Media Analysis Reimagined
In today's rapidly evolving media landscape, understanding how different platforms shape public discourse has become crucial for informed citizenship. Traditional legacy media outlets - newspapers, television networks, and established digital publications - operate under editorial standards developed over decades. Meanwhile, social media platforms like X provide real-time, unfiltered reactions from millions of users, creating a dynamic environment where news spreads instantly but with varying degrees of accuracy and emotional intensity. This AI-powered app monitors top global headlines and finds corresponding X posts, comparing how the same stories are covered, discussed, and received across these fundamentally different media environments. By examining sentiment patterns, emotional responses, and narrative framing, we can discover hidden patterns that influence how society processes information, forms opinions, and makes decisions. This comparative approach doesn't seek to determine which platform is "better," but rather to illuminate how each contributes to the complex tapestry of modern news consumption, helping readers develop media literacy skills essential for navigating our interconnected information age. Specifically, news stories are analyzed for overall sentiment (on a 0-1 scale where 0=very negative, 1=very positive) as well as the dominant emotion expressed, from both legacy sources and X posts. The data is then filtered for news items that are covered substantially differently and conveniently displayed in a table below (see the Latest Data section) for users to peruse, along with a commentary on the differences. The data automatically updates each day at 10pm CST (to capture a full day's worth of news stories), and is automatically posted here as well as on our X account, @h45hb4ng_data. Note that while only stories with a substantial difference in coverage are displayed here (in the latest data section), the full data for each article is in the database in case you'd like to refer to it later. See the source for further details.
Latest Data
Comparison of world news coverage between legacy media outlets and social media discourse on X. Data is automatically updated daily at 10pm to capture evolving narratives based on a full day's worth of news headlines. Daily data collection ended 8/16/26, feel free to view the source and/or archive if curious about how it works. There is also an analysis of what the data reveals here.
| Timestamp | Article Title | Legacy Sentiment | X Sentiment | Legacy Emotion | X Emotion | Analysis |
|---|---|---|---|---|---|---|
| September 29, 2026 @ 12:15AM | Nvidiaβs $150 Billion Buyback: A Lifeline for AI Hype, Not Everyday Investors | 0.25 | 0.62 | Skepticism | Optimism | Article tone is markedly skeptical (0.25 sentiment) framing Nvidia's buyback as propping up AI hype rather than benefiting retail investors. X reactions lean more positive with average 0.62 sentiment, viewing the $150B authorization as strong cash-flow confidence and a bullish signal for sustained AI demand. Common emotion on X is optimism, contrasting the article's caution with investor excitement over shareholder returns and market resilience. |
| September 29, 2026 @ 9:00PM | Mistral CEO Says US AI Safety Debate Masks Competitors' 'Negligence' | 0.3 | 0.48 | Anger | Neutral | The article's negative tone (0.3 sentiment) and anger focus on accusing US firms of using safety debates to hide negligence, while X posts are mostly neutral factual shares or summaries of the CNBC/Slashdot piece with minimal opinion. X reactions lack the article's critical edge, showing slight curiosity or balance instead of anger, as users repost without strong endorsement or critique. Overall, X discourse appears more detached and informational compared to the article's pointed negativity. |
| September 29, 2026 @ 4:01AM | PwC's huge survey of 50,000 workers says AI is splitting employees into 4 distinct groups | 0.59 | 0.61 | Fear | Neutral | Article tone is cautiously positive (0.59) with underlying fear of workforce division from AI. X posts are mostly neutral factual shares of the PwC survey link, highlighting uneven benefits and majority at risk without strong emotional language. Overall X reactions mirror the fear element but with flatter, less alarmed sentiment than the article. |
Archive
To access data for a previous date, please refer to our account on X, which automatically posts the current day's data at 10pm CST daily.
Source Code
This project is completely open source and free. It requires only PHP, MySQL, and API keys for xAI and NewsAPI. It is webserver agnostic, so you can use Apache, nginx, or whatever else will run PHP. The code that automatically posts the data to X also requires an API key from X and the free software packages wkhtmltopdf/wkhtmltoimage and ImageMagick (if you're running Ubuntu, sudo apt install wkhtmltopdf imagemagick is quick and easy).
Here is a quick rundown on the files:
- config.php - set your API keys here
- database.php - database functions to connect to your MySQL server, uses the MySQLi PHP extension
- get_news.php - gets today's global news headlines from legacy source via NewsAPI
- index.php - this file, which acts as the primary frontend
- main.php - responsible for invoking the functions defined in get_news.php, xai_article.php, and xai_x_posts.php. This is the file you want to execute in your crontab
- table.php - handles the latest data table, index.php calls this file as an include
- xai_article.php - passes a legacy news article to the xAI API to perform sentiment and emotion analysis
- xai_x_posts.php - searches X (via xAI API) for posts describing a particular news story. Finds the average sentiment and dominant emotion among these posts
- x_post.php - responsible for automatically posting the data to X. This should also be executed via cron slightly after main.php. Personally, I run them at 10:00 pm and 10:05 pm. Note that this requires an X API key in config.php as well as the TwitterOAuth package (easy install via composer)