According to Bloomberg Business, the reliability of U.S. government data is in a state of crisis, exacerbated by the nation’s longest government shutdown in 2025 under a Democrat-led Congress. That shutdown caused major datasets to be incomplete or disappear, like the National Oceanic and Atmospheric Administration halting its tracking of billion-dollar weather disasters. Delayed data is now trickling out, showing high GDP growth and lower-than-expected inflation, but the numbers are drawn from patchwork sources, making historical comparisons difficult. The politicization of data intensified when President Trump fired Bureau of Labor Statistics Commissioner Erika McEntarfer after a disappointing August jobs report, accusing her of political bias. Furthermore, longstanding metrics like the Census poverty rate and CDC maternal mortality figures are shown to have significant methodological flaws that can mislead the public, such as a senior poverty rate of 9.9% that drops to 5.9% when including retirement account withdrawals.
The boring truth about messy data
Here’s the thing: the real scandal often isn’t conspiracy. It’s complexity. Take the two examples from the piece. The official poverty measure for seniors looks bad, but it ignores money people pull from their 401(k)s. So the picture is wrong. The CDC’s maternal mortality rate looks horrifying, but that’s partly because since 2003 they’ve counted any death with a pregnancy checkbox ticked, even for an 85-year-old. The numbers got inflated.
But so what? Doesn’t that mean the data is garbage? Not exactly. It means we have to be smarter consumers. The “pregnancy box” was added because we were probably undercounting deaths before. There’s always a trade-off. Do you want a consistent measure that lets you track trends over decades, even if it’s imperfect? Or do you want the most accurate snapshot right now, even if it breaks the trendline? Ideally, you want both published side-by-side. Some agencies, like the CBO, do this well with multiple forecast scenarios. Many don’t.
When the spreadsheet just vanishes
But methodological debates are a luxury. We’re now facing something worse: no data at all. The shutdown created huge gaps. And the Trump administration’s habit of discontinuing or delaying inconvenient datasets sets a terrifying precedent. Fire the statistician. Ignore the report. Delete the database. It’s the ultimate “kill the messenger” strategy.
And then what? Policymakers are driving blind. How do you allocate disaster relief if you’re not tracking billion-dollar disasters? How do you fix education if student score data is stuck in a queue? You’re left with anecdotes, vibes, and whatever cherry-picked number supports your pre-existing belief. That’s not governance. It’s guesswork, and it’s incredibly dangerous for everything from public health to the economy.
Our shared reality is at stake
This feels abstract, but it’s not. Thomas Sowell had it right: the difference between cavemen and us is the knowledge we can apply to resources. That knowledge starts with measurement. Reliable, boring, publicly-accessible data is the bedrock. It’s the shared reality that, even when we disagree on solutions, lets us agree on the problem.
When that erodes, everything becomes a conspiracy. The rallying cry to “do your own research” is a joke when the primary sources are gone or doctored. People retreat to their informational silos—a trusted podcaster, a Substack, or just asking an AI to confirm their bias. We lose the ability to be surprised by a fact that challenges our worldview. And that’s how a society stops being able to solve hard problems.
What do we do about it?
So what’s the way out? First, demand transparency and methodological honesty from agencies. Publish the alternative measures alongside the old ones, as the piece suggests. For a deeper dive on the maternal mortality mess, this analysis is crucial. Second, and this is harder, we have to resist the urge to cry “conspiracy!” every time a number we don’t like comes out. Sometimes a bad jobs report is just a bad jobs report.
But most importantly, we have to defend the institutions and the infrastructure itself. We have to fund it, shield it from political interference as much as possible, and treat its caretakers with respect. This isn’t about blind faith in bureaucracy. It’s about understanding that the alternative—a world without agreed-upon facts—is a world where power is the only truth that matters. We got closer to that edge this year than ever before. And pulling back will require more than just restoring a server. It’ll require restoring a shared commitment to the truth, however inconvenient it may be.
