In a recent
diary, Washington Post staff writer Dana Milbank posed the following question to the dKos community:
I'd like to know how you, and your readers, think this could best be done. Some of us have thought about the idea of doing a daily blog report, summarizing what the top blogs are saying and assessing the accuracy/significance. but that's just a small item.
I think we can go WAY further than this if we put our minds to formalizing the process that has worked so effectively in researching the PropaGannon affair.
I'd like to see the emergence of online "distributed research tanks" - groups of people (like those nominally led by SusanG in the PropaGannon matter) who take on vast research tasks and accomplish them in a large-scale, divide-and-conquer manner.
There could be perhaps no more efffective way to mine public resources for specific information than having large numbers of individuals contributing small amounts of their own time and their individual skills, as they see fit.
For instance, in the PropaGannon case, it took just hours for the distributed researchers to accomplish what might have taken an individual, resource-limited reporter weeks or months to do. Need to know if a J.D. Guckert went to some particular college? Well, it's easy if one of your researchers happens to be an alumnus; s/he just types Guckert into the alumni search page, and voila, you have your answer.
The trick is to match tasks up with the individuals for whom those tasks are easiest to accomplish.
Clearly we're a significant way from accomplishing this in a full-fledged manner today. Most of the problems relate to scalability: limitations in how we can prioritize and organize the research efforts, and in how we can effectively collate, organize, and double-check the information gathered. Still, I think much of this could be overcome with some serious thinking and perhaps some modifications to the underlying software.
For instance, web-driven blogs are apparently good for organizing this type of project, but they might be better if augmented with some really task-specific features like databases. Imagine if dKos could semi-automatically construct a graphical mafia-style "Family Tree" as we went along. Or, integrate the information collected into something like this 3-dimensional idea-cloud-thing. Talk about fun!
I can envision something like a blog crossed with a wiki crossed with bugzilla (a system for organizing group software development efforts). All of this could be built around a model resembling a distributed computing project, which leverages the efforts of volunteers, and which automatically protects itself against people (intentionally or unintentionally) corrupting the system with bad information.
I'd love to hear others' thoughts or suggestions! I'm definitely going to keep thinking about this.
Update [2005-2-17 15:15:23 by crushinator]:
Taken from a comment I made below. I decided an example might be informative:
Imagine you want to know something, like, say, "When was the first time Jeff Gannon spoke in the WH briefing room?"
You enter this question as a new research topic. You also enter "research enablers", things that would make this easy to answer. For example "Have access to Lexis-Nexis".
Then, later, someone who wants to help logs in. They click checkboxes on "enablers" that they think apply to them. (i.e. "Hey, I do have access to Lexis-Nexis"). The system then automatically suggests to them the set of highest-priority questions that need to be answered by someone with that access. They look them over, decide the Gannon-presence-timeline question is interesting, do the search, post their results, and are done!
The system could even force a question to be re-answered under some circumstances if the researchers want to be absolutely sure nobody's inputting bogus data.
