jb… a weblog by Jonathan Buys

Personal Information Architecture

For my computer to be useful to me I need to be able to quickly save information, and then easily retrieve it later. Saving and retrieving information sounds like a simple enough use case, but doing both quickly and easily does not. The more information you have saved on your computer, the more difficult it is to effectively retrieve the information you need the moment you need it. Researchers and developers have been tackling this issue for decades, but so far no one has come up with a single best solution that works for everyone. What we need is a way to store and retrieve information without having to stop and think about the method or means of organization. The organizational method should provide an effective affordance without resorting to decoding the method itself.

The first organizational method I tried used a hierarchical folder structure based on dates and file types. For example, inside the “2002” folder, I had a folder for each month, and inside the “January” folder I had a folder for Word documents, and another for Excel, another for text, and so on. However, there was no naming system for files, and no way to associate related files for different projects. It was more of an archive than a persistent and useful file organization system.

I purchased the first version of Yojimbo when it was released, and used it heavily in conjunction with the folder structure. Yojimbo let me quickly save information, and provided a simple search box for retrieving it later. In my opinion, the best feature of Yojimbo was the quick entry window I had mapped to F1. Yojimbo was smart enough to recognize the data I had on the clipboard, and presented an appropriate template filled out with the data. For example, if I copied a URL, Yojimbo would offer to save a bookmark or a web archive. If I copied a serial number, Yojimbo would present the software screen filled out with the serial number and ask for the name of the software. Yojimbo worked well for me up until the iPhone was released, and I realized I wanted ubiquitous access to my data. I contacted Bare Bones several times to ask about an iPhone companion, but the answer was always that there were no plans for one.

Another important change in thinking happened about this time. I began to care quite a bit about the format of my data, and the ability of my data to remain in a retrievable format over time. Yojimbo worked out well for this, since with it I was able to export the data exactly as I put it in. However, the most attractive competitor to Yojimbo, Evernote, did not respect the original format. The third player in the information management apps is DEVONthink.

DEVONthink is a beast of an application with some very special, specific use cases. It is similar to both Evernote and Yojimbo in that it is an information organizer, but where it departs is how deep into the information DEVOnthink goes. DEVONthink includes an form of AI that can help you sort and search your information, but to help the AI engine along, the developers suggest separating your data into several topic specific databases. DEVONthink then lets you open and close individual databases and sync between Macs or the iOS companion application, DEVONthink To Go. I recently gave DEVONthink another try for a few days, and found it a worthwhile experience. DEVONthink helped me get a new perspective on the information I keep, and in doing so I found that I didn’t really need an application with this much power.

DEVONthink suggested that I separate everything out into topics, which could be viewed as top-level folders. Inside each folder could be several groups, and groups within groups. After using DEVONthink for a few days, the similarities between it and the Finder were obvious, and for my use case, DEVONthink became too limiting. It is an application that can handle almost everything, but not quite. Anything that needs its own folder structure cannot live inside DEVONthink, so I eventually wind up maintaining two distinct folder hierarchies. One of databases and groups inside DEVONthink, and one of standard folders in the filesystem. I realized, again, that if I wish to simplify my information storage and retrieval, I needed a system that could accommodate everything. The only system that fits the bill is maintaining a nested folder structure in the filesystem.

I have tried several file naming conventions, including the The Bunsen Method. Seth Brown is extremely intelligent, but I found his method a bit boggling. Keeping all of your files in one of three folders assumes two things: 1, that you are going to be able to remember when a file was created, and 2, that you are accessing the file using an interface convenient enough to accommodate the obscure file name. My experience when faced with a wall of text, all of it starting with a long string of numbers was that I was confused, and had trouble scanning for what I was looking for. My eyes would instantly scan past the date, time, and token fields and straight to the title. What I decided on was much closer to David Sparks’ method, using a simple date stamp followed by a descriptive name.


The folder hierarchy itself is based off of areas of my life. For example, I have a top level folder in Documents for work, and another for home, and a third for Farmdog. Inside each of those are project specific folders, for example inside home I have one for Appliances, another for Oliver (our dog), and another for Finances. Separating out my files this way provides easy backups, separation (in case I ever get a new computer for work), and an easy way to quickly identify where a file might belong. Where the folder hierarchy falls apart is when I avoid stopping to think about what the information I’m saving is and what use it might be to me. When this happens what I used to do was save it to the Desktop, or to the top level of my Documents folder with a default name and forget about it. To avoid this situation, I use a top level folder named “Inbox”, and Hazel. Hazel lets me define what types of files I might have ahead of time, and define rules for what the files should be named and where to put them in the folder structure. I’ve adopted this system from David Sparks’ book Paperless, and for the most part it works well. I set a weekly reminder in OmniFocus to sort through the Inbox folder and update the Hazel rules accordingly, or simply throw the files where they need to be.

The folder hierarchy and naming system I use allows me to think about where I am, what I am working on at the moment, and what I need. I find this structure far better suited to my computer use than thinking about what search terms I should use to find what I am looking for.

But what about ubiquitous access? I could use Dropbox for a portion of the data, but I have found that my needs for information away from my computer are actually so small that I don’t need all of it. What I want is not really what I need, and until what I want is available in to me in an acceptable format, I’m not going to worry about it. The closest thing I’ve seen to exactly what I would like is the Transporter, I’m considering picking one up.

Even with a standard naming convention, an environment and topic based folder hierarchy, and Hazel rules enforcing the standard, there are still things that I find that I want to keep. Just stuff. For this, I’ve gone back to saving random bits of things in Yojimbo. I’m not sure how long I’ll keep this up, I may wind up exporting all the notes into the filesystem and letting Hazel sort them out, but for now I’m enjoying the speed and ease of using Yojimbo. Yojimbo is looking a bit long in the tooth, I hope to see an update soon.

The unlimited flexibility of the filesystem is both a blessing and a curse. You can arrange your data in any way you choose, and save any type of information you wish, but doing so without strict guidelines on naming conventions and placement within a ridged folder structure will cause clutter and lost information. Many people find that relying on a third party application like Evernote or DEVONthink to be perfect for their needs, but I found both of them far too limiting for my needs. Giving up access to data on my iOS devices in return for the secure knowledge that there is a place for everything, and everything is in its place, is a trade off I’m willing to make.

productivity mac