For the past six months, we’ve been meeting with reporters from some of the nation’s largest news organizations to demo Everlaw for Journalists, our program to support investigative reporting through pro bono access to the Everlaw platform.
We’ve learned a lot from those conversations—in particular, what reporting a big story has in common with building a case. We’ve heard how reporters need to sift massive amounts of data and figure out what matters and what doesn’t. How they scour archives for important documents and fit them into a timeline. How they analyze who was speaking to whom, and when. How they look for anomalies, anti-patterns—the places in evidence where things get weird.
We heard a lot about the differences, too. Unlike attorneys, reporters often untangle complicated stories with little more than Google Docs and a wall full of Post-Its.
Attorneys have it easy by comparison. They have software that automatically ingests documents, extracts text, scans PDFs with optical-character recognition, and organizes everything for rapid search. It can weed out duplicates and analyze email threads to identify gaps in your records. It can even predict what you’re looking for before you ask for it.
Our goal with Everlaw for Journalists is to put this power in the hands of reporters. Our theory is, you can find a story with Post-Its, but you’ll find it faster with machine learning and a cloud supercomputer on your side. Do we know that for sure? No, but we want to give it a shot.
The journalists we talked to are at the top of their game. They have their methodology down. But they immediately saw the possibilities of adding Everlaw to their workflow. Here are some of the features that sparked the most ideas.
Email threads and other related documents
In reporting, as in law, you don’t always get archives in tidy machine-readable format. Email threads are often delivered as disconnected PDFs. Chaining them together by hand can be a huge undertaking. But Everlaw can import those documents, extract the text and piece them back together as if you had the original email files. In addition, it identifies holes in the thread where separate emails should appear but were not delivered—useful for revealing potential sources of additional information. All the other related documents in the collection, from near-duplicates to email attachments, are displayed alongside each document in the same way.
Everlaw’s StoryBuilder tool for case-building and deposition prep is a great way to create a timeline, construct a gallery of players, and directly tie key documents to events, people, locations, and issues, with everything automatically cross-referenced so you can investigate connections between entities just by clicking on them. It’s also an easy way to outline a story even as it’s coming together, associating each point in the outline with the salient documents from the collection. We designed StoryBuilder for collaboration between attorneys in far-flung locations, so it’s a powerful tool for teams working together to report a story.
This is where AI meets reporting. Everlaw’s “predictive coding” technology uses machine learning to help attorneys identify key documents in vast archives of data. It can do the same for reporters. It runs in the background, monitoring documents you mark important, figuring out what they have in common and finding more just like them. The more documents your team classifies, the better the AI gets at finding more interesting ones. If there’s a smoking gun hidden in your archive, there’s a good chance predictive coding will bring it to the surface.
Also, as a practical matter, editors (we hear) can get impatient about the progress of investigations. With predictive coding, you’re constantly finding new and interesting documents that move the story along. You may not have it all pieced together, but the pieces keep coming.
Sound intriguing? If you’re a journalist and think Everlaw could help your reporting, we’d love to hear from you.