[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-81694-en":3,"doc-seo-81694-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},81694,1649267921044,"Ava Thompson","https://us-avatar.wpscdn.com/avatar/1800007509477c92dfb?_k=1782875107921204101",8,"Research & Report","Do Recommendation Algorithms Work When Users Are LLM Agents A Case Study on Moltbook","Large language model (LLM) agents are increasingly active on web platforms, challenging a core assumption of recommender systems: that users maintain stable, learnable content preferences. This study investigates forum recommendation on Moltbook, a large-scale social media platform for autonomous agents running on OpenClaw. Nine recommendation approaches are evaluated for predicting the next forum an agent will engage with. Results show that popularity-based rules and item-side collaborative filtering outperform models that learn user representations, while static persona descriptions add little value. Evidence indicates Moltbook agent consumption differs from patterns in human recommendation datasets, informing robust algorithm design.","Do Recommendation Algorithms Work When Users Are LLM Agents? A Case Study on Moltbook  \nDaming Li  \nIndependent Researcher Mountain View, CA, USA[damingliyale22@gmail.com](damingliyale22@gmail.com)  \nSimeng Han Stanford University Palo Alto, CA, USA [shan6@stanford.edu](shan6@stanford.edu)  \nJialu Zhang∗ University of Waterloo Waterloo, ON, Canada [jialu.zhang@uwaterloo.ca](jialu.zhang@uwaterloo.ca)  \narXiv :2606 .29762v2 [ cs .IR] 10 Jul 2026  \nAbstract  \nLarge language model (LLM) agents are increasingly populating web platforms, raising a fundamental question for recommender systems: do algorithms designed for human users still work when users are LLM agents that may not have well-defined content consumption preferences? We study this question by formulating a forum recommendation problem on Moltbook, a large-scale social media platform exclusively for autonomous AI agents running on the OpenClaw framework. We evaluate nine recommendation methods spanning simple heuristic rules, matrix factorization, itemand user-based collaborative filtering, graph-based, and sequential models on the task of predicting which forums an agent will engage with next.  \nWe find that simple popularity-based rules or item-side collaborative filtering leveraging the platform and item structural information outperform techniques that explicitly learn a user representation. The static agent persona descriptions, the closest analog to a preference profile, fail to add value in predicting engagement. These results suggest that, on Moltbook, recommendation depends more on platform-and item-level structural signals than on user-specific personalization. We present multiple lines of empirical evidence that the observed content consumption patterns on Moltbook differ from well-established findings on human recommendation datasets, providing a new angle for studying agent societies and designing robust recommendation algorithms as agents increasingly populate the web.  \nCCS Concepts  \n• Information systems → Recommender systems; World Wide Web; • Computing methodologies → Multi-agent systems.  \nKeywords  \nRecommender systems, Moltbook, multi-agent systems, collaborative filtering, personalization, user modeling, social networks  \nACM Reference Format:  \nDaming Li, Simeng Han, and Jialu Zhang. 2026. Do Recommendation Algorithms Work When Users Are LLM Agents? A Case Study on Moltbook. In . ACM, New York, NY, USA, 11 pages. [https://doi.org/10.1145/nnnnnnn](https://doi.org/10.1145/nnnnnnn). nnnnnnn  \n∗ Corresponding author.  \nConference’17, Washington, DC, USA  \n2026. ACM ISBN 978-x-xxxx-xxxx-x/YYYY/MM [https://doi.org/10.1145/nnnnnnn.nnnnnnn](https://doi.org/10.1145/nnnnnnn.nnnnnnn)  \n1 Introduction  \nRecommender systems have been a cornerstone of modern web platforms for the past two decades, enabling users to discover relevant content by learning from their behavioral history [4, 33, 51] . From early collaborative filtering [18, 25, 30, 60] to modern deep learning [22, 34, 75] and transformer-based sequential models [29, 61, 70, 79], the field has made remarkable progress. The assumption underlying all these advances is that users possess latent yet learnable preferences that persist across interactions and evolve gradually over time. Human users’ interests tend to be consistent within sessions and drift slowly across weeks or months [5, 7, 64], enabling increasingly sophisticated personalization.  \nHowever, a new class of web users, autonomous LLM agents [44, 65], is emerging that may weaken or alter this assumption. These agents, powered by LLMs, are increasingly deployed on web platforms, where they perform tasks and consume content alongside human users [15, 39, 57, 78] . Still, it is unclear whether they possess a clear content preference profile. Many deployed agent architectures [1, 58, 69, 73] have limited or no persistent memory across sessions—each conversation turn may reset the agent’s context, and even agents configured with external memory t","cbCaicbhmFChkTG0","https://ap.wps.com/l/cbCaicbhmFChkTG0","pdf",728974,3,1,11,"English","en",105,"# Introduction\n## Problem motivation\n## Empirical study on Moltbook\n# Methodology\n## Recommendation task formulation\n## Candidate recommendation methods\n# Results and findings\n## Performance comparisons\n## Role of personalization and persona descriptions\n# Conclusion and implications\n## Robust recommendation for agent-populated platforms","[{\"question\":\"What core assumption of recommender systems is challenged by LLM agents?\",\"answer\":\"Recommender systems typically assume users have latent yet learnable preferences that persist and evolve slowly. The document questions whether LLM agents have a clear, persistent preference profile across sessions.\"},{\"question\":\"How does the study set up the recommendation task on Moltbook?\",\"answer\":\"It formulates forum recommendation as predicting which submolts an agent will engage with next, using the Moltbook Observatory Archive dataset to capture agent posts and interactions over an observation period.\"},{\"question\":\"Which recommendation approaches work best on Moltbook and why?\",\"answer\":\"Simple popularity-based rules and item-side collaborative filtering that leverage platform and item structural information outperform methods that explicitly learn user representations. Static agent persona descriptions do not significantly improve engagement prediction.\"}]",1784175452,28,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"do-recommendation-algorithms-work-when-users-are-llm-agents-a-case-study-on-moltbook","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":20},"https://docshare.wps.com/document/research-report/",{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/do-recommendation-algorithms-work-when-users-are-llm-agents-a-case-study-on-moltbook/81694/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-25","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What core assumption of recommender systems is challenged by LLM agents?","Question",{"text":75,"@type":76},"Recommender systems typically assume users have latent yet learnable preferences that persist and evolve slowly. The document questions whether LLM agents have a clear, persistent preference profile across sessions.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the study set up the recommendation task on Moltbook?",{"text":80,"@type":76},"It formulates forum recommendation as predicting which submolts an agent will engage with next, using the Moltbook Observatory Archive dataset to capture agent posts and interactions over an observation period.",{"name":82,"@type":73,"acceptedAnswer":83},"Which recommendation approaches work best on Moltbook and why?",{"text":84,"@type":76},"Simple popularity-based rules and item-side collaborative filtering that leverage platform and item structural information outperform methods that explicitly learn user representations. Static agent persona descriptions do not significantly improve engagement prediction.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]