The Circle and the Squiggle
There is a moment, familiar to anyone who has ever fallen down an internet rabbit hole, when a stray link leads somewhere unexpected. A Wikipedia article about Byzantine architecture leads to a recipe for Ottoman Empire-era dessert. A late-night search for acoustic guitar tutorials surfaces a documentary about the last luthier in rural Japan. These moments feel like luck, but they are also, arguably, a kind of design problem.
In the early 2010s, a developer began asking a question that sounds almost paradoxical: what if serendipity could be engineered? Not manufactured, exactly, but cultivated through a combination of human judgment and algorithmic pattern-matching. The answer became The Serendipity Engine first a paper form and a squiggle drawing, later an open-source global discovery network, and now an academic research project at Vrije Universiteit Brussel that has been quietly building tools for unexpected discovery for more than a decade.
The project began with a deceptively simple premise: computers are extraordinary at cross-referencing information in ways the human brain cannot manage, which makes them inherently better at making connections and making connections is a prerequisite for serendipity. But computers lack the ability to assess human qualities like creativity or physical attractiveness. Those require human judgment. So the original Engine combined both: automated keyword extraction paired with human-powered assessment through Mechanical Turk, Amazon's online marketplace that farms out small tasks to distributed workers around the world.
How the Engine Works
The first version of the Serendipity Engine was created with Kat Jungnickel and tackled the foundational question of what serendipity actually is. The second version, built with Ben Hammersley, attempted to answer the harder question: how might serendipity be produced by a digital technology?
The process starts with a paper form. Users are asked to draw a circle, create a drawing from a squiggle, and photograph their own face. These artifacts are then sent to Mechanical Turk, where workers ordinary people working from their homes, offices, and bedrooms around the world are paid a small sum to evaluate them. According to the project's own documentation, someone somewhere in the world was paid $1.50 to look at each submitted photograph and assess it across four dimensions: how the circle compares to a Japanese ensō (a symbol representing universality, openness, and creativity), the creativity of the squiggle-based drawing, physical attractiveness, and psychological well-being.
Simultaneously, an online form feeds the Engine two types of information: keywords identifying what should be relevant and valuable to the user, and a personal "Serendipitousness" rating measured across seven scales. The form splits into three sections. The first four questions authenticate the user by cross-referencing basic demographic information against a Google search of their name. After identifying the user in search results, the Engine extracts keywords from the top three results that define who they are in public. Social media handles are used to identify what is most relevant to them right now, by extracting nouns from the ten most recent status updates on Twitter, Facebook, or Google+. The remaining questions extract keywords about the user that online databases are unlikely to have on record questions about favorite subjects in school or personal preferences that exist only in memory.
The result is a personalized Serendipity Recipe: a curated set of connections and keywords designed to surface unexpected but relevant discoveries. It is, in essence, a machine for manufacturing the conditions of luck.
From Personal Experiment to Open-Source Project
The Serendipity Engine has evolved significantly since those early paper forms. The current open-source project frames information discovery not merely as a technical challenge but as a matter of global equity. A 2025 white paper titled "Breaking Information Apartheid" articulates the problem in stark terms: the current information discovery system, the authors argue, creates artificial scarcity in an age of abundance. Search engines, they contend, operate on the assumption that users already know what they are looking for a fundamental design flaw that traps curiosity within the boundaries of existing knowledge.
The white paper describes what it calls "The Search Bar Prison." Every search begins with a query, which means every search assumes prior knowledge. This creates a paradox: the most valuable discoveries are, by definition, the things you did not know to search for. The document estimates that 130 trillion web pages are artificially fragmented, 7.8 billion people are limited by IP restrictions, and more than 6,000 languages create what the authors call "knowledge islands" vast territories of human understanding that remain inaccessible through conventional search.
The proposed solution is an open-source, multi-network discovery system that transforms isolated search queries into collaborative expeditions across the full spectrum of human knowledge. The platform acts, according to its IndieDB profile, as "a digital telescope for curiosity, surfacing content users rarely find via conventional search engines." It is lightweight and privacy-first, using a globally distributed crawler network and IPFS storage to bypass geographic blocks while preserving anonymity through Tor, I2P, and end-to-end encryption.
The project is open-source under the GNU GPL-3.0 license, inviting developers, designers, security researchers, translators, and anyone passionate about free information to contribute. Contributors can shape the architecture, improve the user interface, add multilingual support, or audit the security model. The live demo runs on GitHub Pages, and the full source resides in a public GitHub repository. Funding comes from community donations, multi-national grant discovery and application assistance, and a legal-defense fund protecting the project's independence.
The Academic Dimension
While the open-source project develops practical tools, an academic research program has grown alongside it. The Serendipity Engine project is based at Vrije Universiteit Brussel and funded by the Research Foundation - Flanders (FWO) under grant number S006323N. The research team including Annelien Smets, Lore Binst, Lien Michiels, and Umberto Maes has produced a substantial body of work examining serendipity in recommender systems, user modeling, and platform urbanism.
The publications list from the project's website reveals a serious academic engagement with questions of discovery, recommendation, and human curiosity. A 2025 paper in Ethics and Information Technology titled "Intended, Afforded, and Experienced Serendipity: Overcoming the Paradox of Artificial Serendipity" directly addresses the central challenge the project faces: how can designers create conditions for unexpected discovery without undermining the very serendipity they seek to foster? The paradox is that serendipity, by definition, cannot be fully planned or engineered yet the project is premised on the possibility of creating the conditions for it to occur.
Another 2025 publication, "What is Serendipity? An Interview Study to Conceptualize Experienced Serendipity in Recommender Systems," takes an empirical approach, conducting interviews to understand how users actually experience unexpected discoveries in recommendation contexts. A 2026 paper titled "Beyond Rational Choice: A Typology of Urban Discovery" extends this thinking into physical space, examining how people navigate cities in ways that cannot be reduced to optimal path-finding.
The team has also produced practical tools from this research. The Serendipity Cards translate the idea of designing for serendipity into an accessible tool for workshops and ideation sessions. These cards offer a concrete, hands-on method for practitioners to apply academic insights about unexpected discovery to their own work.
In October 2025, the team published the Handbook of Platform Urbanism through Edward Elgar Publishing, edited by Annelien Smets and Pieter Ballon. The handbook represents a broader effort to understand how digital platforms shape urban life, with serendipity and discovery as recurring themes throughout.
Why This Matters for Lnk2It Readers
For readers interested in link curation and resource discovery, the Serendipity Engine offers both a philosophical framework and practical methods worth understanding. The project sits at the intersection of several concerns that Lnk2It covers: how to surface valuable resources that users do not know to search for, how to balance algorithmic efficiency with human judgment, and how to design discovery systems that serve consciousness expansion rather than narrow engagement metrics.
The original Engine's use of Mechanical Turk to incorporate human assessment into algorithmic processes is a model for how link curators might think about quality and relevance. Rather than relying solely on automated ranking signals, the Engine suggests that human judgment even judgment as seemingly subjective as assessing creativity from a squiggle drawing can be meaningfully integrated into discovery systems.
The open-source project's emphasis on privacy, anonymity, and decentralized storage also speaks to concerns about how discovery systems can be designed to serve users rather than extract value from them. In a landscape dominated by platforms that hoard curiosity patterns and knowledge gaps for profit, the Serendipity Engine proposes an alternative: collaborative discovery networks that belong to the communities they serve.
For practitioners building link collections, directories, or recommendation systems, the Serendipity Engine's research on the paradox of artificial serendipity is particularly relevant. The team's finding that intended serendipity often fails to produce experienced serendipity that users do not always recognize or appreciate the discoveries that algorithms surface for them is a reminder that discovery is not merely a technical problem but a human one.
The Brussels Event and What's Next
The project has marked a significant milestone on its calendar: October 27, 2026, when the Serendipity Engine will hold an event in Brussels. This gathering represents an opportunity for researchers, developers, and practitioners to come together around questions of discovery, recommendation, and the future of information access. The event builds on the team's existing publication record and workshop tools, creating a space for the academic and practical dimensions of the work to intersect.
The Serendipity Engine's trajectory from a personal experiment with paper forms and squiggle drawings to an open-source global discovery network and academic research program offers a model for how small, curious projects can grow into something larger without losing their original impulse. The core question remains the same: how do we create the conditions for unexpected discovery in a world increasingly shaped by algorithms designed to show us more of what we already know?
The answer, the project suggests, lies in combining the strengths of machines and humans the processing power of algorithms with the judgment of real people and in building systems that serve curiosity rather than constraining it. Whether that answer scales to 130 trillion web pages and 7.8 billion people remains to be seen. But for anyone who has ever been grateful for a link that led somewhere unexpected, the Serendipity Engine offers a framework for thinking about how that luck is made.
Where to Read Further
Those interested in exploring the Serendipity Engine's work in more depth can start with the project's official homepage at Vrije Universiteit Brussel's Serendipity Engine project site, which includes information about the team, upcoming events, and practical tools like the Serendipity Cards. The "Breaking Information Apartheid" white paper provides the project's full philosophical and technical framework, including the analysis of current search limitations and the proposed alternative. For the academic research underlying the project, the publications page lists the team's papers on serendipity in recommender systems, urban discovery, and platform urbanism, including the 2025 Ethics and Information Technology article on the paradox of artificial serendipity.
| Key Milestones in the Serendipity Engine's Development | | --- | | Early 2010s First version created with Kat Jungnickel, addressing the question "what is serendipity?" | | Early 2010s Second version created with Ben Hammersley, addressing "how might serendipity be produced by digital technology?" | | November 2011 Aleks Krotoski presents serendipity and the web on BBC Two's The Culture Show | | 2011-2012 Research published as part of Nominet Trust's "The Personal (Computer) is Political" report | | 2025 Open-source project launches with privacy-first architecture using IPFS, Tor, and I2P | | 2025 "Breaking Information Apartheid" white paper published | | 2025 Handbook of Platform Urbanism published through Edward Elgar Publishing | | 2025 "Intended, Afforded, and Experienced Serendipity" published in Ethics and Information Technology | | October 27, 2026 Serendipity Engine event in Brussels |