The Race to Preserve Government Websites and Data
‘We are on very thin ice.’
Fearless reporting, a behind-the-curtains look at how journalism is made — and an unabashed point of view. Welcome to Chills.
Nothing is ever really gone from the internet. That’s how the saying goes, but experts racing to capture online government data as it is being slashed and burned at terrific speed say that if you want something published on the internet to stick around, make sure you have your own copy somewhere.
But right now, President Trump’s administration is rapidly deleting web pages and entire data sets, and there is not always a copy somewhere of this important information.
As of Feb. 2, more than 8,000 government web pages had been removed since Trump took office, The New York Times reported. That number continues to multiply every day. It’s an unprecedented deletion of critical government data — everything from community pollution maps to a Centers for Disease Control tool that evaluates social factors that make communities vulnerable during disasters. In particular, government agencies’ pages on climate change are being obliterated.
Then there are the insidious changes to government sites that remain up but now have shifted or missing language. NPR points to pages caught up in the culture wars on the CDC website that previously referred to “pregnant people” — it now calls them “pregnant women.” Also, CDC references to COVID-19 and pneumococcal vaccines on a recommended immunizations page, here captured before Trump took office, are now gone.
“We’ve lost something in the internet age by thinking that everything that is online, that means it’s going to stay forever,” Jack Cushman, the director of the Harvard Library Innovation Lab, told me. “In fact, it is very fragile. In fact, we are on very thin ice.”
White hat groups across the country like Cushman’s are scrambling to preserve everything from data sets, as at Harvard, to pages with language used before Trump took office. The Innovation Lab said in a statement that, so far, it has preserved 311,000 datasets copied between 2024 and 2025, which is 16 terabytes of data.
Still, Christina Gosnell, cofounder and president of Catalyst, a data analysis agency, called all this archiving work “a temporary Band-Aid.”
There is a kind of universal brittleness on the internet created by our dependence on central locations for content, like YouTube. After you post a video on YouTube, for example, the company can take it offline whenever it wants. You may not have a copy of the original, and YouTube may or may not have a backup, which may or may not be in great condition. This brittleness is particularly treacherous when it comes to preserving data.
“If data sets are removed and are no longer updated,” Gosnell said, “our archived data will become increasingly stale and thus ineffective at informing decisions over time.”
Another challenge groups recovering websites and their data are facing is the sheer amount of space government data take up. No one has the server power of the U.S. government. There are more than 307,000 data sets on data.gov. On top of that, there is no single place keeping track of all (quickly disappearing) government websites. Aka the endeavor to preserve data and pages as they were before Trump took office is massive and complicated.
“We’re just a hair’s breadth away from having large, important parts of information vanish,” Cushman said, “because either someone decides to make them vanish or because no one decided not to.”
In this new era of Trump, where illegal government acts are taking place daily, the race to capture vital public, taxpayer-funded information is critical not only because we need such health and other information, but because there is a record that needs rescuing before it is lost to history. I like how Cushman put it to me:
“If you’re archiving, you’re also witnessing,” he said.
It’s one of the very least, but perhaps one of the most important, things we can do right now — stand witness to the destruction of our country’s scientific and other research, and perhaps even of our very own democratic ideals.
Chills is self-funded, without ads. If you want to be a part of this effort, of revealing how difficult reporting is made — of sending me to places like Ukraine to report for you — I hope you will consider subscribing for $50/year or $7/month.
I don't think we are on the ice anymore. I feel like I'm drowning.