[Infowarrior] - EFF digs deep into the FBI's "everything bucket"

[Richard Forno]

The EFF digs deep into the FBI's "everything bucket"

A new EFF report pulls together everything that's now known about the  
FBI's monster internal records system.
By Jon Stokes | Last updated April 30, 2009 10:01 PM CT

http://arstechnica.com/tech-policy/news/2009/04/the-eff-digs-deep-into-the-fbis-everything-bucket.ars

Earlier this week, the EFF published a new report detailing the FBI's  
Investigative Data Warehouse, which appears to be something like a  
combination of Google and a university's slightly out-of-date custom  
card catalog with a front-end written for Windows 2000 that uses  
cartoon icons that some work-study student made in Microsoft Paint. I  
guess I'm supposed to fear the IDW as an invasion of privacy, and  
indeed I do, but given the report's description of it and my  
experiences with the internal-facing software products of large,  
sprawling, unaccountable bureaucracies, I mostly just fear for our  
collective safety.

The idea behind the system, which the FBI has been working on since at  
least 2002, is that the Bureau can dump all of its information in  
there so that it can be easily searched and shared. IDW contains more  
documents than the library of congress—a stew of TIFFs with OCRed  
text, multiple Oracle databases, news streamed in from the Internet,  
reports and records in various in-house data formats, watch lists,  
telephone data, and an alphabet soup of smaller databases and records  
repositories—all accessible as one sprawling system that processes  
batch jobs, runs queries, and issues alerts. In short, the IDW is an  
"everything bucket" for the FBI.

Complicating the picture is the fact that some parts of the system are  
classified as "secret," while others aren't. I'm sure the entire thing  
is a joy to use.

The EFF's report is based on information obtained over the past three  
years through litigating a FOIA request; the organization didn't get  
everything it wanted from the FOIA, but it got quite a bit. Some of  
the e-mails obtained are bureaucratic classics, in which  
correspondents are fussing over phrasing to be used when testifying  
before Congress so as to give the proper impression (e.g., that they  
care about privacy) and generally stay under the radar.

Ultimately, though, the EFF still doesn't have a complete picture of  
all of the data sources that have been added to the IDW, but the group  
is pretty clear on the direction that the expanding database is  
headed: data mining for the purpose of catching bad guys before they  
commit crimes or acts of terror.

Last year I wrote a pretty detailed explanation of why these attempts  
to use data mining to catch bad guys before-the-fact are all doomed to  
fail, based on an National Research Council report that made the same  
point, so I won't recap that here. It suffices to say that the  
precrime stuff does not work, and will never work, and government  
should take the money they spend on these projects and hire linguists  
and other human agents instead.