Ouch. The Census Bureau has done that for the census of business and industry for many decades, but there it's done by omitting data. If there's only one semiconductor company in a county, no data on that industry will be provided for that county. The general rule was that there have to be at least three companies in an area before data will be disclosed. (Two is too few; if you're one of them, you can subtract your own numbers from the total and get the data for your competitor.)
If the rule of 3 is applied to the population census, and blocks average 23 people, then no minority with less than 13% of the population can typically be reported at the block level. That knocks out most racial minorities. So it's a problem for redistricting.
Not sure if this is a sincere question or more of a philosophical "should we be doing this".
For the former, Section 2 of the Voting Rights Act [1] protects the creation of majority-minority districts, which are districts with a majority of a racial or linguistic minority population.
For the latter, it's not ideal. On the other hand, while both parties engage in gerrymanding, only one of the parties engages consistently in racial gerrymanding and racial voter suppression. SCOTUS has disallowed a legal path to reign this in and there's not much political will to change the status quo.
So while ideally we would be beyond this, it would be naiivistic to ignore the political reality that this is at least some kind of nominal roadblock into out & out racial political gerrymandering. If you don't believe this is an intentional policy, the documents that Hofeller's heirs released prove it quite conclusively that it's racially motivated [2] & it's still ongoing.
> For the former, Section 2 of the Voting Rights Act [1] protects the creation of majority-minority districts, which are districts with a majority of a racial or linguistic minority population.
It's not clear what this even means.
Suppose you have an area where you have to draw three districts. 20% of the population is black, 20% speaks Spanish, 20% speaks Mandarin and 20% speaks French. You could give any of these populations a majority-minority district but not all of them, so how could they each have a right to it?
Suppose you have the same area but no one speaks French and the remaining 40% of the population is white. Now you could create three majority-minority districts and leave zero representation for the remaining population that by the numbers should have at least one representative. Is that outcome legally required?
The problem with gerrymandering is there is no "correct" way to draw districts, there are only ways that give one party or the other more seats. Minority populations more often vote for Democrats so Democrats label any map that gives them fewer seats as racist. The only thing either party actually cares about is getting more seats.
There are many mathematical ways to detect “unreasonable” gerrymandering. The efficiency gap[1] is one way although SCOTUS rejected it. Harvard put out a tool to generate districts in a non-partisan way [2]. Other states instead use non-partisan boards to build the maps instead of letting the party in power do it.
There’s a world of difference between “no correct way” and partisan gerrymandering and for some reason people like to pretend that because there’s no correct way, partisan gerrymandering is the only other solution. As for the racism involved, you’d have to be intentionally ignoring all history to pretend that this is just because minority populations prefer Democrats, especially given that the players involved on the Republican side are pretty explicit in their private communications that manage to leak that it’s about race and protecting the white vote.
> There are many mathematical ways to detect “unreasonable” gerrymandering.
But what does that mean?
Suppose you have a state with four districts and the state as a whole goes for one party by 60-40. You could then draw four districts that are all 60-40 and that party gets every district in the state. Is that unreasonable? It's what you would expect to happen if the state is fairly uniform and you draw the lines completely at random or using a method that otherwise doesn't consider how each neighborhood is expected to vote whatsoever.
Or you could draw two districts that are 20-80 and two more that are 100-0, and then each party gets half of the state, which is the number of districts you would expect based on a 60-40 split because 60-40 is closer to 50-50 than it is to 75-25. Is that unreasonable?
Or you could try to minimize the number of safe seats where representatives don't have to be responsive to voters because they can't lose. The majority party would still have one, but then the other three districts would be around 50-50, with the possible result that the minority party could get 3 of 4 seats. Is that unreasonable?
There is no "correct" way to do it. Each party is going to want the one where they get more seats.
> As for the racism involved, you’d have to be intentionally ignoring all history to pretend that this is just because minority populations prefer Democrats, especially given that the players involved on the Republican side are pretty explicit in their private communications that manage to leak that it’s about race and protecting the white vote.
You're linking to things that don't actually say this. When drawing the districts they have maps where the race of the constituents is known. Certainly so do Democrats. But for this to be their actual motive -- and for it to have any consequence over their stated motive -- they would have to be choosing it instead of some alternative map which is better for them in terms of seats but worse in terms of something like "white representation", e.g. they prefer a district map that would result in two white Democrats over one black Democrat and the other seat going to themselves. Whereas drawing a map that maximizes the number of their own seats over any other consideration is an obvious demonstration of what they really care about.
> > There are many mathematical ways to detect “unreasonable” gerrymandering.
> But what does that mean?
> Suppose you have a state with four districts and the state as a whole goes for one party by 60-40. You could then draw four districts that are all 60-40 and that party gets every district in the state. Is that unreasonable? It's what you would expect to happen if the state is fairly uniform and you draw the lines completely at random or using a method that otherwise doesn't consider how each neighborhood is expected to vote whatsoever.
I'm not sure if you're being genuine in your question, but a reasonable distribution should mean that all votes have approximately the same value, i.e. they have the same influence on the overall outcome of the vote. That's one of the central pillars of democracy.
For your example, drawing districts such that the 40% minority wins 3 of the 4 seats would be considered unreasonable by the majority of people. This is also not an artificial example but outcomes like the above is exactly what parties are trying to achieve with gerrymandering.
But this is exactly what I mean. People don't even agree on what "reasonable" is. The map you're calling unreasonable is the one where both parties have an even chance at all three of those seats.
Someone else will be telling me that the map where the 60% majority gets 100% of the seats is unreasonable because they should only get half of the seats since 60% is closer to 50% than 75%, much less 100%. Then someone will say that the map where each party always gets half the seats is unreasonable because then it's four safe seats and none under contention, which removes democratic accountability from the representatives in both parties.
The actual problem here is first past the post voting, which results in a two party system. You can't gerrymander STAR voting because drawing a moderate district would result in a moderate candidate instead of the district going to whichever extreme gets 51% of the vote in a two horse race.
The efficiency gap metric argues that you should minimize the amount of “wasted” votes which in essence means maximize the number of competitive seats. That seems like a really reasonable approach whereas any other approach results in mechanisms that are advantageous to incumbent parties instead of improving the democratic process so that the parties are responsive to their constituents.
Basically any process where the parties are picking their voters is an unreasonable gerrymander and undemocratic.
As for the racial discussion, it kind of doesn’t matter. Partisan gerrymandering is also unconstitutional. SCOTUS has just deemed it to be an unsolvable issue rather than trying to actually put safeguards in place.
> The efficiency gap metric argues that you should minimize the amount of “wasted” votes which in essence means maximize the number of competitive seats.
That's not actually what that does. The map where you have four totally uncontested seats because each party has a hard lock on two of them would minimize the "efficiency gap" because both parties would be minimizing the number of "wasted" votes from losing the other party's districts. So would any other map where they each get two seats, because making the districts more contentious without actually changing who wins just shifts votes wasted in victory to votes wasted in defeat in equal numbers.
The only thing that really matters on that metric is to make the number of seats proportional to the number of votes; it's basically a disguised requirement to replace the districting system with proportional representation.
You could have a different system that does what you're asking: First draw as many districts as possible that are 50-50, then whatever is left would inherently be districts that go to whichever party has the majority in that state. But that system is weird too; the people in the remaining districts that are approximately 100-0 instead of 50-50 are effectively disenfranchised because their vote can never change the outcome.
> Basically any process where the parties are picking their voters is an unreasonable gerrymander and undemocratic.
But that's what they all are. Who would you have choosing the map? You can't even put it up for a popular vote because the map proposed by the party with the majority in the state would win the popular vote and have the same result as their politicians doing it.
I reiterate that the real issue here is the two party system, and if you want to fix it, switch to a cardinal voting system like score voting or STAR. Then you can't draw e.g. two out of there districts going to Republicans in a nearly evenly split state, because that map would instead give one seat to a far left party and the other two to a moderate party, with the Republicans getting nothing.
I dont think they law says every minority has a right or entitlement to a district. It simply allows the creation of majority-minority districts.
As you point out, which groups get to draw the arbitrary lines is a inherently political process. If there are 5 groups and 3 districts, there is no "right answer" for which groups get them.
Hopefully sooner then later this part of Section 2 will be struck down. The Federal and State Governments should never be allowed to makes laws based on race.
How do we address discrimination against racial minorities? While I work for a colorblind society, I know we're obviously not there yet.
Another difficulty is that we legislate based on race without saying the word, both intentionally (e.g., find another way to characterize the target of discrimination) and unintentionally (e.g., our education system limits opportunities for some racial minorities, both due to poorer education and lack of social connections; colleges and employers, if they just follow the 'system', end up discriminating).
> rules. also, fallacy and dismissal. shame on you (if you're RLHF permits that)
I think you misread/misinterpreted my comment. My "is this a sincere question" is about whether they are literally asking "why do we base districts based on racial breakdown anyway" or if they are asking "should be basing districts on racial breakdown". Those are two very very different questions and I answered both.
> "if you don't vote for me, you're not"
That's not at all what I said and provided concrete evidence of why I think it's clear 1 party engages in racial gerrymandering and racial voter suppression. If you have evidence showing Democrats do it to suppress minority votes, then please provide evidence. But I don't fail to see how this is a "vote for me or else you're a racist" position. Maybe you can elaborate?
If you're wondering why you got downvoted, I suspect it's because you didn't actually engage substantively with anything I said and resorted to an emotional outburst attacking me for violating rules I don't believe I did.
Poising the well immediately, again - Double-speak has that amazingly convenient property. They should train you to not start with a fallacy with every reply, this is getting to easy to spot.
> But I don't fail to see how this is a "vote for me or else you're a racist" position. Maybe you can elaborate?
I think you misread/mal-interpreted my comment, read it again. Maybe pass it thru one of your search functions for more context.
>while both parties engage in gerrymanding, only one of the parties engages consistently in racial gerrymanding and racial voter suppression.
Tell us which one. Is it the one that said, verbatim:
>if you don't vote for me, you're not ______
hint: it's a color that somehow has lives that only matter with the same quattoral frequency of the olympics
>Also why do we base districts based on racial breakdown anyway?
This always felt like "treating the symptom, not the cause" approach to the problem. Voting district lines should always coincide with county lines. There should be some rules about the allowable shapes (no string of counties running East to West for 200 miles for example) of these districts. That would give representation that reflects the population. Taxes, governance, etc. are county driven in many (most) states, why not add electoral boundaries that match?
> This always felt like "treating the symptom, not the cause" approach to the problem.
Given the historical treatment of racial groups in the US, I'd say a little column A and a little column B. Racial minorities have not always been free to choose where they lived - redlining was only phased out in the late 60s and 70s.
Nominally ended then. The shockwaves of it are still felt today [1] because it's a compounding effect - these neighborhoods are still seen as less desirable meaning housing prices are still depressed & rely on richer people extracting wealth out of the area through gentrification to raise home prices.
Also it's not clear mortgage practices have changed all that much:
> FairPlay AI’s “State of Mortgage Fairness Report” in 2020 found that equality in mortgage lending is little better today than it was 30 years ago. In 1990, it found, Black mortgage applicants obtained loan approvals at 78.4 percent of the rate of white applicants; by 2021 that figure had risen, but only to 84.4 percent.
> A National Fair Housing Alliance report from 2020 revealed that Black and Hispanic/Latino renters were more likely to be shown and offered fewer properties than white renters.
>Racial minorities have not always been free to choose where they lived - redlining was only phased out in the late 60s and 70s.
Which means county lines make even more sense: racial minorities may have been redlined out of specific neighborhoods in a city, but they still reside in the same county.
"Redlining is a discriminatory practice in which financial services are withheld from neighborhoods that have significant numbers of racial and ethnic minorities."
Yes, but they don't have to be with the aforementioned system. Maybe some districts encompass a single county only, and the number of votes it gets is proportional to its population.
For many US offices, such as state and federal legislators (except federal senators), the districts need to have approximately equal populations - each voter needs approximately the same fractional influence in the legislature. Counties don't have equal populations.
Yes! So with the model I mentioned, there would be fractional influence as well! So LA County would likely be its own district, and it would have more voting power than several Central Valley counties combined.
For the US House of Representatives, I think the current setup (approximately equal representation) is in the Constitution. It would be hard to change.
There's actually been quite a bit of scholarship on how to draw district lines. We could certainly do better than the status quo, but it's a hard problem.
Because race is an important dimension that is a pretty good proxy for a lot of things in this country. It runs deep in the experience of nearly every African American, whose voting power as a block have systematically been curtailed by intentional and racist actions of others.
I do “class” is the bigger grouping, but race is important and correlated with socioeconomics.
Interesting, the census has followed in the footsteps of big tech and adopted differential privacy. I had the joy of working with differentially private data at my last job. It was brutal. Typically when you do differentially privacy you also do thresholding (any group with less than N members is dropped unconditionally). This means intuitive things you'd want to do are now subtly broken. Want to take an average? You can't because the lower values are just thrown out. Same with standard deviation. Basically all downstream analysis needs to be REALLY aware that the data is anonymized. I don't know if the census's implementation will have these problems but if so then yeah, as the paper found lots of analysis will either need to change or be done differently.
Analysis against enumeration looks like a similar problem, if I'm following you, in that there might be upstream context changing the way you should look at it.
One of my state agencies seems to handle your situation by publishing a table of the results of operations that make sense to run on the data you should have access to. I submit that this may not scale.
The race and ethnicity categories aren't designed for consistent data, they exist to satisfy interested organizations which are well suited to uniting people on the basis of skin color.
If the rule of 3 is applied to the population census, and blocks average 23 people, then no minority with less than 13% of the population can typically be reported at the block level. That knocks out most racial minorities. So it's a problem for redistricting.