File Hierarchies: Images & Email

Every now and again, I come across something that makes me realize there is something about me that is different than most people. It happened this week while reading A World Ordered Only By Search by L.M. Sacasas, which is, in part, a commentary on an article Monica Chin’s File Not Found in The Verge. Here’s the relevant bit:

There comes a point when our capacity to store information outpaces our ability to actively organize it, no matter how prodigious our effort to do so. Consider our collections of digital images. No one pretends to order their collections. I’m not sure what the number might be, maybe 10,000, at which our efforts to organize images falters. Of course, our apps do this for us. They can self-sort by a number of parameters: date, file size, faces, etc. And Apple or Google Photos offer a host of other algorithmically curated collections to make our image databases meaningful. We outsource not only remembering but also the ordering function.

I order my images. I find it is pretty easy actually. There’s the main principle, which is by year. Then, you group them into relevant categories. Smugmug essentially uses symbolic links to files that allows for easy hierarchical ordering. So, I can keep files in a specific year. Then, another folder that points to a sequence within that year that corresponds to a vacation or event. You can also create folders for particular people, such as scans of older photos of relatives.

This is really only a problem if you only have one method, such as year, as an organizing principle, which forces you into a search paradigm. You need to be able to organize in multiple places within a hierarchy, and you need to choose software that enables you to do that. However, if you think in terms of search, then you are less likely to think in terms of symbolic links that allow for multiple points of categorization and points of entry to the same file or set of files.

It may be more “natural” to me for a number of reasons. One, *nix systems incorporate symbolic links into their file structure exactly for this reason. So, anyone that primarily uses a *nix system is going to think of hierarchies and using symbolic links to create different points of access to the same files. Two, I used to be a librarian. So, I might claim that I have a greater familiarity with classification systems, indexing, and other ways of organizing information than most people. This blog is probably a good illustration of this point. Another is email. I use a fairly simple structure in email that looks like this:

  • Archives
    • Blog
    • Correspondence
    • Goods
    • Information
    • Jobs
    • Statements
    • Subscriptions
    • Travel
  • Family
    • Name 1
    • Name 2
  • Friends
    • Name 1
    • Name 2

Blog is anything related to this blog. Correspondence is something official, like a politician. Subscriptions tend to be when I sign up for something online. If it is a friend or family member that doesn’t have their own folder, they get dropped into the main folder. If there’s enough email exchanged, then I create a new folder and add all their emails to it later. The rest is self-explanatory.

It’s easy to keep this file system in your head and to quickly categorize a piece of email. The aesthetic for edge cases is to try to use an existing category to cover it, if it is unclear. Information and Goods covers a lot of territory.

Of course, some people may have large families with different branches or different friend groups. It is easy enough to categorize these under sub-headings. But, I think trying to over-categorize is probably why a lot of people end up giving up on file hierarchies. They are trying to impose too much order on a system that should be fuzzy, allowing for rough categorization without too much detail.

The same principles apply to this blog, personal files, and elsewhere. For cafebedouin.org, for instance, there are the categories that act as a rough heirarchy and tags are symbolic links that give other ways to access a particular post. When you start thinking of files as having a main organizing principle and then other avenues for access, hierarchies become a useful supplemental (or starting) strategy that complements search.

L.M. Sacasas made me realize that this kind of mental model for organizing information is unusual. I thought discussing it might spark ideas or be useful for someone else.

2 thoughts on “File Hierarchies: Images & Email

  1. Interesting. What occurs to me that that as we continue to (try to) model computer systems and AI after human cognition, we’re on a parallel mission to reorganize human cognition after machine systems. Neither is a particularly good fit, and it’s arguable that human cognition suffers more as it outsources quite a lot of memory and processing to machines, though with some obvious gains. Basic modes of operating in the world are now subordinated to electronic devices and their dynamics, which is a far greater transformation of human activity than simply developing and using tools.

    Side note: I also used to be a librarian and so think of information categories differently, though I’m certain I’m far less technical than are you.

    1. The A.I. getting out of the box post from today is hitting on this point too. There’s something obviously good about setting A.I. on a problem space and letting it find dimensions to it on volume alone. If A.I. can play a billion Go games against itself, it’s going to learn things no human is capable of.

      One problem, for people, is adaptation-response. If you outsource some core decision-making that you need to function in the world to a machine, are you seeding the path to some H.G. Wells to some variation to Eloi and Morlock dichotomy? Such concerns are probably overwrought at this point, but will they always be?

      On your side note: Curious. Did you go to what was formerly called GSLIS? Drop me an email if you’d rather not have talk about that in a comment section on WordPress.

Comments are closed.