The Digital Archive of Everything

Posted on September 20, 2007 by psu

I was walking around in my local Borders a couple of months ago just after they rearranged the Music and DVD section. My eyes scanned through the shelves of DVDs, and I realized something that was at once horrible and amazing. There, on the shelves, was most of the history of broadcast television archived in little boxes of silver disks. You could get anything, almost no show was too obscure or too low quality to miss the cut. All of this is made possible by the digital encoding of the content. With enough storage, you can capture anything in bits and then instead of being available only at the whim of the content producer, you can watch it anywhere.

Of course, storage is a tricky problem. For most of my lifetime, the cost of storage made it impractical to manipulate large amounts of video in an affordable computer. But things change. 25 years ago 1GB of storage took up a 8 by 20 foot collection of rack mount disk drives in a machine room somewhere at CMU. Today, I can easily lose a chip holding 1GB of storage in a pile of lint in my pockets. When DVDs were introduced, the amount of storage (4GB) that each disk represented was much larger than what you carried with you in a typical computer. These days, a single DVD is a tiny fraction of what you find walking around in a cheap laptop.

Given this, it’s not hard to imagine a future where a single computer, or at least a single external disk drive will be able to hold all the music and video that you might ever listen to and watch. In fact, you could probably store all the music and video ever produced without that much trouble.

Essentially infinite storage also leads to other obvious ideas. For example, many of us like to keep journals and diaries. Others record their experiences with still or video cameras. Since disk isn’t really infinite, these records are usually fairly selective. But, you might ask, why be selective? The New Yorker recently published a profile of one Gordon Bell, who has been thinking about this very question. The article is an interesting look at an interesting person, but I found the conclusions to be disturbing.

The article correctly points out that a complete archive of everything brings up tricky questions of privacy and civil liberties. They also correctly realize that a raw archive of everything we see will never be that interesting. Therefore, they avoid the naive notion that if you strap a microphone and video camera to your person and record everything you see, the result will be like your own personal episode of Seinfeld. You might hope that your life archive would be an entertaining series of little stories full of humor, or drama or deep insights into the human condition. This, of course, ignores the essential fact that almost all of our lives are a banal exercise in tedium. The reason television stories are interesting is that all the boring stuff has been edited out by the filmmaker. The full archive of everything lacks this editor.

Of course, maybe you could build a computer system to find the interesting stuff for you. This is where I think the article goes wrong. There is this quote:

Similarly, Bell and Gemmell would like software that organized the contents of the archive into movies–something, at least, to compress and shape it, to summarize its parts. “Auto-storytelling,” Gemmell calls it. “My dream is I go on vacation and take my pictures and come home and tell the computer, ‘Go blog it,’ so that my mother can see it. I don’t have to do anything; the story is there in the pattern of the images.”

Personally, I find this vision horrifying. It’s good that the article picked photography as an example, because it’s easy to use photography to explain why I hate this idea so much.

Photography, more than any other artisitic endeavor, is at its core an exercise in selectivity. You look at a subject, you choose how to frame the subject in the picture to capture what you find interesting about it. You wait until the context is just right. You hit the button. Later, you look through all the pictures and filter them again. This time, you are picking out the ones that “worked”, meaning that the picture actually captured what you originally visualized when you were shooting. The resulting set of “select” images (even the jargon of the field emphasizes selectivity) are the ones that you personally feel are the strongest pictures.

All of this selection is important. Not only does it serve to ensure that you don’t show your weak pictures to people, it is also at the core of photographic style. Style can best be defined as systematically capturing pictures that you like. Part of cultivating your own personality in your pictures is to be able to pick the best ones out of the pile and make people pay more attention to them. In other words, the selectivity is at the heart of photographic style.

Now, imagine that the machine is making these choices for you. Why would you want that? I think that the key thing that makes important memories important is that I picked them to be important. I want to sift through my life on my own. I want to look at it all and pick out what I think is interesting by myself and then synthesize it into some story that I can go share with the world. Outsourcing this to a machine is tantamount to calling down the robot holocaust and replacing all the people on earth with little a collection of machine learning robots. No thanks.