# Build Your Own Timeline Algorithm with BYOTA > [! note]- > The content of this page is generated by audio/video transcription and text transformation from the content and links of this source. Source: [https://fosdem.org/2025/schedule/event/fosdem-2025-5601-build-your-own-timeline-algorithm/](https://fosdem.org/2025/schedule/event/fosdem-2025-5601-build-your-own-timeline-algorithm/) <video src="https://video.fosdem.org/2025/ud2208/fosdem-2025-5601-build-your-own-timeline-algorithm.av1.webm" controls></video> ## Summary & Highlights: The session 'Build Your Own Timeline Algorithm' presented by Davide Eynard at FOSDEM 2025 focuses on creating personalized timeline algorithms that prioritize user needs over corporate interests. It introduces a stack using Mastodon.py, Llamafile, and Marimo to enable local computation and privacy-respecting solutions. **Introduction to BYOTA** BYOTA, a project by Davide Eynard, is not a new algorithm or end-user application but a call for community-driven development of ethical timeline algorithms. It emerged from the need for user-controlled social networking experiences, particularly following the Twitter migration. **Challenges of Current Timeline Algorithms** Traditional timeline algorithms often prioritize engagement metrics over user needs, leading to issues like bias, lack of transparency, and privacy concerns. BYOTA aims to address these by offering open, transparent, and customizable solutions that can run locally. **Technical Approach and Tools** BYOTA leverages tools like Mastodon.py for data retrieval, Llamafile for local embedding calculations, and Marimo for a browser-based UI. This setup allows users to perform searches and recommendations without data leaving their devices. **Practical Applications and Future Directions** The project encourages experimentation with timeline algorithms and aims to grow as a tool for community use. It highlights the importance of protocols over platforms and invites contributions to enhance its capabilities. ## Importance for an eco-social transformation The session underscores the significance of user-centric design in social networks, promoting privacy, transparency, and control over personal data. It aligns with eco-social transformation by advocating for open-source tools that empower users and communities. Eco-social designers can leverage BYOTA's approach to create inclusive and ethical digital environments. Challenges include overcoming technical barriers for non-experts and ensuring widespread adoption. The session highlights the need for policies that support user rights and ethical tech development. ## Slides: | | | | --- | --- | | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_001.jpg\|300]] | The slide introduces the session 'Build Your Own Timeline Algorithm' by Davide Eynard at FOSDEM on February 1st, 2025. It includes links to Eynard's Fosstodon and GitHub profiles, suggesting a focus on open-source and community engagement. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_002.jpg\|300]] | The slide clarifies what BYOTA is not: it is not a new algorithm, an end-user application, or a service. This sets the stage for understanding BYOTA as a framework or toolset rather than a standalone product. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_003.jpg\|300]] | BYOTA is presented as a 'pacifist call for arms,' inviting community participation. The imagery of the Blues Brothers suggests collaboration and creativity in developing ethical timeline algorithms. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_004.jpg\|300]] | The slide titled 'How it started' likely discusses the origins of the BYOTA project, possibly referencing the Twitter migration as a catalyst for seeking user-controlled social network solutions. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_005.jpg\|300]] | The slide dated October 26, 2022, possibly marks a significant event or turning point related to the development of BYOTA, such as the beginning of the Twitter migration. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_006.jpg\|300]] | The slide labeled '#TwitterMigration' indicates the broader context of users moving from Twitter to other platforms, which influenced the creation of BYOTA to provide user-controlled algorithmic solutions. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_007.jpg\|300]] | The slide 'Six months earlier...' likely refers to events preceding the Twitter migration, highlighting the development timeline and motivations for BYOTA. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_008.jpg\|300]] | The '#MyMigration' slide, accompanied by a salute emoji, suggests a personal narrative or decision by Davide Eynard to move away from traditional social media platforms and develop BYOTA. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_009.jpg\|300]] | The slide introduces the concept of timeline algorithms, setting the stage for discussing their role and limitations in social media platforms. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_010.jpg\|300]] | The slide explores how timeline algorithms determine post visibility, addressing issues like spam, moderation, and engagement-driven recommendations, and questioning assumptions about their necessity. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_011.jpg\|300]] | The slide references a work by Luke Thorburn et al. on platform recommendation systems, suggesting an academic foundation for the discussion on timeline algorithms. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_012.jpg\|300]] | The 'On the shoulders of giants' slide acknowledges contributors and inspirations for BYOTA, emphasizing the collaborative and open-source nature of the project. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_013.jpg\|300]] | The slide lists common problems with timeline algorithms, such as bias, lack of transparency, and privacy issues, highlighting the need for solutions like BYOTA. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_014.jpg\|300]] | The slide discusses breaking assumptions about timeline algorithms by using open models, simple methods, and local computation to enhance transparency and control. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_015.jpg\|300]] | The slide titled 'BYOTA' likely provides an overview of the project's goals and features, emphasizing user empowerment and ethical algorithm design. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_016.jpg\|300]] | The 'Architecture' slide likely outlines the technical framework of BYOTA, detailing how various tools and components interact to create a customizable timeline algorithm. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_017.jpg\|300]] | The 'Llamafile' slide describes a single-file language model optimized for local computation, used for calculating status embeddings in BYOTA. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_018.jpg\|300]] | The 'marimo' slide introduces a reactive Python notebook that can be shared as a standalone HTML file, facilitating local and browser-based execution of BYOTA. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_019.jpg\|300]] | The slide likely continues the discussion on technical components or practical applications of BYOTA, emphasizing its adaptability and user-friendliness. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_020.jpg\|300]] | This slide might expand on the technical details or use cases of BYOTA, further illustrating its potential for personalized timeline algorithms. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_021.jpg\|300]] | The slide could discuss additional features or benefits of BYOTA, reinforcing its role as a tool for ethical and user-centric social media experiences. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_022.jpg\|300]] | The 'BYOTA in practice' slide likely showcases real-world applications or demonstrations of the tool, illustrating its functionality and user benefits. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_023.jpg\|300]] | The 'Embeddings visualization' slide presents visualizations of data embeddings, demonstrating how BYOTA organizes and analyzes social media content. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_024.jpg\|300]] | The slide continues the discussion on embeddings visualization, possibly highlighting how different timelines or datasets are represented and interpreted. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_025.jpg\|300]] | This slide further explores embeddings visualization, showing how BYOTA can group and analyze social media posts to provide insights and recommendations. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_026.jpg\|300]] | The slide likely continues to discuss embeddings visualization, illustrating the comprehensive data analysis capabilities of BYOTA. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_027.jpg\|300]] | The slide provides a detailed map of embeddings visualization, creatively naming different regions to represent various types of social media content. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_028.jpg\|300]] | The slide highlights a specific example of embeddings visualization, showing how different topics, such as family and NSFW content, are grouped. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_029.jpg\|300]] | The 'Semantic search' slide introduces a feature of BYOTA that allows users to search for content based on semantic similarity, enhancing user experience. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_030.jpg\|300]] | The 'Re-ranking' slide discusses an algorithmic feature of BYOTA that allows users to reorder content based on personalized criteria, offering more control. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_031.jpg\|300]] | This slide likely continues the discussion on re-ranking, providing examples or technical details on how users can customize their content display. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_032.jpg\|300]] | The 'What about performance?' slide addresses the efficiency of BYOTA, comparing different models and showcasing its ability to perform local computations quickly. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_033.jpg\|300]] | The slide discusses the local nature of BYOTA, emphasizing that all computations, including embedding servers, run on the user's device for privacy. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_034.jpg\|300]] | The slide continues to discuss local execution, highlighting how BYOTA's components, like Marimo, operate within the user's browser for enhanced control. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_035.jpg\|300]] | This slide likely continues the discussion on local execution, reinforcing the privacy and autonomy benefits of BYOTA's design. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_036.jpg\|300]] | The 'What next?' slide outlines future directions for BYOTA, encouraging community involvement and suggesting areas for further development and experimentation. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_037.jpg\|300]] | The 'Next steps' slide lists potential improvements and research areas for BYOTA, including model comparison, new algorithmic approaches, and protocol development. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_038.jpg\|300]] | The 'Thank you!' slide provides links to Davide Eynard's social and GitHub profiles, inviting further engagement and collaboration with the BYOTA project. | ![[FOSDEM 2025/assets/Build-your-own-timeline-algorithm/preview_039.jpg\|300]] | The 'Modular' slide likely discusses the modular design of BYOTA, highlighting its adaptability and potential for integration with various models and datasets. ## Links [Talk Slides](https://fosdem.org/2025/events/attachments/fosdem-2025-5601-build-your-own-timeline-algorithm/slides/238176/20250201_FKfeAcS.pdf) [Video recording (AV1/WebM)](https://video.fosdem.org/2025/ud2208/fosdem-2025-5601-build-your-own-timeline-algorithm.av1.webm) [Video recording (MP4)](https://video.fosdem.org/2025/ud2208/fosdem-2025-5601-build-your-own-timeline-algorithm.av1.mp4) [Video recording subtitle file (VTT)](https://video.fosdem.org/2025/ud2208/fosdem-2025-5601-build-your-own-timeline-algorithm.av1.vtt) [Chat room(web)](https://fosdem.org/2025/chat/ud2208) [Chat room(app)](https://fosdem.org/2025/chat/app/ud2208)