NICAR — the National Institute for Computer-Assisted Reporting ^{— is one of my favorite annual journalism conferences, even if I haven’t been in a few years. That’s because it’s uniquely easy to benefit from even without attending. As the nation’s news nerds descend on a city (this year Minneapolis), they come bearing PDFs and slide decks and GitHub repos and all manner of tipsheets. The digital trail they leave across the web allows those of us at home to benefit from all the tools and techniques they share.}

Usually, there’s someone online who tries to gather up all the best NICAR material each year, to assemble them all in one place. Chrys Wu used to do it; Brent Jones used to do it; IRE itself has done it. But the reigning champion is Sharon Machlis, the freshly retired journalist responsible for Practical R for Mass Communication and Journalism — a sort of gateway drug into full-on news-nerdery. On her NICAR site, Machlis has assembled all the best of the past six NICARs, all the way back to the cursed 2020 edition in New Orleans, which you remember served as the first COVID superspreader event in the news industry. That’s 225 tipsheets, presentations, or repos in all; if there’s any theoretical journalism + code amalgamation you’ve ever wondered about, it’s probably in there somewhere. The 2025 batch is as rich as any, so dive in; here are a few that stood out to me, with apologies to those I’ve missed.

AI: I hear it’s becoming a thing

If you want to convince your editor you’re ahead of the AI curve — or at least within spitting distance of it — NICAR had you covered. Our old friend Matt Waite posted an R + LLM starter kit, useful for wrangling open-source models and cleaning messy data. Mike Reilly (author of The Journalist’s Toolbox: A Guide to Digital Reporting and AI) offered up a suite of AI tools and resources, especially for newsroom trainers. Jonathan Soma of Columbia’s j-school posted an AI-in-the-newsroom guide and a tutorial on image and video analysis. ProPublica’s Brandon Roberts showed how to integrate it all through Python. (New to Python? Try Mindy McAdams’ beginner’s guide.) Just want the big picture on where we are with AI? Simon Willison’s LLM state-of-the-union is your guide.

Tracking down digital secrets

Another big thread through the conference was specific tools for digital investigations, of the sort often labeled OSINT, or open-source intelligence. (Meaning publicly available tools that can help you find out things someone would rather keep hidden.) Prianjana Bengani and Jon Keegan taught how to use public data to figure out who’s behind a mysterious website. Tristan Lee and Logan Williams from Bellingcat (the OSINT experts) shared their favorite investigative tools. The Baltimore Banner’s Greg Morton showed how to use OSINT to investigate your local transit system. Caitlin Ostroff and Jeremy Merrill went over demystifying cryptocurrency and other blockchain data. A Times-Post-ProPublica triad show how to do the same with nonprofit data.

Maps, maps, and more maps

Maps — and their fancier-sounding big sibling, spatial analysis — are always a big topic at NICAR. The Houston Chronicle’s Alexandra Kanik showed how to build maps through scripting. Brandon Liu showed how to make them interactive. Shreya Vuttaluru of the Tampa Bay Times and Ryan Little of The Baltimore Banner went over spatial analysis in R. Sandhya Kambhampati of the L.A. Times and Michael Corey
of the Star Tribune shared how to apply it to census data. Speaking of, Andrew Ba Chan showed off the Tidycensus library.

The grab bag

Geoff Hing will get you comfortable with the command line. Derek Willis gives a tour of the worst PDFs on the planet (and has some solutions). (So does Jeremy Singer-Vine.) Derek and Matt Waite talk about bringing data journalism into the sports section. Christian McDonald will finally make regex make sense to you.

And that’s just a taste! If all that’s not enough, Machlis also built a scraper that pulled in all NICAR-related Bluesky threads.