakwood ward is in the abomination that is Mid Derbyshire. I would prefer that Chaddesden were reunited, but the guidance says otherwise.
Grimsby and Cleethorpes together form one urban whole, but their residents know exactly which settlement they live in. Given the grotesque regional boundary that snakes through the East and West Ridings of the Kingdom of Lindsey, Cleethorpes has to be included in a seat that includes most, but not all, Grimsby wards. One way of doing this divides the village of Humberston as well as the town of Grimsby. My preferred option is to include the Freshney, West Marsh and Yarborough wards in the seat that extents to the Trent. This then leaves the rest of Grimsby, the whole of Cleethorpes, and the whole of the villages of Humberston and New Waltham in the more compact seat.
A sensible algorithm would seek to minimise the number of settlements divided, but I don’t know how we would teach it to identify a settlement.
We can all recognise a neat constituency proposal. Some seats look beautiful on the map. Others look horrid. But I’m not sure how we teach these aesthetics to the computer.
A sensible algorithm would seek to minimise the length of constituency boundaries.
|
|
|
|
Post by greenchristian on Oct 29, 2020 20:03:06 GMT
Existing constituencies within quota might have a presumption of remaining unchanged, but this presumption is not a prescription. This is a bad idea. There are several examples in the two zombie reviews where leaving a constituency within quota caused some really bad knock-on effects. Everything else gives you a decent basis for an algorithm, though. |
|
ilerda
Conservative
Posts: 1,104
|
Post by ilerda on Oct 29, 2020 21:15:14 GMT
Existing constituencies within quota might have a presumption of remaining unchanged, but this presumption is not a prescription. This is a bad idea. There are several examples in the two zombie reviews where leaving a constituency within quota caused some really bad knock-on effects. Everything else gives you a decent basis for an algorithm, though. Also it assumes that existing constituencies are “good” constituencies that ought to be kept. The reality is that that are numerous highly imperfect and undesirable existing constituencies that ought to be scrapped even if they are within quota at every review for the next 200 years. |
|
J.G.Harston
Lib Dem
Leave-voting Brexit-supporting Liberal Democrat
Posts: 14,808
|
Post by J.G.Harston on Oct 29, 2020 22:24:58 GMT
This is a bad idea. There are several examples in the two zombie reviews where leaving a constituency within quota caused some really bad knock-on effects. Everything else gives you a decent basis for an algorithm, though. Also it assumes that existing constituencies are “good” constituencies that ought to be kept. The reality is that that are numerous highly imperfect and undesirable existing constituencies that ought to be scrapped even if they are within quota at every review for the next 200 years. Case in point: Sheffield Attercliffe has been creeping further and further south because at every review "minimum change" has taken the broken boundaries and broken them more, but just minimally more. Similarly, the status quo assumes the nonsense of Stannington being in Hallam. |
|
Khunanup
Lib Dem
Portsmouth Liberal Democrats
Posts: 12,039
Member is Online
|
Post by Khunanup on Oct 30, 2020 3:39:30 GMT
There should be no algorithm.
No systems level bureaucracy here please, because it always ends in tears (and is magnificently inefficient).
|
|
J.G.Harston
Lib Dem
Leave-voting Brexit-supporting Liberal Democrat
Posts: 14,808
|
Post by J.G.Harston on Oct 30, 2020 9:25:04 GMT
There should be no algorithm. No systems level bureaucracy here please, because it always ends in tears (and is magnificently inefficient). But "within +/- 10%" is an algorithm. "No crossing waterways" is an algorithm. "Prioritise local authority boundaries" is an algorithm. Even "don't make any decisions, do it all at random" is an algorithm. |
|
Khunanup
Lib Dem
Portsmouth Liberal Democrats
Posts: 12,039
Member is Online
|
Post by Khunanup on Oct 30, 2020 13:16:20 GMT
There should be no algorithm. No systems level bureaucracy here please, because it always ends in tears (and is magnificently inefficient). But "within +/- 10%" is an algorithm. "No crossing waterways" is an algorithm. "Prioritise local authority boundaries" is an algorithm. Even "don't make any decisions, do it all at random" is an algorithm. Not what's being discussed here though is it. We're talking an algorithm that's fee into a computer and spits out constituencies. |
|
J.G.Harston
Lib Dem
Leave-voting Brexit-supporting Liberal Democrat
Posts: 14,808
|
Post by J.G.Harston on Oct 30, 2020 19:43:11 GMT
But "within +/- 10%" is an algorithm. "No crossing waterways" is an algorithm. "Prioritise local authority boundaries" is an algorithm. Even "don't make any decisions, do it all at random" is an algorithm. Not what's being discussed here though is it. We're talking an algorithm that's fee into a computer and spits out constituencies. Whether an algorithm is fed into a computer or into a human brain, it will spit out the same results, and it's still an algorithm regardless of the processor that's processing it. "Remember the sums of single digits. Start at the rightmost column Using those pre-remembered sums, add the column of digits, write the rightmost digit of the total at the bottom, and if the total is more than 9 write the number of tens at the top of the next column to the left Move over to the next column to the left and repeat, including the carry over, until you run out of columns" ...is an algorithm. |
|
|
|
Post by Pete Whitehead on Oct 30, 2020 21:00:06 GMT
Not what's being discussed here though is it. We're talking an algorithm that's fee into a computer and spits out constituencies. Whether an algorithm is fed into a computer or into a human brain, it will spit out the same results, and it's still an algorithm regardless of the processor that's processing it. "Remember the sums of single digits. Start at the rightmost column Using those pre-remembered sums, add the column of digits, write the rightmost digit of the total at the bottom, and if the total is more than 9 write the number of tens at the top of the next column to the left Move over to the next column to the left and repeat, including the carry over, until you run out of columns" ...is an algorithm. I think the point is that this exercise should be regarded as an art as much as (or more than) a science |
|
|
|
Post by islington on Oct 31, 2020 8:24:51 GMT
I'd suggest it's not so much an art as a craft. We're striving for an outcome that is functional and utile, rather than something beautiful. Ingenuity, if we ever achieve it, is a virtue; but creativity wouldn't be.
|
|
|
|
Post by East Anglian Lefty on Oct 31, 2020 9:13:48 GMT
This is a bad idea. There are several examples in the two zombie reviews where leaving a constituency within quota caused some really bad knock-on effects. Everything else gives you a decent basis for an algorithm, though. Also it assumes that existing constituencies are “good” constituencies that ought to be kept. The reality is that that are numerous highly imperfect and undesirable existing constituencies that ought to be scrapped even if they are within quota at every review for the next 200 years. This is true, but the Boundary Commissions traditionally work on the assumption that it is not, so it would make sense for an algorithm trying to work out what they'd propose to implement something like this. |
|
YL
Non-Aligned
Either Labour leaning or Lib Dem leaning but not sure which
Posts: 4,915
|
Post by YL on Nov 1, 2020 8:59:36 GMT
I suspect a well-designed algorithm could give reasonable results as far as the "minimal change" and "respect local government boundaries" criteria are concerned, though decisions would have to be made about how to prioritise them, but I suspect it would always struggle with "local ties". The Electoral Calculus algorithm completely ignores this, with towns being split all over the place, and unlike some of its other flaws I'm not sure that solving this would be that easy.
You could, I suppose, tell it to try to keep wards together when they're currently in the same constituency and in the same civil parish (or partly so), or perhaps if they share part of their name (with compass points and the like being excluded!). But there are plenty of towns which aren't parished and don't appear in ward names: for example, neither of those would detect the existence of Halifax or Huddersfield. And I suspect there would be other oddities too.
|
|
|
|
Post by greenchristian on Nov 1, 2020 9:08:24 GMT
I suspect a well-designed algorithm could give reasonable results as far as the "minimal change" and "respect local government boundaries" criteria are concerned, though decisions would have to be made about how to prioritise them, but I suspect it would always struggle with "local ties". The Electoral Calculus algorithm completely ignores this, with towns being split all over the place, and unlike some of its other flaws I'm not sure that solving this would be that easy. You could, I suppose, tell it to try to keep wards together when they're currently in the same constituency and in the same civil parish (or partly so), or perhaps if they share part of their name (with compass points and the like being excluded!). But there are plenty of towns which aren't parished and don't appear in ward names: for example, neither of those would detect the existence of Halifax or Huddersfield. And I suspect there would be other oddities too. There are ways to tag wards with the information needed to identify if there are connections between them, and if they are part of the same town or city. That doesn't entirely solve the local ties issue, though, since it does require local knowledge to know which parts of towns and cities go together well or which town a rural ward looks to. |
|
|
|
Post by minionofmidas on Nov 1, 2020 9:55:30 GMT
I suspect a well-designed algorithm could give reasonable results as far as the "minimal change" and "respect local government boundaries" criteria are concerned, though decisions would have to be made about how to prioritise them, but I suspect it would always struggle with "local ties". The Electoral Calculus algorithm completely ignores this, with towns being split all over the place, and unlike some of its other flaws I'm not sure that solving this would be that easy. You could, I suppose, tell it to try to keep wards together when they're currently in the same constituency and in the same civil parish (or partly so), or perhaps if they share part of their name (with compass points and the like being excluded!). But there are plenty of towns which aren't parished and don't appear in ward names: for example, neither of those would detect the existence of Halifax or Huddersfield. And I suspect there would be other oddities too. you could always treat a contiguous unparished area within a district/ua as a town. Going to miss on some places and be no help at all in London, obviously, but would generally fix the issue. You could even use pre74 (or better pre60s) borough/ud boundaries for unparished areas. All of this is still quantifiable and can be dealt with by an algorithm. It's at "which of these suburban parishes ahould be included with the town and which with the rural area" and "we do need to split one town, but which one? And which area?" that a mathematical algorithm is of no help at all. It can be trained to use the area leading to the most compact boundary, or the least deviation, or the least disruption to current pattern, or even a complex tradeoff (gonna take an insane amount of computing power though) but it can't ever make a qualitative judgment on ties. |
|
