Privacy concerns are breaking the Census
slowboring.com
slowboring.com
Being able to doxx with PII was shown to be able to be done in the suburbs, so this is more accurately phrased as "anyone in low-density suburbs and below are at-risk." If the author doesn't care about that, then just say it. There's many deanonymization attacks, and this is just one, but it showcases just how little is required with an actual dataset: https://www.cs.utexas.edu/~shmat/shmat_oak08netflix.pdf
The authors basic argument can be summarized as, "Well, companies already list all your personal data, or just flat out abuse it. So, let the government also list your personal data so we can make congressional seats as accurate as possible." Congressional seats being affected by census data scrubbing is not good, it should be fixed, but open data to the public that can be enumerated and deanonymized isn't a good thing either.
If the census can't be anonymized, or is rolled back then I imagine we'll have to start warning people about participating in the census. The irony at that point is your census data also won't be accurate because of a lack of participation.
You can deanonymize block level data and the study shows as much. You just need to have some amount of information that is either unique or low enough in distribution.
There's a tutorial on Twitter if you'd like to follow: https://mobile.twitter.com/john_abowd/status/111494218027827...
The deanonymization/reidentification attacks rely on joining census data against commercial datasets to produce something beyond what either contains. It's not accurate to say this information is already semi-public, because a good deal of it is not.
[1]: https://www.ncsl.org/research/elections-and-campaigns/access...
Maybe it would make it ok to donate if you also found a likable person on the other side to give money to at the same time.
I believe small, cash-only donations are anonymous. So that might be a good way.
It could be the other party trying to make you hate their opponent (a "False Flag" operation).
This was one of the biggest shocks I had moving to California. I eventually got fed up and actually replied telling them never to text me again and someone actually replied. That was when I learned that real people are on the other end of these things. It made it even creepier that my phone number was handed out, not to a machine, but to a real person.
Ross Anderson's Security Engineering Chapter 9 is all about this https://www.cl.cam.ac.uk/~rja14/Papers/SEv2-c09.pdf (Anderson releases the textbook free online)
This is a fairly well developed field (i.e., in medicine there highly screened "Medical Statisticians" who give data to research groups only after it's properly anonymized. They know beforehand what the researchers are interested in, and introduce noise that effects all the measures _except_ the ones the researchers said they were interested in looking at beforehand.
> The idea behind the differential privacy strategy, as I understand it, is that you can basically introduce statistical noise into the block-level results in a way that will make it impossible to reverse-engineer the demographic profile of people who live at specific addresses.
>This has, I think, roughly zero real-world privacy value
This was painful to read...
This is a really inaccurate way to characterize what researchers were able to obtain from reidentification attacks on block-level data from the 2010 census: https://www.science.org/content/article/can-set-equations-ke...
Isn't there a danger in the census statisticians doing the same thing? That they'll introduce noise that affects looking at the data in ways that they didn't anticipate or allow for. Or even that hinders drawing conclusions that go against current dogma.
The data they are collecting related to demographics is beyond what is necessary for congressional apportionment. If they collected only the required data; then there would not be a need to fuzz it.
If I'm trying to avoid catching Covid-19 by limiting exposure but have to go the the grocery story, the doctors, and the gas station it doesn't mean I should also go to a music concert since I'm already unable to avoid exposure.
Yglesias once again looking at the right thing and getting exactly the wrong conclusion out of it.
In general we would have to be much less guarded about the collection of personal information if we had laws prohibiting its harmful use ala the GDPR. Presently, anything that is collected becomes fair game for the advertising-surveillance industry's monetization, manipulation, and indefinite storage. So an entity with a bona fide use for personal information (eg The Census) cannot be trusted because resisting collection of such data is currently the only way to prevent yourself from being abused. And while I agree practically the Census isn't really collecting much data that the surveillance industry doesn't already have, this dynamic still informs societal norms.
The census is how they have that estimate...
This is superficial "privacy" which affords no marginal improvement in your actual privacy but does concretely break the Census.
Also, why is the Census (or any collective action) more important than an individual's subjective view of privacy (or any private thought)?
The term defines itself. I assume you understand what is meant by Privacy Derangement Syndrome, something like: valuing privacy above all other virtues. For example, somebody who refuses to use a mobile phone due to iOS privacy concerns, and falls out of touch with family and friends.
But I don't even really appreciate this point even as a joke. The idea that every possible mental malady is constrained and gatekept by a single book seems entirely too rigid.
The subjective view of privacy has some value, I'll grant that, but so to does a municipality preparing to lay infrastructure to support a growing population. When it comes to tangible value, the census is one of the most well-known value-adding resources I can think of.
I almost never argue against privacy, and policies and measures protecting individual privacy. But the US census is value-adding for research, policy, and individual purposes, let there be no doubt.
Giving purposely bad data is much worse than no data.
It's unfortunate that this means the Census isn't accurately presenting information needed for fair representation, but I don't think this is something that can be fixed by attempting to shame the Census for putting out "inaccurate information"?
The Census tried to move heaven and earth to contact us. We avoided them like the plague.
Imagine my surprise when i heard that our state had lost a house seat, due to being 80 people short of the required population count to keep its prior state house seat apportionment.
By avoiding the Census, we were able to change the machine of bureacuracy far better than by regular voting.
Sometimes the winning move, is really not to play.
>Given that baseline reality, compromising the usefulness of Census data for the sake of minimal privacy gains seems really bad.
The difference is that you can opt out of various companies collecting data. Can you opt out of the census? With this sort of argument, you can basically justify any sort of government privacy intrusion because private companies are already doing it, so "compromising the usefulness of NSA surveillance data for the sake of minimal privacy gains seems really bad". You do want to stop terrorists and pedophiles, don't you?
I'm going to assume that this isn't a rhetorical question.
Yes. Just don't fill out the form. And if the Census folks send folks to your house, you can pretend not to be home, politely ask them to leave, or in states that allow it, shoot such trespassers dead.
In fact, I'd prefer it if those who live in areas that consistently elect folks from $PARTY whose platform is more disagreeable to me do exactly that.
That way, when congressional districts are redrawn, the $PARTY that annoys me more will have a harder time getting elected. Redistricting is, after all, the primary purpose of the Census.
Edit: Less is more.
Just like paying taxes is optional because all you have to do is not send money to the IRS, and if they come you shoot them dead?
>The participation of every citizen in the United States Census of 2020 was mandatory by the law by Article I, Section 2 of the American constitution. Census is carried out in the U.S. every ten years to count the correct population in the country. The figures obtained in the 2020 Census were particularly helpful in determining congressional representation and community funding.
After a certain number of tries without getting a response, census takers are supposed to impute how you would have responded and use that.
One thing that annoyed me in particular was their creepy obsession around race. Even once you answer, "White", there are several more questions aimed at determining precisely what subcategory of white person you are.
If anyone at the census is listening, here's some advice. If you only get to ask Americans a handful of question once every ten years, don't waste half of them trying to figure out someone's 50 shades of whiteness.
That's a bit circular. One of the main reasons that districts are so easy to gerrymander is precisely because there is so much rich census data that can be used to predict voting patterns.
Also, I did some reading. Apparently the census has been asking these questions since 1790 https://www.census.gov/acs/www/about/why-we-ask-each-questio.... Though, I imagine the nature of the question has changed over time.
Watchdogs and groups trying to fight gerrymandering (racially-tilted or otherwise) usually don't have as many resources. It's harder for the public to keep an eye on this without public data, while keeping it private-only barely inconveniences the most relevant bad actors.
Parties know their votes by precinct, those data are available.
Apportionment of voting districts must be done by population, so the Census is necessary. Discrimination on the basis of race violates multiple Constitutional amendments and voting laws, for which enforcement requires clear knowledge of demographics.
The justirications are somewhat involved, but not strictly circular.
For a discussion of the background, from a conservative viewpoint, see: "Redistricting, Race, and the Voting Rights Act" (2010). This touches on the question multiple times, this passage captures much of the circumstance, though I recommend reading the full piece:
[T]he South's swift and effective circumvention of the Voting Rights Act in the wake of its passage demonstrated the truth of what many civil-rights activists had argued: that merely providing access to the ballot was insufficient after centuries of slavery, another century of segregation, ongoing white racism, and persistent resistance to black political power. In the Deep South in 1969 (and beyond), the alternative to de facto reserved legislative seats for black candidates would have been the perpetuation of whites-only politics.
https://www.nationalaffairs.com/publications/detail/redistri...
Demographic info adds almost no predictive value over this.
Unless I misunderstood your question and you meant to say that the race portion specifically is not.
This isn't really on the census. This is just the census trying to account for Americans' creepy obsession with race. Let's not forget that for most of this country's history, your race was everything.
You might not like the idea of being accounted for, but that negates exactly zero benefit to society that census brings. How can the rulemakers make informed decisions if they're blind?
If you dislike the census because you hate big gov't then it should be left there, even if you know it's not a popular opinion.
This is the latest version, but it was originally published in 1977 and based on the idea that humanity consisted of only four races. The reason it asks what "kind of white person" you are is the 1997 update was in response to criticism that this wasn't fine-grained enough, so they split Pacific Islander into its own group and split white people into Hispanic and Non-Hispanic. The idea of there only being four races in the world comes from Carleton Coon's publications from the early 60s. He had a theory that human races all evolved independently from Homo erectus to Homo sapiens, in contrast to the more widely accepted "out of Africa" theory, and that this accounts for why some races, according to him, are "more civilized" than others. That is, humans left Africa in a savage state and evolved into sapiens form and developed civilization much earlier in Europe and Asia, again, according to him.
I think it should be obvious why making a theory like this the basis of US official federal standards for race is somewhat controversial, but as we know, laws and regulations change slowly, if ever, so the Census is kind of stuck defending work they're legally charged with doing, and this data is used to assess things like racial equity of federal spending and assessment of the fair housing act and what not. It's kind of a relic of our shitty history, but it's there.
Truth be told, seeing the state of our governement, I fully expect all of it to leak out sooner or later. :|