Saturday, July 28, 2007

The Wiki as a Meta-cortex

The human mind is a wonderful thing. It performs complex visual analysis, can restructure itself on the fly and and determine solutions to extraordinarily complex problems. It does however have problems with retaining discrete packets of information. There are basic ways to increase ones capacity for remembering arbitrary stings of information, such as mnemonics, but often these require invoking longer strings with stronger associations before being able to recall the specific data needed. (This can be reversed, especially if the information in question is an abstract, as the associated mnemonic can be an abbreviation of the key elements in the abstract.) However, these types of associations work best when they are accessed frequently. For those that are used infrequently it can be much harder to retain a useful memory. Personally, I feel that this is acceptable, at least in that it does not really hinder any day to day interactions. The area where infrequent recall of arbitrary strings is most vital is in the use of computers. Passwords and user names are not necessarily used on a daily basis, especially for things such as online banking. However, it is important to be able to recall these rapidly and accurately, as they control the flow (or lack thereof) of your money, and can impact other portions of our lives as we function more in the digital realm.

Mitigating the impact of the shortcomings of short term memory can likely be implemented in a software setting. Using a user populated personal wiki to manage data outside of one's head is a viable option. The customizable file structure, the ability to embed content and then search it using tags or keywords makes wikis ideally suited to the task of personal data storage. Coupling a reasonably secure wiki with a sandboxed hardware device would allow the user to carry a significant amount of data with them at any given time. In and of itself having readily search able data is useful, if it could be tied to location or visual cues it would also be a prompting system, providing pertinent information without user input.

The main limitation thus far is twofold: commonly available webcams do not have a high enough resolution to accurately identify visual cues and its hard to put enough processing power in a portable to analyze video in real time, especially at high res.

None of this is new. And its not necessarily hard either. It just simply hasn't been done widely yet. I think that it should be.

No comments: