PhD Wikis
Note: Updated version of this, and with clearer images, now on Medium at https://davidjhburden.medium.com/my-second-brain-f66b03f2a0fa.
I'm interested in the whole idea of doing a PhD “in public” and the use of Wikis and other Personal Knowledge Management tools to support them.
The most basic approach (but still novel to most) is to just use a Wiki (like this one) to track/showcase/garner comment on your PhD. Here are some examples (and with some ideas I might steal):
Digital Gardens is a term that has emerged quite recently to describe these uncurated collections of notes and thoughts on the web (unlike a more curated web site or even blog). There's a good article on it at https://maggieappleton.com/garden-history.
This is sort of related to what some of of us used to call “data gardens” in Second Life, where you'd create a 3D space/park/garden which people could wander round and it would be full of the information and data (often as 3D objects and 3D visualisations) about a particular topic. You can find an article on that at https://thinkbalm.com/thinkbalm-data-garden-is-live/. Would be great to re-produce something like that for the PhD in Frames or Mozilla Hubs so that anyone can access it on computer, mobile or HMD.
Another emerging form is the Personal Knowledge Management (PKM) system. Most people are familiar with OneNote and Evernote, but these go a step further, typically offering some form of tagging and a set of layout tools that rivals professional Wiki systems like Confluence. Systems in this category include Notion and RevNote. RevNote seems more aimed at students and revision. Notion I really liked the look of when I tried it, working on PC and iPad is great, and the layout tools are ace. It did though seem to have templates that forced structure on you that you had to then undo. The real killer thought was that it was missing the one feature I really wanted…
For ages I've played around with Ray Kurzweil's The Brain application. It looked gorgeous, with lovely dynamic force-graph driven semantic knowledge graphs of your notes, but the note taking was cumbersome and featureless and I always struggled with it. A few years ago I heard of Roam Research, which was a list driven note-taking app, but which features backlinks (so as to build the graph), block level tagging and a slightly static knowledge graph visualisation. I played with it until the demo licence expired, but again I found its enforced list approach a bit off-putting, and the graphics underwhelming, even if some of the features were top rate.
Then last week (March 2022) I decided the time had come to make a choice between Notion and Roam, but in hunting out tutorial and comparison videos (I'd recommend this and this) I discovered Obsidian.
Obsidian may not look the prettiest (at least in note taking mode) but it's the best implementation I've seen yet of this “second brain” concept. Note taking is quick and fast into a blank page, using Markdown syntax. The only thing I really miss are the tables from Notion, but there are hundreds of community extensions for Obsidian, some of which have better tables. Notes can be put in folders, but that is purely for human admin. What binds notes are the links you put in from one to another (again backlinks) and the tags. I'm already finding that taking notes is just as fast as taking them into a Word or text document, and the fact that I can do it on the iPad ideal when working one handed whilst reading a book. The key though is to enter every important concept you mention as a hashtag, as that not only links the notes in a graph, but also lets you just click on a hashtag in a note to see every other note with that hashtag!
As you build your notes the graph view build automatically.
It not only looks lovely but hovering on a node highlights all its links, helping you to see connectivity between ideas more clearly.
The one downside is that although the graph shows the tag links and note links on the same graph, and connects pages to their own tags, you do sort of have two different ontologies to manage, pages and tags. Within those community plug-ins I mentioned there are already a number that implement different network graph algorithms, so I should be able to use them to start to analyse the graph and perhaps yield new insights that I hadn't spotted.
One of the other big plusses of Obsidian is how it stores your data - it's all flat-file markdown files, one per note, on your PC. For something like a PhD with a 5-6 year life I need to know my data is safe and if the company folds I can get the data. This approach delivers that - particularly as I store my “vaults” on my Dropbox account, so they're backed up, and versioned, there. Obsidian sync then provides near real time (~2-3 seconds!) word by word sync between your PC and any other devices, and with iOS and Android apps that means you can have your whole Second Brain on your mobile device.
The proof of the pudding is probably in the fact that I've already started using to support my paid work (OK there's a fair amount of overlap topic wise).
But of course I can't stop there. Our big step forward at Daden was when we started using knowledge graphs to store the “brain” of the chatbot. I'm now dumping my PhD into a Second Brain which supports something like a knowledge graph (OK not to RDF/OWL standards but I'm sure that can be fixed). So that's got me thinking, if I crack on with my new architecture chatbot (NodeJS+ML+Grammatical Parser+SPARL+Triplestore) then can I eventually add the Obsidian knowledge graph to it - and get the bot to sit my viva for me?!
Let's see how I get on….
