Why pay $20 for your personalized Letterboxd Wrapped when you can use years of study and spend 20 hours developing a similar analysis yourself? Some might say it's a waste of time (and they’re not wrong), but it's more about the joy of discovery than convenience.
Over the past weeks, I developed a proof of concept to recreate Letterboxd Wrapped using tools available to any user.
For those unfamiliar, Letterboxd is a social media platform for movie lovers to log, rate, and review the films they've watched. Each year, much like Spotify, Letterboxd also releases a "wrapped" feature that aggregates the most interesting data from your account.
Initially, I aimed to create a simple Jupyter notebook to calculate the results. However, after discovering this post by one of Marimo's founders, the analysis quickly turned into a full-fledged app.
With Marimo, it's possible to create a prototype quickly—a process that would require many more steps if developed with Jupyter notebooks. Marimo allows exporting notebooks to formats like WebAssembly (WASM), making it possible to host a GitHub Pages project using this template.
The result? A fully interactive app where you can experiment with values, input files, and much more:
The issues…
I thought I'd found the holy grail of prototyping, but there are some drawbacks:
- Load Times: Every access requires downloading Python dependencies, resulting in slow load times—unacceptable for a modern app.
- Integration Issues: Libraries like Matplotlib had issues, so I switched to Plotly. Some libraries don't integrate well with this workflow.
Final thoughts
In conclusion, while there are challenges such as slow load times and integration issues with certain libraries, the process of developing a personalized Letterboxd Wrapped using Marimo and other modern tools is both rewarding and insightful. The ability to create a fully interactive app and explore your movie data in depth offers a unique experience that goes beyond convenience. Despite the hurdles, the potential of using a modern stack like UV, Polars, Marimo, and DuckDB for Python prototyping is promising and worth exploring for those passionate about data analysis and app development.
For technical details, refer to the GitHub repository.