Beyond C.S.I.: The Rise of Computational Forensics

This is an interesting article.

"The fact is that many forensic tests…have never been exposed to stringent scientific scrutiny," a committee convened by the U.S. National Academy of Sciences concluded last year.

It is good to see that there's an interest in improving on forensic practices!

The concept of presenting other types of evidence with a "percent certainty" rating, as is done with DNA, has been discussed in forensic circles for some time. But as this article points out, the data has to be crunched first, to make a mathematical rating calculable, for a specific type of evidence.

There is something seriously wrong, when the types of 'evidence' or data which could reasonably be subject to a mathematical evaluation, are presented without the proper assessment of uncertainty values. Instead are accepted without question as 100% certain truths, or else fall back on the subjective certainty of an "expert" which may or may not be a good match for the actual but uncomputed mathematical values.

The flip side, of course, is that experts who give subjective testimony are sometimes encouraged to "mathematize" the presentation of their opinions, even if there is no mathematical basis whatsoever for doing so.

For a (rather amusing) rationalization of this practice in the context of criminal threat assessments, see page 17 of CISC's SEWS protocol. Highly likely, better than even, very good chance of a chuckle.... (oh what the hell, lets say, 72% chance of a hee haw, not to give the false impression that this number is the product of, or related to, statistical practice. It is simply a means to give the false impression that this number is the product of, or related to, statistical practice communicate my subjective judgement in a way that is more objectively meaningful.. ?? )