The future of blogging
Dec. 19th, 2003 11:11 amSome time in 2000 I had a conversation with a group of researchers that were showing off their intelligent agent personalized news system. It was designed to search for news across multiple sources according to the preferences/criteria explicitly expressed by the user. Basically, it was a glorified search engine married to the agent technology.
I argued that the system had two fundamental flaws:
1. scalability - as the number of users and sources increases, the amount of computation/profiling increases exponentially.
2. (more serious) - very low discovery value, since the user's query/agent will tend to position the results toward the description of the known phenomena. There's no easy, intuitive way to tell a machine: please bring me something I don't know, without expressing at least a rough outline of one's knowledge.
The solution, as I saw it then, and as I still see it now, lies in automated collaborative filtering, where the most intelligent part of content filtering is done by people, but the process of aggregation is automated to avoid the pitfalls of the traditional browsing.
I believe syndicated blogging fits this criteria. For example, your LJ friends, real or virtual, work for you to throw away most of the environmental info-noise. The process of friend selection (I'm not talking about real-life friends here) boils down to choosing content creators and/or content gatherers capable of making the information worth the discovery effort.
Unfortunately, content presentation in the majority of the blogging services is still tied to the "browsing/web-surfing" paradigm. This should change. A content management and presentation packages should emerge enabling one to completely replace a variety of traditional media sources such as newspapers, magazines and etc. Friend grouping in LJ is a very crude, rudimentary tool for creating a truly personal "living paper". All of its segments are still transient, so you can't set them up to reflect different time scales for different subjects, e.g. for technology, art, literature, politics.
In the end I expect a flexible software for personal and group communications that provides structure for information management in space (e.g. layout, physical location, interests, and etc... ) and time (what we now call news reports, newspapers, magazines, books, and etc...)
--------
the current thinking is at least partially based on information found in journals![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif)
centralasian, urbansheep, ![[profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
behrk, and a short comment exchange with joshuagay. thanks.
I argued that the system had two fundamental flaws:
1. scalability - as the number of users and sources increases, the amount of computation/profiling increases exponentially.
2. (more serious) - very low discovery value, since the user's query/agent will tend to position the results toward the description of the known phenomena. There's no easy, intuitive way to tell a machine: please bring me something I don't know, without expressing at least a rough outline of one's knowledge.
The solution, as I saw it then, and as I still see it now, lies in automated collaborative filtering, where the most intelligent part of content filtering is done by people, but the process of aggregation is automated to avoid the pitfalls of the traditional browsing.
I believe syndicated blogging fits this criteria. For example, your LJ friends, real or virtual, work for you to throw away most of the environmental info-noise. The process of friend selection (I'm not talking about real-life friends here) boils down to choosing content creators and/or content gatherers capable of making the information worth the discovery effort.
Unfortunately, content presentation in the majority of the blogging services is still tied to the "browsing/web-surfing" paradigm. This should change. A content management and presentation packages should emerge enabling one to completely replace a variety of traditional media sources such as newspapers, magazines and etc. Friend grouping in LJ is a very crude, rudimentary tool for creating a truly personal "living paper". All of its segments are still transient, so you can't set them up to reflect different time scales for different subjects, e.g. for technology, art, literature, politics.
In the end I expect a flexible software for personal and group communications that provides structure for information management in space (e.g. layout, physical location, interests, and etc... ) and time (what we now call news reports, newspapers, magazines, books, and etc...)
--------
the current thinking is at least partially based on information found in journals
centralasian, urbansheep, behrk, and a short comment exchange with joshuagay. thanks.
