This discussion on the usefulness of date organised folders prompted me to post a bit more on my filing system. I already discussed how I embed meaning into the file itself, but now how I sort the files.
I came across an interesting index card filing system designed by a neo-luddite from Japan who wanted a better PDA. He carries around a stack of index cards to keep all of his thoughts on, and at the end of the day, moves those cards to a filing system. For filing, he uses Noguchi’s strict chronological system with one exception: Accessed cards are not moved to the front, but marked on the top with a dark pen. This way, when looking at a large stack of cards, you can quickly see which ones have been accessed before. Frequently accessed cards will have a number of marks (he stops at four). So cards are stored strictly chronologically, and highlighted by re-use. Next, when creating the card he makes a mark along the top left in one of four places. The position of this mark indicates what kind of card it is. Is it information researched for some project? An original thought or idea? Something that needs to be done? Et cetera. That is along the top of the card so it can be picked out from a stack. On the front of the card he puts a small icon to denote the type, a title for the card, and the date stamp. The date stamp is also used as a unique identifier for cross-referencing.
And that’s it! But it is surprisingly effective. Knowing roughly when something was written, and what category it is can quickly narrow down a large stack of cards to a handful, and the title at the top of the card makes it easy to relocate the precise card you were looking for. Putting the date in a prominent position makes it easy to re-sort if you accidentally drop a handful of cards.
I tried out this technique for a while, and then decided that while I quite liked capturing my thoughts onto cards during the day, I really wanted things stored digitally in the end. So what I do now is simply type in these cards using MMD, and file them into my archival software. I use the same principles of accessibility. Everything is stored strictly by chronology, and the only visible data is the tag, title, and date. I don’t mark for access because that is only something that is useful in a physical stack of cards where the title cannot be available looking at 1000 of them in a filing drawer. Note, I periodically use card instead of file in the rest of this discussion because I often think of individual files as being cards; sorry for the confusion. Assume I mean files when I say cards, unless I’m specifically talking about physical index cards.
What I have found liberating is the concept of only have four tags. I put things into a year/90-day cycle structure, give it a rough category, title and date, and it is done. No more filing. Filing is drop dead simple. With cards you just put the day’s deck on the top of the stack. Done. With a computer, you just save them into the current date folder. Done.
The thing I discovered is that I simply never used extensive tags. I spent hours and hours entering in lists of keywords, and never once used them in an actual recall situation. With this current system, I have never lost a thought. Using date+type+title is exceptionally comprehensive in a single-person situation.
Now, I have gone and sub-typed a bit. For example, my second tag type is @Creative. Since I am very often looking for thoughts on a specific project within the @Creative tag, I’ll jot down the name of the project after that. So the tag becomes @Creative-book-X or whatever. But I am very careful to only make these sub-types an exception.
Another thing I like about this system is that it is well suited for the computer’s file system. The three main filing axis can be embedded right into the file system. By example:
07193687-R-Digital Photographer as Chemist, Photographer, and Darkroom Tech.md
This file name has the date at the front so it will always be hard sorted by date. The ‘R’ means ‘Record’ which means ideas or thoughts I’ve had, the title, and then the file extension (MultiMarkdown). It might look a little cluttered by itself, but in a list of several hundred, those dashes and numbers at the front form visual columns that the eye can quickly pick out data from. What to use for the tag marker is something I have not completely settled on. I’d love to use Unicode characters, but that is still not completely cross-platform and stable. Perhaps in the future that will be a valid technique. I experimented with numbers, but found letters are easier to pick out in a crowd if the letters are well chosen. C and R look quite a bit different, where 2 and 3 can be missed when rapidly viewing a large list. I use R C M and I. The nice thing is that if you do change your mind, it is very easy to change them all at once using a bulk file renaming tool. So if/when Unicode becomes a valid technique. I can search for -R- and replace it with some icon surrounded by hyphens. For those rare cases where I sub-type:
07193687-R-Dreams-Analysis of highschool reunion theme.md
That way, it doesn’t get in the way of the primary filing columns, but is still in a visible slot. Using Smart Folders, I can search for -Dreams- and quickly isolate all files by sub-type. Or agents in Tinderbox, or whatever.
This is the type of extremely durable meta-data that isn’t going anywhere. With MMD on the inside, and most of the filing data on the “cover,” a directory full of this stuff will last for as long as we have “files” and “folders” on our systems, and conceivably even beyond that. This filename will work on any modern operating system, but to make it even more durable you could replace spaces with underscores and remove grammatical punctuation. This would increase its scope to the web. To reference this card from another card, I would simply type in [ic07193687]. This is its unique number derived from the date (ic- for Index Card). If I combine all of these cards into a single file, that will create a hyperlink between the linked text and this card. It’s also a human usable link. I can double-click, press Cmd-E to move the selection to the search buffer, and press Cmd-G to find the next instance of it. Or I can visually scan the directory for that number and find the card/file manually. Another efficiency benefit of using an archival software (be it integrated with the Finder, or something entirely separate that can recreate the finder structure later), is that you can now search the contents of files as well as the titles. This should not be underestimated, and it is a big reason why I ultimately wanted to stick with digital instead of going all paper.
Exporting my knowledge archive to another computer is a simple matter of zipping up the entire top level directory and moving it to another computer. Any further filing on top of this system is gravy, and not unlike the principle I described in an earlier post where modern technology is used to add efficiency to a system that has a low-technology back up. If my archival software fails, I simply lose a little efficiency, but I do not lose any data, and the primary retrieval system still exists.
There are several important concepts for making this system work:
The first carries over from the index card paradigm and that is to keep data parcels small. These files are not big. Rarely are they more than a few paragraphs long. In this sense, they are much like a library index card method. This is the same technique that Tinderbox, Scrivener, and other programs use to keep search results relevant. Since there are very few systems that can actually link to points within a file, you must assume that you can only link to the file itself from other files. If each file is many pages long, your cross-references reduces its usefulness. But full files can be placed into the index directory without any problems. I just reference them differently internally so that I know what I am getting in to from the cross-reference link. That little ic- in front is part of that. I will rarely link to files on the drive from a card file, because this is prone to link breakage in the long term, but occasionally I will do that. More often, I’ll just copy the source file into the filing directory and give it a new name. Then I know it is going nowhere, and linking to it is not nearly as risky. The main reason I’ll link to an “external” file is if the placement of that file is important. For example, a Scrivener template needs to be in a certain location to work. That it must be there is also a safety mechanism.
Second: Picking out threads can be done using the system itself, not embedding further complexity into the system. To explain what I mean by that, consider the most common method for threading. Say for a while you have a series of ideas revolving around one project. You want to somehow mark that these files/cards are relating to that project. The typical response would be to add a keyword. There are no keywords in this system though, so you would have to increase the complexity somewhere by adding another axis. Now, excepting the rare cases where I create a sub-type, the way I approach this is by creating an index of these cards and storing it as a new file. This might look something like:
07193724-R-INDEX for modern photography techniques.md
The capitalised “INDEX” makes it very easy to spot, and since it is filed using the same techniques as everything else, it is just as easy to isolate an index as it is any other file. This file will simply contain a list of all the files/cards using cross-referencing. Subsequent index files will refer to the prior index file at the top of the list, OR, they can simply add up all of the other lists and then continue the list. The key is that the old ones are not replaced or modified.
This leads right into the third principle, and that is: Once you file something, you never ever touch it again, except to look at it! That is perhaps the most bizarre in this modern computing age. If you want to add ideas to an older file, you make a new one and cross-reference to the older one. I’ll admit, this one is probably more up to taste than being an absolute concept, but here is the philosophy behind it. The digital age has made modification of data a largely corrective instead of additive task. The original, once modified, is lost. Now in a perfect world that is fine. 99 times out of 100, you changed it for a reason and the old data is no longer necessary. But there is that 1 time where you have an accident and lose data that is now gone forever (unless you are extremely diligent in backing things up), or realise a year later that you really wish you had version one instead of version two, and so on. If you leave things strictly chronologically, where old data is never modified only referenced, then you will always have a full record of any idea’s evolution. In this day and age, when drives are large enough to store most of the textual data in the entire world, it really makes no practical sense to delete or overwrite your own thoughts with new ones. The one weak point in this, is that if you take this seriously and never touch old files, you can create “future scanning” blind spots. While back-referencing from newer files is possible, your brain might only remember the original idea formulation’s date, and not the revision date. But the original does not reference the revision, only the revision references the original. For this reason, I will sometimes, very carefully, add future-references to old data cards, if I feel it is important to do so. But I always add these as an addenda, so it is clear that anything below that “line” was written after the original drafting of the file. Another point of contention for some is whether or not anything should ever be culled from the system. I do a natural sort of culling when I transfer paper cards to digital. During the day I’ll jot down things I need to do; grocery lists and such, and these have no use in a knowledge archive so I do not store them. The individual that developed this system, on the other hand, does store them. He keeps everything, and consequently has thousands upon thousands of index cards in his archives. I believe in the retention of data that seems irrelevant because you never know, but I do draw the line with some things. What I will sometimes do is fold these into my daily diary entry. Sometimes it is fun to go back and see what you picked up at the grocery store ten years ago. You can scoff at your old eating habits or whatever. But to store it in its own file, I personally think that is overkill. I never delete anything I am unsure of though. If there is any hesitation, I save it. This philosophy has saved me many times in the past. We usually hesitate for a reason, and I’ve learned to trust that. I’m sure there is a lot of junk I’ll never need, but I’d rather have a lot of junk than a missing idea. In theory, the filing system should be able to accommodate the junk. If you can still find the important stuff amongst the junk, then it is okay to leave the junk. That is my philosophy on the matter.
Another principle is that this is a personal filing system. When we communicate with ourselves, we can take shortcuts that we could not otherwise take with other people. I think this is part of why the 4 tag category system works as well as it does. The whole concept of using large keyword lists evolved from a need to make data portable between large groups of humans. If you are indexing a catalogue of images, you want to place as many searchable terms as possible, because it is not practical to predict what everyone will think when they search for something. We have one very critical bit of information that another person will never have a complete knowledge of, when we first thought of something. The chronological aspect is useless to pretty much anyone but ourselves and maybe a very close friend or two. But to ourselves, it is an incredibly useful map. Once you get the list down to a few dozen entries, finding a title is an exceptionally rapid process, even if we do not remember the title. As for what to use for the four tags, I believe that is a personal choice. I developed my own four-tag system instead of using the original, because one of his tags was GTD. While I do use GTD, I don’t want to store GTD tasks in my knowledge archive. He actually uses the card system as his GTD system, while I use a dedicated program for that. I replaced that tag with “Communications,” which by the way, this entry will be filed under in my system. It could probably be filed as Record, but I’ll remember it in the future as “that forum post I made in the summer of 2007.” Somebody else might just put it in Record if they do not have a communication tag.
It was a little scary adopting this system at first. I discovered that my extensive filing techniques had become a bit of a “safety blanket,” that in no way reflected upon my actual needs. I continue to evolve the system here and there, but am very careful to not tread on the core concepts set forth by the Noguchi method. I’ve been using it for a year now, and there are thousands of files set up this way. I have yet to spend more than a few minutes locating anything in that archive. Usually it is more in the order of seconds.
That, for me, is all I need to continue in its usage.