Researchers can now sort Congressional tweets by whether senators talk about problems or solutions

Computer scientists built a tool that reads US senator tweets and automatically labels them as either identifying a problem or proposing a solution. The tool works with 80% accuracy, which means it could let researchers measure whether senators spend more time complaining or problem-solving, and whether that changes over time or by party.

Until now, measuring what Congress actually talks about required humans to read thousands of documents by hand. This tool lets researchers analyze millions of social media posts automatically, turning a tedious manual task into something repeatable. That matters because if you can measure something at scale, you can actually study whether legislative behavior changes, whether it correlates with elections or crises, or whether different senators have consistent patterns.