Serendipity for Serendipity
It all started with frustration, which motivated efforts to find a solution. The solution worked. More than that, it turned out to be an occasional generator of serendipity events. Serendipity wasn't rare in my life. What was different this time was that it caused a serendipitous episode about serendipity itself—about making serendipity happen more frequently.
When the pandemic started at the beginning of 2020, I moved to Roam Research as my main knowledge management tool. Roam Research was one of the catalysts for the explosion of tools for personal knowledge management (PKM). From barely a dozen PKM tools in 2019, they exceeded 70 within the next two years. Not only did their number grow fast, but so did the amount of innovation the PKM toolmakers came up with. One interesting trend was the transition from advanced note-taking apps towards tools for thought.
That and Roam's graph nature was what attracted me at first. Cultivating a serendipity generator, like Luhmann’s Zettelkasten, was a long-time pursuit, but I wasn't satisfied with what I could achieve with the previous tools. Now, it was just one of the many things that could be done effortlessly. Linked with that, my appreciation of Roam as a knowledge graph came from a decade of working with and experiencing the benefits of enterprise and open knowledge graphs.
What I failed to achieve with previous tools happened naturally with Roam, so it was no surprise that, after replacing my previous PKM tools, it also replaced many non-PKM tools I've been using. I have shared some of what fascinated me about Roam in a series of articles, the last one explaining how I use it. Yet, this was before I adopted the practices, which led to a more serious study and experimentation with serendipity.
Serendipity comes to the prepared mind, Pasteur said, and that meta-serendipity accident would have gone unnoticed if there hadn't been a long-time engagement with serendipity already. Serendipity drove my interest in personal knowledge graphs, independent of specific tools. This led to the book on Personal Knowledge Graphs, where in the first two chapters, from different perspectives, I tried to explain why PKG are conducive to serendipity, in what way they can start having a life of their own and how that’s what enables them to surprise their creator.
When using Roam for research, managing projects, productivity, collecting and linking annotations, writing, training courses and workshops (not just preparing and managing but also as a replacement of presentation tools like PowerPoint and GoogleSlides), tracking expenses, repairs, and finding the right grinding levels for different coffee beans, it's not surprising that it quickly grew. With the growth came the problems and the frustration. The bulk wasn’t only due to collecting and linking valuable things. I wasn’t immune to what I kept warning others about: information hoarding. The result was clutter, a lot of noise in search results, and the tool working slower every month.
That frustration was maybe enhanced by my aversion to clutter of any kind. I find 99% of the web pages too busy and 99% of the tools badly designed for having way too many menus, icons, and — that appalling Microsoft invention — ribbons (to me, Ribbons feel like Jabba the Hutt is taking the upper part of the screen). And so naturally, another thing that attracted me to Roam was its minimalistic design. That was still far from sufficient but with the help of
’s wonderful themes and, later, Roam Studio and Auto Hide, I cleaned the residual clutter. Yet, while I achieved an uncluttered interface once and for all, uncluttering the graph required continuous effort.At first, I scheduled clean-up sessions, where I'd spend some controlled amount of time reviewing pages that were not referenced, old or met some combination of criteria. That had an effect but required effort. I wasn't able to keep up with this practice. I tend to avoid whatever needs maintenance or any overhead. That attitude is additionally enhanced by many years spent, in another life, optimising processes, identifying and clearing waste. Not only for clients. Speaking of clutter, at one point, I went as far as putting red markers on objects at home, mainly devices and tools, to put them away if I see that they have not been used for a long time (that came from the Japanese seiri practice, one of the 5S). And apart from finding maintenance wasteful, to me, it simply doesn't feel right to serve a tool. The tool should serve me and somehow disappear when doing so. In Heideggerian terms, it should be ready-to-hand, not present-at-hand.
If some maintenance is necessary, it should be minimised and made fun to do. This is how I decided to include a new automated practice, which I called graph pruning.
Maybe the digital gardening metaphor inspired the name. But then, look at it! Doesn’t it look like a bush that went into jungle mode in the spring and now urgently needs pruning?
Technically, the graph pruning practice is supported by a simple functionality that brings a set of random pages to my daily note page. In Roam, that is done using the SmartBlocks extension for building custom workflows. A popular type of Smart Block is one that automatically runs every day so that your daily note page, called journal in other tools, is populated with events, meetings and routines along with whatever nodes of the graph would appear by having been tagged with that date. In other words, the daily notes page is pre-populated with expected and surprising items. These items range from meetings and tasks, through messages from your past self, to items for prompting an action or memorisation, controlled by space repetition algorithms.
And so, one new item on my day page was this graph pruning block, which contained a number of randomly selected pages to review. That didn't take much time and was attractive to do because of not knowing what the new selection would be. I'd expand the list in the sidebar and, one by one, open the randomly selected pages in my main working area for review. If that was something not needed anymore, I’d delete it. That was the main value of the routine. Sometimes, a page would be valuable and kept, but removing unnecessary blocks or re-grouping them was still graph pruning, just cutting smaller branches. And then, from time to time, a serendipitous event happened. A forgotten page would appear that unexpectedly held or pointed to the key of a problem I was working on. Or it might be a page of a person I haven't seen for a long time or forgotten that they were involved in something that I happened to have a note about. That note looked insignificant at the time and hadn't stayed in my memory, but it gained relevance in the current context. On some days, it was a matter of a block note appearing as part of some workflow on that particular day, which, when seen next to a randomly pulled page, reviewed as part of the graph pruning routine, made sense for something valuable and unexpected.
However, serendipity happens anyway when using well-designed tools for thought. That the pruning practice brought fortuitous accidents on some days wasn't a revelation. It was only an additional motivation to keep and improve this practice, and nothing more than that, until I started tinkering with this workflow.
I tried varying the number of random pages fetched, the days of the week it will be run, and where it will appear on the daily note page. I'll give more details about this in a future essay. What's worth mentioning now is that I slowly increased the number of pages I'd tolerate for a review session, which means bringing down when reaching a threshold that made me skip the routine. It stabilised at a number bigger than the initial. To my surprise, at that point, there was something like a phase transition, and the graph pruning routine started producing more frequent serendipitous events.
I was quite pleased with that, and I continued improving it and started experimenting with other techniques to enhance the chance for serendipitous events (that’s how practices like ideas dating and reverse prompting were born). Yet, that surprising by-product of my graph pruning technique did not motivate me to start seriously working with serendipity. It was an event that marked a serendipitous episode I would later call a pair pattern (that, and two other patterns, penny and path, I’ll explain in a forthcoming essay). It was like stumbling upon a lost left shoe. The right shoe was a Google Scholar alert mail. It brought a paper that, for the first time, looked at serendipity from an understanding of cognition that I share. The two accidents were the catalyst for systematically working on serendipity. The goal of this work is creating serendipity framework and protocols, independent of particular tools and platforms.
In this series, I'll share perspectives, patterns, and practices that may help increase the frequency of serendipitous events. Such events depend on experience, curiosity, risk tolerance, playfulness, sensitivity, and attention training. Yet, a lot depends on the affordances of the environment. This quest is driven by the belief that the digital, workspace and urban environments can be designed to increase the likelihood of unexpected, beneficial discoveries and encounters, not only by stimulating unusual combinations to happen but also by influencing the behaviours of agents to provoke them and their ability to notice them.
If you would like to contribute, please share your stories of serendipity. Another way to support this work is to become a paid subscriber.