• by Aaron Goldman , Op-Ed Contributor, September 5, 2007

In my last column I covered the ground-breaking MyLifeBits project currently underway at Microsoft. Led by Gordon Bell, a team of researchers is attempting to build a memex (memory extender) by digitizing and archiving everything from personal emails and documents to phone calls and TV shows.

The goal is to create a platform by which a person's entire life can be indexed with the information available for on-demand retrieval. Then, taking it one step further, Bell and team hope to layer in artificial intelligence to use the data to complete tasks in much the same way a personal assistant might   today  -- with the intimate knowledge of one's preferences, availability, etc.

Last time I focused on the MyLifeBits approach to constructing a memex as well as the trends working in its favor. I also discussed a few of the ways a successful memex could impact society at large.

Today I'll spend some time on the challenges Bell and team face in creating a memex and the issues we'll encounter if/when such a tool is ready to roll out. Then, in my next column, I'll explore the implications of MyLifeBits for those of us in marketing and search should the project succeed.

The Hurdles To Getting There

 One of the challenges with mass-memexing is the barrier to entry. To build his personal archive, Bell spent over three years scanning everything he owned  --including his health records, and even labels from bottles of wine he liked.   

Clearly, the value in a memex is tied directly to its comprehensiveness. It is (relatively) easy to commit to digitizing all one's belongings on a go-forward basis. As Bell points out in this Sciam submission, once you create a framework for digitizing and annotating, it becomes (again, relatively) easy to work it into your everyday routine, not to mention that most "documents, images, and videos are now created in digital formats so capture is automatic."

To help with the indexing process, Bell now wears a "SenseCam" around his neck that snaps pictures when it detects a warm body nearby, or a change in light suggesting new scenery.

Going back and scanning and tagging everything from your past, however, is quite a daunting task. And we don't all have the bankroll of Microsoft to hire assistants to do it for us.

Other hurdles to adoption that Bell has identified include:

1. Poor speech-to-text conversion tools - allowing users to "search for words or phrases in phone conversations or other recordings" would provide great utility.
2. Lack of automatic face recognition - this would "solve the pesky problem of photograph labeling."
3. Inability of systems to recognize formats - if the system "automatically identified the nature of each of the several hundred document types [it could] analyze their form and content [and aid in the information retrieval process.]"
4. Restrictions against recording and photographing people - some countries and U.S. states flat-out don't allow it.

The Issues Once We Get There

 For now, though, let's suppose that we can overcome these challenges and successfully build and maintain a memex through MyLifeBits or some other application. Now we encounter an entirely different, but no less overwhelming, predicament.

If you're scared by the notion of a search engine saving your prior search history, I can't imagine how you'd feel about a "hypothetical company called LifeBits, Inc." managing the hosting and storage of your digital memories.

It's pretty ironic that the same company that is earning praise for allowing users to opt out of cookie-based targeting and making search data anonymous after 18 months is behind an initiative to create the largest personally identifiable index in history.

The telling insight here, though, may lie in the statement made by Brendon Lynch, director of privacy strategy for Microsoft: "Something we recognize is that the customers [need to be] in control of their personal information."

Control is clearly the crux of the issue. If people are empowered to make decisions about how their data is collected, stored, and used -- certainly, in Bell's case, active participation is required to generate the memex -- they will be more comfortable with the process.

Another key consideration is the safety of the data. Bell acknowledges the validity of such concerns, noting that "digital memories, unlike those in our brains, would be fair game in a legal proceeding."

Bell's proposed solution? New technologies that "obscure images or speech to avoid illegal recording." Great, so now my memories can look like they're straight out of an episode of  "COPS."

Another alternative Bell suggests is that "particularly sensitive information that might put someone in legal jeopardy can be kept in an offshore data storage account -- a 'Swiss data bank,' if you will -- to place it beyond the reach of U.S .courts." Sounds pretty out there, right? We may as well stash it at Gringotts.

What's Your Memex Got To Do With Me?

Between my exploration of Ambient Findability and MyLifeBits, I've now devoted my past five columns to some pretty heady concepts which are not immediately actionable to search marketers. Hopefully you agree that it's important to be thinking about the future of finding, and have enjoyed the diversion from day-to-day search tactics.

In my next column I'll cover some specific ways those of us in marketing and search could potentially capitalize on widespread adoption and scale of MyLifeBits. After all, it's not inconceivable that Microsoft would offer a free ad-supported version of this product. It will just have to make sure the value exchange nets out in favor of consumers -- or else they'll be saying MyLifeBits Bites.