Big Redistricting News Out Of PA!

[muon2]

Back in 2013 I looked at PA using DRA as the muon rules were first being fleshed out. With help from Torie, traininthedistance, and jimrtex, I came up with a balanced neutral plan. I couldn't find my drf file from back then, so I reconstructed it with a couple of tweaks to reflect the current version of the rules.



Chester is the only macrochopped county smaller than a CD, and only three other small counties have regular chops. No city/borough/township is chopped except Philly, and no ward within Philly is chopped. Obama won 11 of the 18 CDs in 2008. DRA population deviations and current PVIs for the CDs are:

CD 1: (-42) D+24
CD 2: (-497) D+39; BVAP 52.6%
CD 3: (+180) R+8
CD 4: (-1808) R+17
CD 5: (+1704) R+15
CD 6: (-691) R+3
CD 7: (-616) D+8
CD 8: (+1566) R+1
CD 9: (+1078) R+20
CD 10: (-1173) R+17
CD 11: (-491) D+0
CD 12: (+1919) R+13
CD 13: (-636) D+17
CD 14: (+921) D+10
CD 15: (-2359) R+0
CD 16: (+963) R+9
CD 17: (-1047) R+9
CD 18: (+1006) R+3

This might give some sense as to what a new neutral map might produce.

Edit: fixed the PVIs by using old voting districts

[muon2]

CD 1: (-42) D+22
CD 2: (-497) D+39; BVAP 52.6%
CD 3: (+180) R+5
CD 4: (-1808) R+5
CD 5: (+1704) R+16
CD 6: (-691) R+0
CD 7: (-616) D+8
CD 8: (+1566) R+2
CD 9: (+1078) R+15
CD 10: (-1173) R+20
CD 11: (-491) R+2
CD 12: (+1919) R+8
CD 13: (-636) D+17
CD 14: (+921) D+12
CD 15: (-2359) R+7
CD 16: (+963) R+13
CD 17: (-1047) D+0
CD 18: (+1006) D+2

2, 1, 13, 14, 7, 18, 17, 6, 8, 11, 3, 4, 15, 12, 16, 9, 5, 10
<more dem...................more gop>
Median seats are 8 and 11, R+2, pretty much in line with the state at large.

Again he is using a piecemeal map, with only about 1/3 of the precincts having 2012/2016 data - HE NEEDS TO USE THE SECOND ON THE LIST. Want to confirm? Open PA in DRA and then mouse around over precincts - you will easily see many missing ones. You need to pick 2010 voting districts, not 2010 voting districts State Updated.

This is a problem because DRA ads up PVI per precinct and averages them to produce a final result. If only a third of precincts have data - you districts will only reflect that third.

Unfortunately I was not aware that the precincts were incomplete. I know that precincts have changed in some locations so I thought the updated list would better reflect the current numbers. I also see that I can't change the precinct choice on a saved map, so I will have to recreate it, again.

You asked about Philly, and while I am recreating the map, I can tell you that the city is divided such that everything north of South and east of Broad except ward 61 is in CD 1. Everything south of South or west of Broad except for wards 9, 10, 22 and 50 is in CD 2. Those other wards not in CD 1 or 2 are in CD 13.

[muon2]

CD 1: (-42) D+24
CD 2: (-497) D+39; BVAP 52.6%
CD 3: (+180) R+8
CD 4: (-1808) R+17
CD 5: (+1704) R+15
CD 6: (-691) R+3
CD 7: (-616) D+8
CD 8: (+1566) R+1
CD 9: (+1078) R+20
CD 10: (-1173) R+17
CD 11: (-491) D+0
CD 12: (+1919) R+13
CD 13: (-636) D+17
CD 14: (+921) D+10
CD 15: (-2359) R+0
CD 16: (+963) R+9
CD 17: (-1047) R+9
CD 18: (+1006) R+3

2, 1, 13, 14, 7, 18, 17, 6, 8, 11, 3, 4, 15, 12, 16, 9, 5, 10
<more dem...................more gop>
Median seats are 8 and 11, R+2, pretty much in line with the state at large.

I edited the PVIs to match the correct VTD choice on DRA. Here's the updated sequence:

2, 1, 13, 14, 7, 11, 15, 8, 6, 18, 3, 16, 17, 12, 5, 10, 4, 9
Median (6, 18) R+3, which is a couple points more Pub than the state as a whole. However, if the votes from the VRA district are removed the rest of the state performs at R+3 which is the median without CD 2.

[muon2]

Or a new swing seat in a court-drawn map where Chester has a D+ PVI. What a time to be alive

Chester is interesting - in 2008/2012 the county was easily still to the right of the state. In 2016 the county swung very hard to to the left, and it looks to be staying that way. Democrats appear to have made strides in the row offices last week, and the county backed the failed Liberal bid at the Supreme court, even while the Liberal Judge lost Lakawanna, Erie, and Lehigh all went for the Conservative. For reference, the race was 52-47 Conservative Judge.

The weird thing is that when drawing fair districts in that region of PA, there is a Gordian Knot in the lines. You have the numbers for two districts (6/16) and a few extraneous pops to balance out the 13th or the 7th. However, the counties make things nice. Of Chester, Berks, and Lancaster, only two can get a district based around them. The third has to get brutally cut up between her neighbors. There is no way to avoid it. We have a map in this thread that cuts Chester, and a map that cuts Lancaster - and I always have liked to cut Berks. There is no real solution which is the better county to cut.

If two chops are needed and the chop is inevitably large, eg more than 5% of the district population, I find that the shapes can be best managed by putting both chops in one county. Metrics that look at chops sometimes build that in intentionally. For instance FL looks at how many counties are chopped, not how many chops are required and that forces all the chops into a few counties if applied rigorously.

[muon2]

Holy hell is that a lot of cut counties - I have drawn maps that only cut 6.

No it isn't, only 16 counties were split between two or more congressional districts, and with 3 being larger than a congressional district, that leaves but 13 counties split.
Also, I don't draw my maps with the sole objective of splitting as few counties as possible, I also take into account urban areas, statistical areas, cultural regions, existing districts and many other factors.

Counties are very important as political units, and except in New England probably the most important unit in the US. It is one of the most common items to protect for states that have rules against gerrymandering.

[muon2]

Holy hell is that a lot of cut counties - I have drawn maps that only cut 6.

No it isn't, only 16 counties were split between two or more congressional districts, and with 3 being larger than a congressional district, that leaves but 13 counties split.
Also, I don't draw my maps with the sole objective of splitting as few counties as possible, I also take into account urban areas, statistical areas, cultural regions, existing districts and many other factors.

Counties are very important as political units, and except in New England probably the most important unit in the US. It is one of the most common items to protect for states that have rules against gerrymandering.

Most certainly, I completely agree with you, however, as I said, I don't think the sole metric of a fair redistricting plan is that it splits as few counties as possible, yes that is a key metric, however it shouldn't be the only aim of redistricting.

The problem I have observed is that some criteria that sound good actually are quite subjective. These criteria are excellent covers for political gerrymandering. The best way to guarantee fair maps is to stick to measurable criteria.

For instance we developed measurable criteria based on Census stats to define urban metro areas (see the Urban County Cluster sticky thread). I haven't seen a good way to specify an item like cultural areas that isn't open to abuse by subjectivity.

[muon2]

Holy hell is that a lot of cut counties - I have drawn maps that only cut 6.

No it isn't, only 16 counties were split between two or more congressional districts, and with 3 being larger than a congressional district, that leaves but 13 counties split.
Also, I don't draw my maps with the sole objective of splitting as few counties as possible, I also take into account urban areas, statistical areas, cultural regions, existing districts and many other factors.

Counties are very important as political units, and except in New England probably the most important unit in the US. It is one of the most common items to protect for states that have rules against gerrymandering.

Most certainly, I completely agree with you, however, as I said, I don't think the sole metric of a fair redistricting plan is that it splits as few counties as possible, yes that is a key metric, however it shouldn't be the only aim of redistricting.

The problem I have observed is that some criteria that sound good actually are quite subjective. These criteria are excellent covers for political gerrymandering. The best way to guarantee fair maps is to stick to measurable criteria.

For instance we developed measurable criteria based on Census stats to define urban metro areas (see the Urban County Cluster sticky thread). I haven't seen a good way to specify an item like cultural areas that isn't open to abuse by subjectivity.

My approach to redistricting is a British one.
I have for a long time taken a significant interest in British Boundary Reviews and have made multiple submissions to reviews.
The British boundary reviews currently have only one requirement, and that is that constituencies must be within quota, other than that there are no steadfast rules, only guidelines and aims, yet despite this the British system works perfectly well, free of gerrymandering.
In Australia we use the British system. The Redistribution Committee do not aim, when conducting redistributions, to split as few local government areas as physically possible. Their aim is for minimum change, while simultaneously keeping communities of interest together and maintaining compact and logical districts.
There is no gerrymandering in the UK or Australia. In Australia we did have had the Playmander in SA and the Bjelkemander in Queensland, however neither of these were gerrymanders, they were egregious cases of Malapportionment. The electoral districts were not gerrymandered, they were still logical and compact, the problem was that rural districts were far smaller than urban districts, in some cases being only a tenth of the size of the urban districts.

Unfortunately, even some independent panels in the US have suffered from subjective biases. I'm a fan of the method IA uses where the criteria are well defined in statute and an independent body draws the map based on the criteria.

[muon2]

For instance we developed measurable criteria based on Census stats to define urban metro areas (see the Urban County Cluster sticky thread). I haven't seen a good way to specify an item like cultural areas that isn't open to abuse by subjectivity.

Communities of interest should be identified now for use following the 2020 Census. This would permit objective criteria to be proposed and evaluated. Otherwise you have people playing hocus pocus with the redistricting commission.

You must be logged in to read this quote.

Most of the census data that could be used is from the ACS. The Census Bureau will do special tabulations of the ACS for those willing to pay for it. Race or ethnicity is a social interest. I suspect they deliberately avoided it as an example.

Communities of interest overlap and constantly change. That is why you shouldn't just draw a map and use it just like that. You need to develop a plan, then consult with local communities, then improve or change the plan, then consult again, and so on.

For instance in the UK they develop then publish the Initial Proposals, then there is an Initial Consultation period, with written representations and public hearings, then there is a Secondary Consultation period, with written representations and comments with regards to the written representations and public hearings of the Initial Consultation period, then, using the community input the proposals are then revised, and then published as the Revised Proposals (we are currently here), then there is a Final Consultation period with written representations, and then finally the proposals are further revised and then published as the Final Report which is then brought before Parliament to become law.

Meanwhile in Australia the process is even more thorough, with the timetable being:

Electoral Commission directs commencement of redistribution by way of Notice in the Commonwealth Government Notices Gazette.
Electoral Commissioner determines current enrolment quota Electoral Commission appoints Redistribution Committee.
Production and checking of enrolment projections.
The Electoral Commissioner invites written suggestions to the redistribution from the public.
Suggestions available for public to make written comments on suggestions.
Redistribution Committee considers suggestions and comments on suggestions and develops a set of boundary proposals.
Production of Redistribution Committee’s report and maps showing proposed names and boundaries.
Redistribution Committee publishes and exhibits maps showing proposed boundaries and names and reasons for proposal. Public invited to make written objections to the proposed redistribution.
Objections available for public to make written comments on objections.
Augmented Electoral Commission considers objections and comments on objections. As part of these considerations, a public inquiry into objections may be held.
Announcement of augmented Electoral Commission’s proposed redistribution.
Final determination of names and boundaries of electoral divisions by notice published in the Gazette.
Augmented Electoral Commission’s report tabled in Parliament.
Augmented Electoral Commission’s report is made publicly available.

As jimrtex notes, the problem has often been identifying standards that justifies a community so as to achieve a political goal. In 2011 I sat though a great deal of public testimony of the kind that the UK and Oz require. I also watched the mapmakers cherry-pick which testimony to give weight, and then see them identify those favored groups.

Communities do change, but redistricting is once a decade at a very well-defined point in time. There's no reason not to quantify the communities  in advance of that set date.

[muon2]

Iowa is good by American standards but it isn't really that good compared to the rest of the world. In fact it's a great example of my point. The Des Moines urban area is split between all four congressional districts, while if you just split a single county then all of Des Moines could be united in one single congressional district. In the same vein Western Iowa is also split between two districts, when it could easily be united as one.

"urban area" has a quite specific meaning in the United States census. You may be using some entirely different meaning. What is the name of that county that you believe if it were split would unify Des Moines in a single congressional district.

If western Iowa were in a single district, as it was during the 2000s, what would the other three look like? Some of the others during the 2000s were ugly.

Because I am to lazy to draw my own this very second I'll just borrow a map from Dailykos:

Unfortunately for DK, IA has a very strict requirement to keep counties whole. The split county in this map does nothing to keep communities of interest whole. It does serve the political agenda of DK by packing the most conservative parts of the state into a single CD. This is what I mean by my concern over the lack of firm criteria; using soft standards invites subtle gerrymandering.

[muon2]

My Pennsylvania redistricting plan.

Apart from you did to Philly and excessive county splits, it's not too bad overall. PA-01 and PA-02 should both be contained within Philadelphia County. As for having two different districts pick up the remnants, I can see the merits to such a plan, but I'm not sure how others view that. Pulling PA-01 forces a counter-clockwise rotation that pushes a lot of PA-13 out of Philly and into Montco which in turn pushes PA-06 into Chester County and therefore parts of PA-07 out of Chester (as it entirely takes in Delco). I do like your PA-12 though. It's not something I see in Pennsylvania maps very often, but it's a very logical district.

For a plan with N districts, there should not be more than N-1 county chops without a clearly definable reason. In a state like PA with lots of well defined cities, towns and Philly wards (identified by the names of the vote districts in DRA) there is no reason other than trying for exact population equality to chop any of them. Doing so divides a community of interest (less so with some of the wards). There are enough of those county subdivisions that it should always be possible to get within a relatively small deviation, probably under 1000 and certainly less than 0.5%.

The five Philly area counties plus Berks and Lancaster are a near perfect fit for 7 CDs with 2010 data. There should be no more than 6 chops in those counties, and this plan does that (3 chops in Philly, 1 each in Montco, Delco, and Chester. I don't know if towns and wards were preserved whole though, but I presume some minor adjustments could be made to respect those lines.

The county splits out west are a different matter. I know ASV said he thought the communities of interest are more important, but I'm not how that applies there. The western boundary of Cambria and Somerset counties is a significant mountain ridge that is largely protected by state forests. It seems that this plan tries to use the next, lower ridge to the west as the line separating communities of interest, but that doesn't comport with the actually lifestyle patterns in that part of the state from my visits. Those county lines in this case do a pretty good job of marking the separation between regions.

[muon2]

For a plan with N districts, there should not be more than N-1 county chops without a clearly definable reason. In a state like PA with lots of well defined cities, towns and Philly wards (identified by the names of the vote districts in DRA) there is no reason other than trying for exact population equality to chop any of them. Doing so divides a community of interest (less so with some of the wards). There are enough of those county subdivisions that it should always be possible to get within a relatively small deviation, probably under 1000 and certainly less than 0.5%.

The five Philly area counties plus Berks and Lancaster are a near perfect fit for 7 CDs with 2010 data. There should be no more than 6 chops in those counties, and this plan does that (3 chops in Philly, 1 each in Montco, Delco, and Chester. I don't know if towns and wards were preserved whole though, but I presume some minor adjustments could be made to respect those lines.

The county splits out west are a different matter. I know ASV said he thought the communities of interest are more important, but I'm not how that applies there. The western boundary of Cambria and Somerset counties is a significant mountain ridge that is largely protected by state forests. It seems that this plan tries to use the next, lower ridge to the west as the line separating communities of interest, but that doesn't comport with the actually lifestyle patterns in that part of the state from my visits. Those county lines in this case do a pretty good job of marking the separation between regions.

I think that PA-01 is a violation of any fair and reasonable map. Philadelphia has the population for at least two districts. I think that should require, in any fair map, that the city/county should have two districts within its boundaries. That does mean an additional chop of Chester, however I believe the preceding criteria should take precedence. And furthermore, I think the other problem with SEPA in that map is that PA-13 takes in two non-contiguous parts of Philadelphia.

Like I said, I don't think it's an overall unreasonable map apart from what I mentioned above. It just needs some cleaning up in terms of county splits. I would work on it myself, but I don't want to start from scratch. To start, I would swap numbering between PA-11 and PA-17 and I would make the new PA-11 a Lackawanna/Luzerne/Monroe district.

There is a school of thought that says one redistricting principle should be: In any county greater than the population for a whole district there should be a number of districts completely within the county equal to the population of the county divided by the quota for one district, rounded down to the nearest whole number. In PA that would require 2 CDs in wholly in Philly, 1 wholly in Montco, and 1 wholly in Allegheny.

We debated the merits of this principle on threads on this board back in 2012-13. In the end the majority view here was that a count of the total number of chops was a more important principle than considering how the fragments of the chops nested within counties. I shared that view in part from my experience in the 2011 OH Redistricting Competition, which had nesting requirements for legislative districts and rules that created a preference for wholly-contained districts in the congressional plan. I thought it created unneeded skews to other measures of the plan.

[muon2]

For a plan with N districts, there should not be more than N-1 county chops without a clearly definable reason. In a state like PA with lots of well defined cities, towns and Philly wards (identified by the names of the vote districts in DRA) there is no reason other than trying for exact population equality to chop any of them. Doing so divides a community of interest (less so with some of the wards). There are enough of those county subdivisions that it should always be possible to get within a relatively small deviation, probably under 1000 and certainly less than 0.5%.

The five Philly area counties plus Berks and Lancaster are a near perfect fit for 7 CDs with 2010 data. There should be no more than 6 chops in those counties, and this plan does that (3 chops in Philly, 1 each in Montco, Delco, and Chester. I don't know if towns and wards were preserved whole though, but I presume some minor adjustments could be made to respect those lines.

The county splits out west are a different matter. I know ASV said he thought the communities of interest are more important, but I'm not how that applies there. The western boundary of Cambria and Somerset counties is a significant mountain ridge that is largely protected by state forests. It seems that this plan tries to use the next, lower ridge to the west as the line separating communities of interest, but that doesn't comport with the actually lifestyle patterns in that part of the state from my visits. Those county lines in this case do a pretty good job of marking the separation between regions.

I think that PA-01 is a violation of any fair and reasonable map. Philadelphia has the population for at least two districts. I think that should require, in any fair map, that the city/county should have two districts within its boundaries. That does mean an additional chop of Chester, however I believe the preceding criteria should take precedence. And furthermore, I think the other problem with SEPA in that map is that PA-13 takes in two non-contiguous parts of Philadelphia.

Like I said, I don't think it's an overall unreasonable map apart from what I mentioned above. It just needs some cleaning up in terms of county splits. I would work on it myself, but I don't want to start from scratch. To start, I would swap numbering between PA-11 and PA-17 and I would make the new PA-11 a Lackawanna/Luzerne/Monroe district.

There is a school of thought that says one redistricting principle should be: In any county greater than the population for a whole district there should be a number of districts completely within the county equal to the population of the county divided by the quota for one district, rounded down to the nearest whole number. In PA that would require 2 CDs in wholly in Philly, 1 wholly in Montco, and 1 wholly in Allegheny.

We debated the merits of this principle on threads on this board back in 2012-13. In the end the majority view here was that a count of the total number of chops was a more important principle than considering how the fragments of the chops nested within counties. I shared that view in part from my experience in the 2011 OH Redistricting Competition, which had nesting requirements for legislative districts and rules that created a preference for wholly-contained districts in the congressional plan. I thought it created unneeded skews to other measures of the plan.

PA-1 and 2 could be entirely within Philly County, however I should have made it clearer that the arrangement exists solely to allow for 2 Black Majority congressional districts in Philly.

The 2010 Census had 665K blacks (including Hispanic blacks) in the city of Philadelphia. The 2010 Census set a quota of 706K per CD. The black population in Philly isn't large enough to be the majority in 2 CDs.

[muon2]

There is a school of thought that says one redistricting principle should be: In any county greater than the population for a whole district there should be a number of districts completely within the county equal to the population of the county divided by the quota for one district, rounded down to the nearest whole number. In PA that would require 2 CDs in wholly in Philly, 1 wholly in Montco, and 1 wholly in Allegheny.

We debated the merits of this principle on threads on this board back in 2012-13. In the end the majority view here was that a count of the total number of chops was a more important principle than considering how the fragments of the chops nested within counties. I shared that view in part from my experience in the 2011 OH Redistricting Competition, which had nesting requirements for legislative districts and rules that created a preference for wholly-contained districts in the congressional plan. I thought it created unneeded skews to other measures of the plan.

I'm not sure I necessarily agree, but I admit I haven't fully explored the issue. I have always considered it a high priority when I've drawn maps. For PA, I have always followed the requirements you noted of that criteria. For OH, that would mean one district each entirely contained within each of Cuyahoga, Franklin, and Hamilton Counties. I don't really see a strong argument against having such a standard.

In this case, however, the two criteria are not in conflict (unless chopping a county an additional time violates certain criteria). In the map we're discussing, the districts basically move counter-clockwise. PA-01 moves out of Delco, eliminating the chop of that county. And ultimately, through a number of changes, PA-06 moves into Chester County, which adds a chop.

I do have one question about your criteria. Do non-contiguous chops from one district count as one or more (such as PA-13 in the map above)?

Here's the definitions worked out on the subject of chops (edited to reflect current use). Chops involving more than one fragment of a district are still just a chop. However, chops involving multiple fragments will typically have higher erosity, as each fragment counts separately for connections. The creates a small disincentive for fragmented chops, but that fragmentation might preserve towns or wards from additional chops.

Definition: Chop. A single chop is the division of a geographic unit between two districts. A second chop divides the unit between three districts. In general the number of chops is equal to the number of districts in that unit less one.

Definition: Chop size. In units with a single chop, the size of a chop is the population of the smaller district within the unit. For districts with more than one chop, chop sizes are measured in order from the smallest populated district in the unit up to but not including the district with the largest population in the unit.

Definition: Macrochop. A macrochop is one or more chops in a county that has a total size in excess of 5.0% of the quota. When a macrochop of a county occurs, the subunits of the county must be considered as if they were units as well. Note that macrochops may only apply to counties with a population of more than 10% of the quota, and must apply to counties with more than 105% of the quota.

Item 10: CHOP measures the integrity of geographic units in a plan. The CHOP score is the total of all county chops. In counties with a macrochop, chops of county subunits are added to the CHOP score, however VTDs that span county subdivisions do not increase the CHOP score.

[muon2]

For a plan with N districts, there should not be more than N-1 county chops without a clearly definable reason. In a state like PA with lots of well defined cities, towns and Philly wards (identified by the names of the vote districts in DRA) there is no reason other than trying for exact population equality to chop any of them. Doing so divides a community of interest (less so with some of the wards). There are enough of those county subdivisions that it should always be possible to get within a relatively small deviation, probably under 1000 and certainly less than 0.5%.

The five Philly area counties plus Berks and Lancaster are a near perfect fit for 7 CDs with 2010 data. There should be no more than 6 chops in those counties, and this plan does that (3 chops in Philly, 1 each in Montco, Delco, and Chester. I don't know if towns and wards were preserved whole though, but I presume some minor adjustments could be made to respect those lines.

The county splits out west are a different matter. I know ASV said he thought the communities of interest are more important, but I'm not how that applies there. The western boundary of Cambria and Somerset counties is a significant mountain ridge that is largely protected by state forests. It seems that this plan tries to use the next, lower ridge to the west as the line separating communities of interest, but that doesn't comport with the actually lifestyle patterns in that part of the state from my visits. Those county lines in this case do a pretty good job of marking the separation between regions.

I think that PA-01 is a violation of any fair and reasonable map. Philadelphia has the population for at least two districts. I think that should require, in any fair map, that the city/county should have two districts within its boundaries. That does mean an additional chop of Chester, however I believe the preceding criteria should take precedence. And furthermore, I think the other problem with SEPA in that map is that PA-13 takes in two non-contiguous parts of Philadelphia.

Like I said, I don't think it's an overall unreasonable map apart from what I mentioned above. It just needs some cleaning up in terms of county splits. I would work on it myself, but I don't want to start from scratch. To start, I would swap numbering between PA-11 and PA-17 and I would make the new PA-11 a Lackawanna/Luzerne/Monroe district.

There is a school of thought that says one redistricting principle should be: In any county greater than the population for a whole district there should be a number of districts completely within the county equal to the population of the county divided by the quota for one district, rounded down to the nearest whole number. In PA that would require 2 CDs in wholly in Philly, 1 wholly in Montco, and 1 wholly in Allegheny.

We debated the merits of this principle on threads on this board back in 2012-13. In the end the majority view here was that a count of the total number of chops was a more important principle than considering how the fragments of the chops nested within counties. I shared that view in part from my experience in the 2011 OH Redistricting Competition, which had nesting requirements for legislative districts and rules that created a preference for wholly-contained districts in the congressional plan. I thought it created unneeded skews to other measures of the plan.

PA-1 and 2 could be entirely within Philly County, however I should have made it clearer that the arrangement exists solely to allow for 2 Black Majority congressional districts in Philly.

The 2010 Census had 665K blacks (including Hispanic blacks) in the city of Philadelphia. The 2010 Census set a quota of 706K per CD. The black population in Philly isn't large enough to be the majority in 2 CDs.

That is why the 1st congressional district ventures into Delaware County, so that, on DRA figures:

District 1 D+36.23 - 86.0 - 13.5 - 50.5 African American
District 2 D+40.09 - 89.4 - 10.2 - 50.1 African American

Is that black population or black VAP? The latter is what counts in determining if it is majority black.

[muon2]

For a plan with N districts, there should not be more than N-1 county chops without a clearly definable reason. In a state like PA with lots of well defined cities, towns and Philly wards (identified by the names of the vote districts in DRA) there is no reason other than trying for exact population equality to chop any of them. Doing so divides a community of interest (less so with some of the wards). There are enough of those county subdivisions that it should always be possible to get within a relatively small deviation, probably under 1000 and certainly less than 0.5%.

The five Philly area counties plus Berks and Lancaster are a near perfect fit for 7 CDs with 2010 data. There should be no more than 6 chops in those counties, and this plan does that (3 chops in Philly, 1 each in Montco, Delco, and Chester. I don't know if towns and wards were preserved whole though, but I presume some minor adjustments could be made to respect those lines.

The county splits out west are a different matter. I know ASV said he thought the communities of interest are more important, but I'm not how that applies there. The western boundary of Cambria and Somerset counties is a significant mountain ridge that is largely protected by state forests. It seems that this plan tries to use the next, lower ridge to the west as the line separating communities of interest, but that doesn't comport with the actually lifestyle patterns in that part of the state from my visits. Those county lines in this case do a pretty good job of marking the separation between regions.

I think that PA-01 is a violation of any fair and reasonable map. Philadelphia has the population for at least two districts. I think that should require, in any fair map, that the city/county should have two districts within its boundaries. That does mean an additional chop of Chester, however I believe the preceding criteria should take precedence. And furthermore, I think the other problem with SEPA in that map is that PA-13 takes in two non-contiguous parts of Philadelphia.

Like I said, I don't think it's an overall unreasonable map apart from what I mentioned above. It just needs some cleaning up in terms of county splits. I would work on it myself, but I don't want to start from scratch. To start, I would swap numbering between PA-11 and PA-17 and I would make the new PA-11 a Lackawanna/Luzerne/Monroe district.

There is a school of thought that says one redistricting principle should be: In any county greater than the population for a whole district there should be a number of districts completely within the county equal to the population of the county divided by the quota for one district, rounded down to the nearest whole number. In PA that would require 2 CDs in wholly in Philly, 1 wholly in Montco, and 1 wholly in Allegheny.

We debated the merits of this principle on threads on this board back in 2012-13. In the end the majority view here was that a count of the total number of chops was a more important principle than considering how the fragments of the chops nested within counties. I shared that view in part from my experience in the 2011 OH Redistricting Competition, which had nesting requirements for legislative districts and rules that created a preference for wholly-contained districts in the congressional plan. I thought it created unneeded skews to other measures of the plan.

PA-1 and 2 could be entirely within Philly County, however I should have made it clearer that the arrangement exists solely to allow for 2 Black Majority congressional districts in Philly.

The 2010 Census had 665K blacks (including Hispanic blacks) in the city of Philadelphia. The 2010 Census set a quota of 706K per CD. The black population in Philly isn't large enough to be the majority in 2 CDs.

That is why the 1st congressional district ventures into Delaware County, so that, on DRA figures:

District 1 D+36.23 - 86.0 - 13.5 - 50.5 African American
District 2 D+40.09 - 89.4 - 10.2 - 50.1 African American

Is that black population or black VAP? The latter is what counts in determining if it is majority black.

It's whatever DRA gives as African-American

It gives two different boxes for demographics. The upper box is for the whole population. Back around 2010, I pointed out to DRA that the voting age population (VAP) is what the courts look at to determine the minority strength, and they added a second box below the first. It's the percentages in the second box that matter.

[muon2]

It has to do with how you count divisions of geographic units and what is the impact of a large county within its metro area. During the formulation of the definition of chop there were some you agreed with you. Even the OH competition used language that you'd like better (edited for clarity):

A fragment is a contiguous part of a district within a county that is not the entire county or the entire district. The Fragment score is the total count of all fragments.

During the competition I found that this had a defect in the way it scored small counties compared to large counties (those larger than a district). It had a discontinuity when a county was first chopped, such that small counties either had 0 or 2+ fragments, but never 1 fragment. However large counties could never have less than 1 fragment. This created opportunities to pack chops in the large counties.

FL used a different method of counting, and only counted the number of counties that had more than 1 district. This also served to force more fragments into the large counties.

One other problem with forcing the maximum districts into a county has to do with metro areas, and large counties are generally part of a multi-county metro area. Counties in a metro area tend to be larger than average for the state, and the lack of smaller counties make it harder to come up with county combinations that equal a whole number of districts. Forcing districts into a county compounds this problem and increases the likelihood that more chops are needed in the counties around the large central county. This is a bias in favor of the central county integrity at the expense of its neighbors.

My solution is to treat all counties exactly the same, regardless of their size. I do that by counting chops, not the pieces created by the chops.