Weed Out

Inspiration

There are a lot of things going on in the world today, politically and scientifically. It's not uncommon for people to fall out of the loop when it comes to staying updated with the latest news. We wanted to create something that encourages people to stay up to date while also testing their ability to distinguish between fake/real news. We took inspiration from The New York Times Games to create a simple game that can be played whenever you're bored.

What it does

Weed Out is a game where players read news headlines and summaries, then pick out the "weeds"—the ones that contain fake news! There are 2 out 5 articles that are the "weeds". It mimics The New York Times Games where each player gets a certain amount of tries to find the fake news articles. Each day, there are new articles to explore.

Users can also add friends and track their scores on a leaderboard.

How we built it

We started by scraping online articles from reputable news sites, as well as claim reviews from the Google Fact Check Explorer. We processed our false claims to be understandable headlines using Google's generative AI model. We then categorized them into political or scientific news using a simple logistic regression machine learning model. For our backend we used Flask, PostgreSQL, and SQLAlchemy. The frontend was built using React, JavaScript, and Tailwind CSS.

Challenges we ran into

We had some issues with the user authentication process but were able to figure it out eventually. It was also difficult to find data sources for the false vs. true claims. We toyed with having a ML model that would classify fake news from true news, but this could potentially be unreliable, and we wanted the claims presented to be definitively false or very likely to be true.

Accomplishments that we're proud of

We're proud of being able to get a somewhat solid backend working with the way we processed the news articles. Being able to create the game on the frontend was also a good accomplishment.