ASA Connect

     With the emergence of gerrymandering-squared in Texas, I want to make an important point.  We'll never have fair congressional districts until they are derived by algorithm, with no human intervention, depending only on population and location.  I have such an algorithm.  There are many others.  They all aim for reasonably shaped districts with approximately equal populations.  My algorithm clusters US Census block groups with well-defined populations and locations (longitude-latitude) into ideal districts subject to each state's apportionment of 435 House seats.

     It's worth noting that congressional apportionment used to be like the Wild West, as districting is now.  However, since 1941 we have had a legislated apportionment algorithm.  It's not perfect, but at least we have one.  After each decennial census, the Census Bureau runs the algorithm, announces the results, and there is barely a whimper, notwithstanding some questionable calls like Montana and Rhode Island each getting two seats in 2020 at the expense of New York and Ohio.  In any event, we need the same sort of setup for congressional districting, i.e., Census runs a legislated districting algorithm after each decennial census, they announce the results, and the case is closed for the next ten years.

     Finally, the House of Representatives is way too small.  It has had 435 seats for the past hundred years while US population has more than tripled.  I'm not suggesting tripling House seats, but some have suggested adding 150-160.  That in itself would enhance fair, proportional representation in the House, and in the Electoral College.

https://github.com/raagnew/IdealCongressionalDistricting

https://public.tableau.com/app/profile/bob.agnew/vizzes

https://www.nytimes.com/interactive/2018/11/09/opinion/expanded-house-representatives-size.html

https://www.raagnew.com/us-congressional-apportionments.html

Dr. Agnew writes "...derived by algorithm, with no human intervention..."

Speaking generally, not is this specific case only, algorithms have biases. The simple fact of using an algorithm does not preclude human impact. Examples include but are not limited to

• Biased training data, where past human decisions are used to label outcomes employed in training the algorithm. 
• Human impact on feature selection, where the features selected for the algorithm convery human preferences. 

Due to these factors, bias testing mitigation should be an important step in algorithm development. Simply using an algorithm does not automatically make a process free of human influence.

David 

     With the emergence of gerrymandering-squared in Texas, I want to make an important point.  We'll never have fair congressional districts until they are derived by algorithm, with no human intervention, depending only on population and location.  I have such an algorithm.  There are many others.  They all aim for reasonably shaped districts with approximately equal populations.  My algorithm clusters US Census block groups with well-defined populations and locations (longitude-latitude) into ideal districts subject to each state's apportionment of 435 House seats.

     It's worth noting that congressional apportionment used to be like the Wild West, as districting is now.  However, since 1941 we have had a legislated apportionment algorithm.  It's not perfect, but at least we have one.  After each decennial census, the Census Bureau runs the algorithm, announces the results, and there is barely a whimper, notwithstanding some questionable calls like Montana and Rhode Island each getting two seats in 2020 at the expense of New York and Ohio.  In any event, we need the same sort of setup for congressional districting, i.e., Census runs a legislated districting algorithm after each decennial census, they announce the results, and the case is closed for the next ten years.

     Finally, the House of Representatives is way too small.  It has had 435 seats for the past hundred years while US population has more than tripled.  I'm not suggesting tripling House seats, but some have suggested adding 150-160.  That in itself would enhance fair, proportional representation in the House, and in the Electoral College.

https://github.com/raagnew/IdealCongressionalDistricting

https://public.tableau.com/app/profile/bob.agnew/vizzes

https://www.nytimes.com/interactive/2018/11/09/opinion/expanded-house-representatives-size.html

https://www.raagnew.com/us-congressional-apportionments.html

     With the emergence of gerrymandering-squared in Texas, I want to make an important point.  We'll never have fair congressional districts until they are derived by algorithm, with no human intervention, depending only on population and location.  I have such an algorithm.  There are many others.  They all aim for reasonably shaped districts with approximately equal populations.  My algorithm clusters US Census block groups with well-defined populations and locations (longitude-latitude) into ideal districts subject to each state's apportionment of 435 House seats.

     It's worth noting that congressional apportionment used to be like the Wild West, as districting is now.  However, since 1941 we have had a legislated apportionment algorithm.  It's not perfect, but at least we have one.  After each decennial census, the Census Bureau runs the algorithm, announces the results, and there is barely a whimper, notwithstanding some questionable calls like Montana and Rhode Island each getting two seats in 2020 at the expense of New York and Ohio.  In any event, we need the same sort of setup for congressional districting, i.e., Census runs a legislated districting algorithm after each decennial census, they announce the results, and the case is closed for the next ten years.

     Finally, the House of Representatives is way too small.  It has had 435 seats for the past hundred years while US population has more than tripled.  I'm not suggesting tripling House seats, but some have suggested adding 150-160.  That in itself would enhance fair, proportional representation in the House, and in the Electoral College.

https://github.com/raagnew/IdealCongressionalDistricting

https://public.tableau.com/app/profile/bob.agnew/vizzes

https://www.nytimes.com/interactive/2018/11/09/opinion/expanded-house-representatives-size.html

https://www.raagnew.com/us-congressional-apportionments.html

Well said, Robert. However, the Electoral College will only be even approximately fair when States cast their votes in proportion to their popular vote.

     With the emergence of gerrymandering-squared in Texas, I want to make an important point.  We'll never have fair congressional districts until they are derived by algorithm, with no human intervention, depending only on population and location.  I have such an algorithm.  There are many others.  They all aim for reasonably shaped districts with approximately equal populations.  My algorithm clusters US Census block groups with well-defined populations and locations (longitude-latitude) into ideal districts subject to each state's apportionment of 435 House seats.

     It's worth noting that congressional apportionment used to be like the Wild West, as districting is now.  However, since 1941 we have had a legislated apportionment algorithm.  It's not perfect, but at least we have one.  After each decennial census, the Census Bureau runs the algorithm, announces the results, and there is barely a whimper, notwithstanding some questionable calls like Montana and Rhode Island each getting two seats in 2020 at the expense of New York and Ohio.  In any event, we need the same sort of setup for congressional districting, i.e., Census runs a legislated districting algorithm after each decennial census, they announce the results, and the case is closed for the next ten years.

     Finally, the House of Representatives is way too small.  It has had 435 seats for the past hundred years while US population has more than tripled.  I'm not suggesting tripling House seats, but some have suggested adding 150-160.  That in itself would enhance fair, proportional representation in the House, and in the Electoral College.

https://github.com/raagnew/IdealCongressionalDistricting

https://public.tableau.com/app/profile/bob.agnew/vizzes

https://www.nytimes.com/interactive/2018/11/09/opinion/expanded-house-representatives-size.html

https://www.raagnew.com/us-congressional-apportionments.html