A Health Data Surveillance State that Dominates
It's a short trip from "data modernization" to national eugenics registries
Media attention to the illegal, government shutdown-enabled reductions in force (RIFs) that laid waste to the Centers for Disease Control and Prevention (CDC) has rightly focused on the destruction of public health capacity in America. This is an immediate and serious threat to national and international health and security.
That’s pretty obvious for something like outbreak response. When the largest measles outbreak in decades occurs, most people can see how it’s useful to have experts on hand who know how to contain it. That’s probably why these roles were some of the RIFs made “in error.”
But the CDC doesn’t only spring into action when there’s an outbreak. They are always at work, collecting data on anything and everything related to American health: cancer, mental illness, diabetes, Alzheimer’s, smoking-related illness, nutrition, women’s health, children’s health, vaccination, and, of course, infectious diseases. These data directly translate into CDC recommendations, programs, and initiatives geared towards improving health for millions of Americans. CDC collects a LOT of data. Managing that much data is really challenging. In 2020, the Department of Health and Human Services (HHS) contracted with a software company to help manage data systems during the pandemic. This turned into a larger data modernization effort. That contractor was a company called Palantir.
The all-seeing eye
Palantir is named after the magical seeing stones that acted as basically evil walkie-talkies in Lord of the Rings, probably because Palantir’s founder Peter Thiel’s non-evil technofascist bookshelf consists solely of Tolkien. Palantir is a big data analytics company that makes software for managing large and complex datasets. However, the true driving force behind their business is not the software they sell but the data they query.
Google searches for news articles Palantir are like a tour through a techno-dystopian theme park and it’s not hard to see why. Their technology is inherently dual use (meaning it has both civilian and military applications). It’s also somewhat hard to understand intuitively. Even former employees struggle to describe what Palantir’s software actually does. Their business isn’t built as much on acquiring data but doing things with it. Specifically, it allows companies or institutions to rapidly integrate multiple, often very different types of data to answer complex questions using AI models. For example, if I was a CDC employee and I wanted to know who was most at risk from seasonal flu, I might want to query both flu surveillance data, national hospital records, and immunization data, all of which would are separate datasets, then analyze these data using a predictive model.
However, Palantir hasn’t made a name for themselves by helping their customers query health data to assess risk to public health. Their focus has been defense and law enforcement, based in part of founder and human Troll doll Alex Karp’s belief that Palantir’s purpose is defending Western civilization. Karp believes that “wokeness” is a serious threat to America (and his company). He’s right if by “America” he means a technofascist surveillance state and by “wokeness” he means basic Constitutionally guaranteed civil liberties. Palantir has been developing AI software that enables Minority Report-style “predictive policing” since 2013. They have provided AI systems for targeting, battlefield management, and combat decision making to the US military, the Israeli Defense Forces, and NATO. They have quietly been installing the architecture to carry out mass surveillance of citizens and enemy combatants alike, as well as coordinate law enforcement and military strikes. Several months ago, Palantir was awarded a $30 million contract to develop a platform to help Immigrations and Customs Enforcement (ICE) deport people more efficiently.
Palantir has made it pretty obvious that their goals as a company are to: 1. establish a pervasive, omnipresent surveillance state and 2. acquire and analyze so much personal data that people cannot escape its constantly vigilant gaze. I personally do not feel great about this company having unfettered access to my and every other American’s health data.
What does this have to do with the RIFs?
When Palantir got the contract to develop the One CDC Data Platform (1CDP), there were immediate concerns about possible abuses of the technology. For one thing, the Foundry platform that 1CDP is built on allows easy merging of data from different government agencies. So CDC data could be merged with data from, say, ICE, who has been doing business with Palantir since 2011 and is one of their most loyal customers. This is not particularly comforting, considering the type of data that CDC has collected.
CDC’s excellence in public health has largely been because of how much data is has collected on American health, in turn allowing CDC to set the standard for evidence-based public health. CDC databases contain everything from reportable disease data from state and municipal health departments to wastewater surveillance to nutrition and environmental exposure data. It collects data on cancer, antibiotic resistance, pregnancy and maternal health, mental health, and on health equity and the impact of poverty on disease. CDC tracks information on every aspect of vaccination and every disease that affects Americans or poses a threat to American health security. As a result, some of this data contains sensitive personal information.
Although clinical data is deidentified as a matter of practice, it still can be used to identify people with the modeling approaches that Palantir specializes in. I use machine learning classification in my own research to predict disease outcomes based on the gene expression responses that a virus provokes. These models work by “training” the model on gene expression data associated with known outcomes. These models’ predictions become more accurate when the training set is very large, so you can capture all the variation within the data. When trying to predict something more complex—like will someone commit a crime rather than will this mouse die from Ebola—you need a really, really huge training set.
Guess what the CDC has? A really, really huge training set, on the entirety of America, going back decades. This would allow Palantir to develop tools that would enable agencies like ICE to arrest women for seeking an abortion or trans kids for seeking gender-affirming care. It could lead to persecution of people seeking treatment for drug addiction or mental health. People could be targeted based on their medical history or genetic background. Based on the services Palantir has already delivered to military and law enforcement customers, I have no confidence that they can be trusted to manage these sensitive data without responsible oversight. Unfortunately, from what I can tell, there aren’t many people around to provide it.
Some of the RIFs at CDC impacted the Office of Public Health Data, Surveillance, and Technology (OPHDST), which already was without steady leadership since OPHDST Director Jen Layden left the agency in August. These RIFs at OPHDST and the Office of the CDC Chief Information Officer (CIO), as well as employees leaving voluntarily or in prior layoffs, are removing key people from the job who provided oversight and governance to Palantir’s team embedded at CDC.
Although there are guardrails in place to prevent CDC data from being merged with other agencies’ data, the people charged with safeguarding the privacy of Americans’ sensitive health data are no longer on duty. The platform stewards who ensure that 1CDP is used for purposes that align with CDC’s mission are no longer at OPHDST or CIO. Having a governance process in place doesn’t really matter if nobody is around to carry it out.
Palantir has already experienced a windfall from integrating and modeling data across government agencies to support the rise of the burgeoning domestic surveillance state. They have a strong incentive to add some of the most sensitive and personal data possible to their training set: data about our health, which is really data about our identities.
The worst phoenix rising
I think the CDC is no longer capable of serving American health because it has been so viciously attacked and depleted of expertise and knowledge. A lot of people showed up in my Twitter/X mentions to tell me how great this is and that it is a joyous occasion to see CDC burned to the ground. But I don’t think it’s going to be burned to the ground. While it won’t be functional, I think that CDC will still exist as an agency. However, the mission-driven public health professionals upon which CDC built its reputation will be replaced with a smaller crew of political appointees. They will transform CDC into something else.
CDC is useful to both President Donald Trump and HHS Secretary Robert F. Kennedy, Jr. as a mouthpiece for laundering ideological initiatives—that may hinder Constitutional rights, civil liberties, and privacy—as beneficial for American health. Palantir is handing them tools that weaponize Americans’ own health information against them. Palantir can provide a powerful mechanism for allowing Trump to eliminate opposition and allowing Kennedy to fully lean into his eugenicist quackery—at scale.
There are more reasons to fear the destruction of the CDC besides just the consequences for our health. This is one of them.
OMG...the things that happen behind the curtain
JJF Phm 🇨🇦
Best Substack of the day