Investigating wrongdoing sometimes yields unexpected results.
Not Shutting Up
By Stephen Engelberg

Welcome to Not Shutting Up, a newsletter from ProPublica’s leadership. You’re receiving this because you’ve supported ProPublica’s journalism; we’re grateful for that, and we hope to give you some context on how our newsroom works. If this email was forwarded to you, you can sign up to receive it here.

Dear ProPublicans,

There’s an adage in journalism: Never let the facts get in the way of a good story. The advice was never meant to be taken literally. Outside of certain infamous tabloids, the grizzled city editors who would growl this phrase at their charges were not asking for fiction. They just wanted good stories to be done quickly, without what they would derisively dismiss as “over-reporting.’’

At ProPublica, where we pride ourselves on showing the receipts, over-reporting is both a passion and a necessity. Many stories look good in their early phases but fall apart upon closer scrutiny as facts come to light that “get in the way” of what looked like big scoops. Sometimes, that’s because our original premise is off. On occasion, it’s because government agencies have rearranged the facts. Our newsletter today is guest written by Ellis Simani, a data reporter at ProPublica who experienced this phenomenon firsthand. Welcome, Ellis.

— Steve Engelberg


Since the pandemic began, much of my reporting has focused on chronicling the stories of renters facing the threat of eviction and understanding the extent to which policy measures have succeeded in keeping families in their homes.

Five months ago, President Donald Trump signed the $2 trillion pandemic relief package (also known as the CARES Act) into law. The law is well-known for providing many Americans with $1,200 stimulus checks, but it also temporarily protected millions of tenants from being pushed out of their homes for not paying their rent.

The CARES Act applied to properties benefiting from a patchwork of federal programs that cover more than a quarter of all rentals nationwide. It wasn’t perfect: Legislators included no explicit enforcement mechanisms or penalties for owners of federally backed buildings who tried to evict their tenants.

In the weeks after the law passed, my colleagues and I found landlords in several states who violated the law and moved to throw more than a hundred people out of their homes.

What became clear to me as I talked with renters, many of whom had either lost their jobs or had to take care of children and couldn’t return to work, was that there was no clear way for them to find out if they were protected by the federal ban.

I wasn’t the only one to recognize this. The government-sponsored mortgage company Fannie Mae and its sister company, Freddie Mac, created websites in early May where renters could search to see whether their buildings had mortgages backed by either agency.

At the time, I had teamed up with my colleagues Jeff Ernsthausen and Al Shaw to build our own searchable database for renters. Our goal was to build a more comprehensive resource by combining data from Fannie Mae and Freddie Mac with data from a handful of other federal agencies.

We wanted to make sure our work was accurate, so we tested Fannie Mae’s new site by entering in a few hundred property addresses from a list it had provided to the public. We were satisfied when the vast majority of those properties were correctly identified as subject to the eviction ban.

We published our tool and continued reporting on how renters across the country were dealing with the economic downturn. Then, this month, we got a tip from a source who said she was having problems with Fannie Mae’s site. Properties that had initially been tagged as having mortgages backed by Fannie Mae now appeared as not being covered. Jeff and I decided to look into it.

When I rechecked the same addresses that we had back in May, only about a quarter showed up as being backed by Fannie. I was alarmed. If the agency’s own data was not appearing correctly, then I knew renters and their advocates who were entering addresses were likely being misled.

We wanted to make sure we understood the scope of the problem. We checked over 2,000 more apartment buildings from Fannie Mae’s most recent data against the tool. Again, the site showed only about a quarter of those properties as having a mortgage backed by Fannie Mae.

We asked Fannie Mae to comment on our findings. In a statement provided to ProPublica, a spokesperson for the agency suggested that users might get an incorrect result because of differences between an address as it appears in internal records and the information entered into the search tool.

What struck me most, however, was that Fannie Mae emphasized that its tool wasn’t meant to be used by tenants in court, and that it wasn’t offering any guarantees about the accuracy of the results. That didn’t square with our reporting. We spoke with a legal aid attorney in Atlanta who had used information in Freddie Mac’s lookup tool to persuade a landlord’s attorney to drop an eviction case.

Because most tenants in housing court don’t have lawyers, Fannie Mae’s position — “don’t rely on what we put online” — put the entire burden on renters to prove they were protected.

That seemed unfair. Without reliable, centralized ways of determining whether a building receives federal backing, tenants and legal advocates were left to sift through public records that can be hidden behind paywalls or only accessible in person.

Soon after Fannie Mae’s response, I reopened its tool in my browser and knew right away that the agency had been at work. There was now a test known as a CAPTCHA, a common system used to distinguish between human and automated use of a webpage. Such a setup makes it harder for data journalists to do big, automated searches.

Additionally, a note at the bottom of the tool was flagged as an “IMPORTANT DISCLAIMER,” and the text emphasized that the tool “is provided for informational purposes solely as a convenience for renters.”

These were the first hints that we might have achieved what Steve Engelberg called “pre-impact” in his column a few weeks ago.

As we do with all of our reporting that involves robust data analysis, we assembled a team of our colleagues to go over a portion of our work and independently confirm our finding that the government tool was missing many properties.

With the CAPTCHA system blocking us (or any other computer-savvy users) from quickly checking whether a number of properties were showing up as covered, so began the tedious task of searching each address by hand.

Our spot checks showed that a number of the records listed earlier as not being backed by the agency were now showing up as covered –– as they originally had in May. It was clear that a significant change had been made to the tool.

Fannie Mae never told us what changes it made, or whether our findings influenced it. What is clear is that in the days after we reached out to the agency, the tool seemed to have significantly improved in performance, and we no longer had a particularly interesting story on our hands.

At its best, our journalism challenges powerful institutions, holds leaders accountable and empowers you, our readers, to make more informed decisions. If we’re fortunate, our reporting generates impact after it’s been published. Things don’t always happen in that order. And yes, facts can get in the way of seemingly good stories.

— Ellis Simani

Data Reporter, ProPublica

More From Our Newsroom
The Mystery House: How a Suspicious Multimillion Dollar Real Estate Deal Is Connected to California’s Deadliest Fire
Hundreds of Thousands of Nursing Home Residents May Not Be Able to Vote in November Because of the Pandemic
Alaska’s Attorney General Resigns Hours After We Published “Uncomfortable” Texts He Sent to a Younger Colleague
She Was Sued Over Rent She Didn’t Owe. It Took Seven Court Dates to Prove She Was Right.
Federal Prosecutors Have Steve Bannon’s Murky Nonprofit in Their Sights
Black Workers Are More Likely to Be Unemployed but Less Likely to Get Unemployment Benefits
Was this email forwarded to you from a friend? Subscribe.
This email was sent to example@example.com. Update your email preferences or unsubscribe to stop receiving this newsletter. Email not displaying correctly? View it in your browser.