This is one the best articles I have read on gerrymandering regarding its political import, and of course it is by one of the most articulate mathematicians, Jordan Ellenberg:

“Fixing partisan gerrymandering requires some technical calculations. That’s why we filed a mathematicians’ brief to better define the problem—and the solution.”

See Gerrymandering at SCOTUS. (You will have to read the article to understand the picture.)

(Updates 4/8/2019, 6/27/2019, 8/27/2022) 

(Update 4/8/2019) Feedback on this article from a friend suggested that it still was not that clear what there was about gerrymandering that one could say did not occur in Massachusetts but did occur in North Carolina. Apparently modeling Massachusetts without trying to gerrymander resulted in more or less the same outcome that currently exists in the state. Whereas modeling North Caroline without gerrymandering yielded very different results from the current situation, a situation that showed up in the gerrymandered models. What was omitted was an explanation of how one would perform these models with or without gerrymandering, that is, what are the characteristics one would employ to gerrymander and what are the ones used for a fair subdivision? Looks like that will require some follow-up. But the article is still entertaining to read as is.

(Update 6/27//2019) No longer a joke.  Ignorance has its consequences: (HuffPost)  “In a 5-4 decision divided along ideological lines, the court’s conservative justices, led by Chief Justice John Roberts, effectively allowed partisan gerrymandering or redistricting to move forward. Roberts’ majority opinion argued that the high court should not weigh in on the issue and instructed lower courts to dismiss legal challenges in North Carolina and Maryland.”

(Update 8/27/2022) Jordan Ellenberg’s tweet points to a good article in the Washington Post on gerrymandering, involving the application of Monte Carlo simulations.