A wise person once said that people always learn something, from any experience, lesson, article or talk but that quite often what they learn has nothing at all to do with what the “teacher” or organiser intended. I came away from reading a quite enjoyable and well-written article recently with thoughts and responses that were not even slightly touched on in the article itself. That helped my thinking, but was certainly not what the author intended (which is why I am not linking to the article — I am critical, but it feels unfair to criticise someone for what they did not write!)
I was reading yet another article in which someone compared personal knowledge apps (in this case Craft with Roam), which basically came down to a list of features the apps had in common, what differed and how well each feature was implemented.
This writer helpfully went further than many, talking about what he mainly used a note-making app for and so giving insight into how the features and the way they worked related to what he was trying to do.
As I read, it seemed to me that he took many things for granted without explaining or justifying them at all. For example, that writing a “daily note” every day was essential, that there was a clear and fundamental distinction between “permanent notes”, “literature notes” and various other kinds, that any links to other notes needed to show their content so you could “discover” it; that links to people referred to in a note needed to be different to links to other kinds of information; that most notes needed to be created by means of a pre-defined template. I didn’t make a complete list of these assumptions, nor did I really care whether I agreed with any of them or not, but they ended up interesting me far more than any of the points he was actually explaining and arguing well.
Assumptions and Conclusions
The conclusions this author reached (thoughtfully, reasonably and fairly, I thought) were much more to do with his assumptions, which he had only briefly mentioned, than they were much to do with the apps themselves. I profoundly disagreed with his conclusions, which seemed at odds with my experience, and I came to realise that this was because I was starting from a different point — I did not share his assumptions so would not agree with the weight he placed on various aspects and features of the two apps, and there was nothing in the article to help me question or examine my own assumptions, so what could have been a useful article, wasn’t.
I think this taking assumptions for granted approach is common in tech writing, especially articles which review or evaluate specific examples of new technology (or even new or newish apps). The most infamous examples are well known: the IBM executive who could not believe that the USA would ever need more than half a dozen computers in total or a number of industry leaders who saw microchips, personal computers, online shopping, cell phones, web sites or the internet as nothing more than another fad that would never become significant in the real world. Our assumptions about how things work, especially in a field that we know and are personally invested in, may mislead us badly into drawing the wrong conclusions unless we acknowledge and question them.
This doesn’t just apply when trying to draw conclusions about what might happen in the future. It applies to our evaluation of things we are looking at right now, as when we are trying to evaluate or decide between one or more note-taking apps. Where you are starting from (which even you might not realise) will largely determine what conclusions you draw.
One example of how powerful this is is illustrated by what people call apps like Roam or Notion or Craft or Obsidian or Apple Notes or Notability or GoodNotes or Noteshelf or any one of the several hundred competing applications in this “note-taking apps space”. In general, people who call them something like “personal knowledge management” (pkm) apps draw very different conclusions about them from people who call them something like “note-taking apps”. The language used shows they are starting from different places and so often end up in different places when they decide which to use, or try to evaluate them. This is seen most clearly when people confidently assert which are “personal knowledge apps” and which are not (often because of some feature tick-list) as if human knowledge was ever simple. I’ve even seen people say things like “I’m waiting for this app to become a pkm app, but it can’t be one because it doesn’t have this feature”. They are perfectly entitled to their opinions, but it might be more helpful to define your assumptions: what do you mean by pkm and why is that feature essential to it?
There is no easy way to deal with this, apart from the common sense approach of recognising that opinions are exactly that and no matter how well-intentioned or expert someone might be they might not be right, or they may be carrying assumptions that you don’t share. As the internet often puts it: “YMMV” (your mileage might vary).
The Black Box model
A time-honoured way of clarifying and simplifying assumptions in computing is to see a system, no matter how complex, as a black box containing a process, which can be defined by what inputs it receives and what outputs it generates from them. The days are long gone where this can often be restricted just to the level of tightly defined data (e.g. input name, output photo of that named individual). Computing systems are very often “stacks” in which more complex and valuable processes are built on layers of simpler ones. Real, possibly highly valuable, but vague things like brand value or self-esteem or personal productivity are built on layers of ever simpler things like algorithmically generated web pages or databases, which are built on scripts, html and styling, which are built on things like client-server operations, built on exchanging packets of binary data over networks.
It might be productive to clarify and simplify some discussions around “personal knowledge management” by taking the black box approach. Exactly what are we inputting into our choice of pkm system and what do we want to come out? The answers to each question are layered, of course. At the highest level, we want to input our “knowledge”, which would include our reading, thoughts we might have ourselves, examples, even details of contacts, locations and events. In turn, those things will be represented as collections of text, images, locations, dates and so on.
But what are the outputs?
But what are the outputs? This is the aspect which is least well defined, I think. There are many high-level and rather grand claims: that a pkm system can be a “second brain” or that it can avoid the need for us to remember or organise our own thought, or that it can deepen our understanding of our knowledge and clarify our ideas or that it can make us more productive. If these claims are true (and enough people value apps and systems like this, and put effort into using them to suggest that they are helpful in some way) what are the simpler levels on which they are built? What are the outputs? Is something like the “graph” of relationships between items in the pkm database the most important output, or the documents or notes or blocks and their contents themselves, or the new relationships between knowledge created in search results or in structures we have created? Is it even a more traditional kind of output like a printed or finished document summarising, clarifying and exploring what we have input, or even a list of key facts or gallery of images we have discovered?
I am not an academic, but this seems to be in need of real, deep research. I have read a lot about everything from Zettelkasten to Mind Maps and I have found utility and value in some of that reading, enough to adopt some changes to how I do knowledge work. I have found value in some apps in this area, enough to use one of them every day, but I keep running into my own assumptions being vague and ill-defined and I often find myself at cross-purposes in discussion about all this, because there are various and non-defined assumptions all around.
Add to that mix, the commercial pressures to sell apps and services or, more often, to obtain venture capital, which require innovative approaches and even claims that this particular thing is radically new and different, or that it does something unique. There is much more interest in new, shiny and exciting rather than coherent, effective and worthwhile and it is harder and harder to see the wood for the trees (or the reality for the hype).
I’ve spent many hours of my past life telling computing students to solve problems by clarifying in more detail the outputs required for given inputs to the systems they are developing. I have a feeling that it is way past time to start doing that for apps like Roam, Craft, Obsidian, Evernote and many more, even if only as a better way of clarifying assumptions about what it is what we want from apps — which ultimately is the only way to compare and decide which is best suited to our needs.