Kaggle Chronicles: 15 Years of Competitions, Community and Data Science Innovation

Kevin TheItCrow Bönisch, Leandro BwandoWando Losaria · 2025

Read Kaggle Chronicles: 15 Years of Competitions, Community and Data Science Innovation by Kevin TheItCrow Bönisch, Leandro BwandoWando Losaria (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-06304.txt.

Introduction

Kevin Bönisch and Leandro Losaria chronicle fifteen years of Kaggle, the data science competition platform, from 2010 onward. Kaggle grew from a contest site into an ecosystem of forums, shared notebooks, models, and datasets. Newly released platform metadata now allows a data-driven history of that growth.

The study mines millions of shared code notebooks and discussion threads for long-run trends. It tracks how the community grew, which technologies rose and fell, how winners approach problems, and how quickly members adopt new tools for real-world tasks. The findings portray a steadily growing, diversifying platform whose users build models that generalize solidly on average.

Catalog record

CREATORKevin TheItCrow Bönisch, Leandro BwandoWando Losaria
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE71 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-06304

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Kaggle Chronicles: 15 Years of Competitions, Community and Data Science Innovation Kevin TheItCrow Bönisch, Leandro BwandoWando Losaria Abstract Since 2010, Kaggle has been a platform where data scientists from around the world come together to compete, collaborate, and push the boundaries of Data Science. Over these 15 years, it has grown from a purely competition-focused site into a broader ecosystem with forums, notebooks, models, datasets, and more. With the release of the Kaggle Meta Code and Kaggle Meta Datasets, we now have a unique opportunity to explore these competitions, technologies, and real-world applications of Machine Learning and AI. And so in this study, we take a closer look at 15 years of data science on Kaggle–through metadata, shared code, community discussions, and the competitions themselves. We explore Kaggle’s growth, its impact on the data science community, uncover hidden technological trends, analyse competition winners, how Kagglers approach problems in general, and more. We do this by analyzing millions of kernels and discussion threads to perform both longitudinal trend analysis and standard exploratory data analysis. Our findings show that Kaggle is a steadily growing platform with increasingly diverse use cases, and that Kagglers are quick to adapt to new trends and apply them to real-world challenges, while producing–on average–models with solid generalization capabilities. We also offer a snapshot of the platform as a whole, highlighting its history and technological evolution. Finally, this study is accompanied by a video and a Kaggle write-up for your convenience. Created as part of the Meta Kaggle Hackathon Team Bierbingka 1 Introduction TL;DR We begin each section with a TL;DR ( Too long; didn’t read ) box, which provides a rough summary of the insights presented in that section. These summaries are not intended to replace the full content, but rather to serve as a skim guide for identifying the most relevant sections. Kaggle was founded in 2010 by Anthony Goldbloom and Ben Hamner with the vision of creating a platform where data scientists from around the world could compete to develop the best possible solutions–measured by a specified metric–to a given problem within a limited timeframe, driven by the motivation of earning medals, recognition, and prize money. Over time, Kaggle has evolved far beyond these competition-focused roots. With the introduction of dedicated discussion forums, integrated notebooks, public datasets, and most recently, model repositories, it has transformed into a comprehensive data science ecosystem. This growth encouraged a continually expanding community, making Kaggle not only a place for competition but also a hub for learning, collaboration, and innovation in data science. When Google acquired Kaggle in 2017, this development was further accelerated through greater integration with the Google ecosystem and infrastructure. Meanwhile, the field of data science has undergone massive transformation. Machine learning and AI exploded in popularity across industry, daily life, and academic research [ 12 , 2 ] . At the same time, the demand for computational resources has skyrocketed–particularly with the rise of (large) language models, which continue to grow in size and complexity [ 13 ] . For example, recent studies show a dramatic increase in AI-generated content online, with a 57,3 % rise on mainstream websites and an alarming 474 % rise on misinformation sites [ 10 ] , alongside growing concerns around plagiarism [ 3 , 8 , 14 ] . This trend is not limited to language technologies–AI adoption is also accelerating across healthcare, finance, entertainment, and transportation, powered by machine learning algorithms, vision models, and time series forecasting [ 5 ] . Critically reflecting on this brings us to a central–yet often overlooked–necessity: one of the major challenges in managing a technological boom such as AI is the task of accurately measuring and testing its capabilities . That is, translating research and potential breakthroughs into rigorous, real-world application to answer questions such as: How effective is AI for specific tasks? What are its boundaries–not just in theory, but in practice? And what real-world problems can it meaningfully and reliably solve? Blindly deploying state-of-the-art AI technologies in sensitive or high-stakes domains–such as financial forecasting, education, information retrieval, or medical diagnostics–without understanding their actual capabilities can lead to unintended and potentially harmful outcomes. To mitigate this, researchers have turned to benchmark development [ 15 , 11 ] , controlled experiments, and systematic evaluations across domains [ 16 , 6 ] . Yet within this landscape, Kaggle stands out as a uniquely valuable case study: over the past fifteen years, it has gathered a global community that advances AI through open competitions, shared notebooks, public discussion, and collaborative experimentation. With the release of Kaggle’s very own meta-datasets –Meta Kaggle and Meta Kaggle Code–we are now equipped to analyze Kaggle itself, providing a rare opportunity to reflect on how the platform and its community have contributed to the evolution of AI and Data Science as a whole. And so in this study, we take a data-driven look at Kaggle’s development, examining how its users, forums, notebooks, and competitions have collectively helped shape the data science landscape–and how this insight can inform the future of empirically-driven AI development. We follow a general research paper structure, but do not fully adhere to its conventions. Throughout this work, we frequently cite non-traditional sources–such as Kaggle notebooks (often enough created by us for this specific purpose, and cited like: [ * 1 ] ) or other forms of evidence–to support our claims. It is also important to note that this work was conducted within the scope of the Meta Kaggle Hackathon , under a limited time and resource frame. As such, we encourage readers to explore the cited resources directly to validate, extend, or build upon our findings and to acknowledge, that this is not a peer-reviewed scientific paper. 2 Kaggle User Growth TL;DR We’ve analyzed the growth of users in Kaggle over the years and have identified spikes in the daily counts in the registration of users. By identifying the spikes in new users registering, and not just the organic growth over time, we can gauge how influential Kaggle is in the datascience space and how effective is its democratization through associated events, competitions, and collaborations with other platforms. Kaggle was launched in April 2010 as a small platform hosting data science competitions, aiming to connect organizations with skilled individuals to address real-world problems and promote the democratization of data science. Over time, it evolved into a global community where users, regardless of background, can access datasets, share code, and learn from one another. Today, Kaggle plays a significant role in making data science more open and accessible by supporting skill development, career growth, and collaborative problem-solving. The platform also provides free access to high-performance tools such as GPUs, enabling users to run advanced models and experiments with ease. By examining spikes in new user registrations, beyond the platform’s steady organic growth, we can assess Kaggle’s influence within the data science community and evaluate how effectively it promotes democratization through key events, competitions, and partnerships with other platforms. Some questions that we would like to answer with the analysis. 1. When did key Kaggle membership milestones occur (e.g., 1M, 5M, 10M members)? 2. What specific dates show significant spikes in user registrations? 3. Which events, both within

Read the complete text