|interest level vs. link popularity - 01.01||01.02||one user's interest level vs. link popularity - 01.03|
As the amount of information available to everyone increases, it similarly becomes increasingly necessary to seek out
"nodes" that gather and filter a subset of all of the available information.
We see these nodes take the form of Google News, RSS/feed aggregators, "social bookmarking" sites, and blogs that gather/collate in specialized niches.
For example, I used to keep a running collection of sites that featured visualizations of complex networks and interesting interfaces for browsing large sets of data, until I discovered visualcomplexity.com, a labor of love by Manuel Lima that "intends to be a unified resource space for anyone interested in the visualization of complex networks." Because of the narrow focus of the site, for me it has an extremely high signal-to-noise ratio, and I can read this "aggregate blog" through my RSS news-reader alongside other blogs, Basecamp project updates, del.icio.us inbox subscriptions, etc. and know that I have a wide coverage of visualization systems.
It is useful now to consider the empty chart in Figure 01.01 which plots personal interest in a link against its general popularity (on del.icio.us). Consider how many
of the links that you receive/view in the day are popular/random, and/or useful/forgettable.
And when considering how we browse the web, there should also be a distinction drawn between various modalities. There are many, but three modes in particular are relevant here: "information retrieval" where one has a query or roughly knows what they are looking for, "show me whats new" where the aim is to get a feel for the 'news' (or a pulse of the current world vibe), and the "show me what I need to see right now" mode where one has a fixed amount of time allotted to bit-viewing yet is flexible as to the content. (Note of course that any one of these modes can cause a user to branch or split off into submodes and sub-queries, a complex, intertwined dance of information gathering.)
One drawback with most aggregation nodes is that they mostly aggregate on popularity or relevance to the general population, acting
as band-pass filters attuned to a specific nodal resonance. The tradeoffs are between precision in delivery, signal-to-noise ratios,
and volume of information presented. As we wish to get more a wider spectrum of information, these parameters get jostled.
I started adding my links and tags to del.icio.us when I realized what an immense predictive tool it was,
and that by adding my data to the collective whole I would then be able to use the system as a feedback device for my own exploration.
Generally speaking, I think the key shift is to begin to see, understand and utilize each other as these nodes/sources of aggregation, and continue to extend our tools and interfaces to facilitate our group-mind interaction.
In the interest/popularity graphs, perhaps tautologically, I am assuming that the probability that a user will see a given link is proportional to its popularity. This means that there
are links (that fall into the areas above marked "SWEET SPOT" in Figure 01.03) that are currently of medium- to low-popularity and yet are of high interest. These are
the areas that I feel are worth exploring and facilitating access to, and that nodes such as del.icio.us can be utilized in this manner
in a much greater degree than at present.
del.icio.us allows a user to subscribe to another user's book-mark stream. This is a great feature, but out of the thousands of users, how do we pick out the users with the highest (for us) signal-to-noise ratio? And how do we best utilize the disparity in link popularities, to find hidden nuggets of personally-relevant information that might otherwise be missed?
Continue on to part 02 | Graphs