by Jared Goralnick
Why does it take less time to find something on the web than on your hard drive?
It’s because the internet has no order, but we’d like to think we do. Guess again: using an orderly approach to storing and retrieving is similar to paying full price for airline tickets: it made sense twenty year ago but is a costly decision today.
Here’s how to file and find things in the next decade.
Bottom Up vs Top Down
Storage and retrieval ultimately fit into two methods: top down and bottom up. A top down approach is to come up with categories, and perhaps sub-categories, and sub-sub-categories…and then to place each item in exactly one place. For instance, my residency might be classified as United States:Maryland:Prince George’s County:College Park. From a legal perspective that’s the only place you’d find me today. This is how file cabinets and hard drives have been setup forever.
However, in a bottom up system, one places all items into just one bucket—that means all residents aren’t classified into countries, they just get adjectives (or tags). They have their name and their adjectives, some of which might conflict. For instance, I could be a resident of both DC and Maryland in a bottom up system, of both the United States and Spain.
In the physical world it would be impossible to find anything if it were all lined up next to one another, but with fast servers and intelligent algorithms, you can search for “Jared Goralnick” without having to narrow your search to any geographic criteria. This is helpful, because I identify first with DC, second with Baltimore, and not really at all with College Park. So if you met me you might have trouble looking me up in and old-fashioned (top-down) phonebook—you might never look in the College Park edition.
There are numerous reasons why bottom up storage and retrieval are better than top down approaches when it comes to digital information:
- When filing something, there’s only one place to put it. Tags are optional. No choice means a heck of a lot less time filing
- When retrieving something, it’s better to use a (very fast) search tool…rather than guessing the right folder to look in first
- If an item relates to two projects, or is from an old friend but relates to business, etc…there are times when it’s tough to know where to file…or to retrieve it. In a bottom up system, you just archive it in one place and search for it later. End of story
- With tagging, one can tie one item to many categories. So something can be tied to two projects without one having to copy it into both folders
Applying Bottom Up Approaches to Email
Not everything in the world of technology is ready for a flat, bottom-up architecture. But email is, at least if you use Gmail, Outlook 2007, or Postbox. I presume Mail is the same (feel free to verify in the comments). The fewer folders you have, the faster your filing and retrieval will be.
That is because these modern email clients/web email sites use search indexing, the same technology that allows sites like Google to search the web so quickly. With Outlook you can even search within the body of the files that are attached to your emails. And it’s all instant.
That being said, you may still want to have folders for broad categories, but if you do, I’d suggest that you ensure that you can see all the folders on the screen at the same time. That means that if you have more than a dozen you’d probably better rethink things. Try to have fewer than five.
You may be thinking that having fewer folders feels less organized. But in reality there’s less to deal with: fewer places to go, fewer places to look, and less time spent filing or retrieving.
Applying Bottom Up Approaches to the Rest of Your Technology
Data is becoming more bottom-up friendly every year. The search features in Windows have been pretty good since Vista, and get some helpful improvements in Windows 7. Same goes for Mac’s Finder—it gets faster in Snow Leopard.
For a long time you’ve probably paid attention to metadata (which is a fancy word for the tagging taking place inside a file) for your music collection—your artist, title, track, album, and other data have been associated with your music files. That trend is going to be carried through to everything over the next ten years.
So I’d suggest that you think seriously about it the next time you do some reorganization on your computer. We have more files than ever before (even if they’re online). So labeling and tagging will be more relevant. With photos, this is especially true. But all documents deserve better labels, not better folders.
As we’re surrounded by more and more data, bottom up filing and sophisticated searching will be the only approach for us. We need to start shifting our mindset and getting ready for it.
You can start now, with email.
You should really subscribe to Technotheory via 
email or rss.
Jared,
Thanks for the good article. I do think that there’s room for a hybrid system — a few folders, plus the “big pile.”
In my wife’s work, for example, she handles scheduling for co-workers, so she needs to know when they’ll be on vacation. But when they send email to her with that information, they don’t use consistent terminology: one person “will be out” from July 12-15, one person “will be on holiday,” one person “will be at a conference,” etc. So there’s no consistent tag to use for retrieval. In this case, a “Scheduling” folder is very useful.
Unless I’m missing something. Is there a better way?
Hi Dan,
I see nothing wrong with a few highly focused folders for specific purposes. The only problem comes when they write a message that contains multiple points that you might want to refer back to…but one of them relates to a scheduling issue.
When it comes to tagging, I don’t think Outlook does a good job with categories, whereas Gmail does a great job with labels. In either case, a tag could be used in place of putting those messages in a folder. Nonetheless if there aren’t many of these special cases, then there’s no harm done.
For instance, when I hire someone new I create a folder specifically for that job as I usually get a ton of responses in 2-3 days. Then afterwards I’ll dump all those replies into my generic correspondence folder, but just put a tag on them for job-applicant.
So it’s useful to recognize when folders can help to really bring out a particular category and when they can get in the way by adding one more decision to the filing process.
Cheers!
Jared -
Thanks for another very useful post!
You are so right about the importance of metadata. For anyone interested in metadata for images, there is a site that has aggregated all of the information – http://www.photometadata.org/. With a grant from the Library of Congress, they have been on a campaign to get anyone who uses images to start using metadata, and they put on a great seminar in D.C. last week.
That’s a great tip and a cool site, thanks Victoria!
Fantastic post. But just the tip of the iceberg, right?
I ask rather than declaim because I’m further behind you on the curve of this switch to the new, search-based way of doing things. My hard drive looks like a tag-and-folder yard sale, and my delicious, StumbleUpon, etc is worse–I’m terrified of doing it wrong and not being able to find stuff, so I overdo it.
I know this is wasted time. And I know I’m not the only one grappling with this. And while I also know that taxonomy ain’t everything, it would be helpful to have some reassurance in the form of a system–a way of looking at things, i.e. principles, or an actual system.
If you can come up with a way to do this, you will rule at least part of the new known world. I mean, I’d pay for that sh*t and I’m a cheap bastard!
Hi Colleen,
You’re taking the right steps. The best thing to do is in fact to overdo. But overdo with small detail, not with general repetition. In other words, don’t just use high level general categories when tagging…tags that you probably overuse or are just broadstrokes that may not help you find things. 500 articles tagged with one term are hard to sort through.
Instead, use specific tags and use lots of them even if that tag only ends up being used once or twice out of every 500 items you’re categorizing. It’s best to be specific so you can find things, while the general categorization is more useful for machines or for high level analysis (“I tagged 200 articles about productivity so clearly it’s on my mind” is useful for analysis but not for finding specific articles).
I’m using a file naming system that allows me to file everything in one folder. I start with the date, type of file (Excel, email,etc), who created it, and finally a description of the file. Most people just use the description as the file name, and go through the agony of trying to find the file after several months. My system might take a little longer to set up, but using a desktop search makes it easy to find the file that I want.