Let's take a look at Minnesota, a state with ~5.7 million people, and 8 congressional districts. We need to create 8 congressional districts with approximately 700,000 people within each, using existing county lines. Hennepin County, which contains the city of Minneapolis, has about 1,260,000 people, which is 71% more than their appropriate share. Clearly, we can't use Hennepin County as an "already drawn" line as it is, which kills the "use counties" plan immediately, but let's continue.

Ramsey County, the second most populous, has 535,000 people. Dakota County, 443,000. Anoka County, 369,000. Washington County, 275,000. Together, these 5 counties comprise 50% of the State's population. All 5 of these are relatively dense, and surround the Twin Cities area. We need to split these 5 counties into 4 congressional districts, to appropriately represent their share of the congressional districts. Hennepin is out since we obviously need to give them 1 seat on their own. After that, we need 7 districts of 634,000, 4 of which should come from Ramsey, Dakota, Anoka, and Washington counties. The easiest way to get close to this is to give Ramsey and Dakota counties their own seat, and to combine Anoka and Washington Counties into our final district. Conveniently, these two are geographically adjacent, so we don't even have to draw any weird districts with physical separations.

Now. The rest of the state is WAY less dense, so getting those districts to be roughly proportional for our 4 remaining districts is probably not going to be that bad, and I'm going to set them aside. Jumping back to our metropolitan area, we have Hennepin's district that represents 1,260,000 people, and Ramsey County's district which represents 535,000. So, the residents of Hennepin County have approximately half as much representation as the residents of Ramsey County, even though these counties are adjacent to one another, and each contains one of the two Twin Cities.

If we do allow some level of mix-and-matching some portion of certain counties, then presumably we use some other criteria in order to draw the lines. We can come up with metrics we think are reasonably fair, but ultimately it's all subjective at the end of the day. And, when we look at the actual congressional map for Minnesota, we find that it kinda looks like balancing these edge cases is already what the map looks like.

I'm not going to claim that this is a perfect, unbiased congressional map. I'm guessing many very reasonable people would disagree. But it's meant to illustrate that using previous arbitrary lines doesn't really buy us any better of results, since those lines weren't designed to facilitate political representation. https://en.wikipedia.org/wiki/Minnesota%27s_congressional_districts#/media/File:MN_2022_congressional_districts.jpg