Re: CRC32 vd MD5

On Sun, Jan 05, 2003 at 08:47:42PM +0100, Ronald Prins wrote:
>
> In my experience a lot of "forensic investigators" are using hashing

Is this risk (that investigators may be deliberately fabricating or altering evidence) a particularly serious or important one?

It's my impression that existing adversarial/legal systems cope with this risk by allowing for impeachment of witnesses for bias, and by correlating (or failing to correlate) evidence across witnesses or forensic artifacts - e.g., it's one thing for one investigator to say he saw evidence in one logfile or one one hard disk, and another (less likely circumstance) for a team of investigators, perhaps from different agencies or departments, to agree to falsify testimony and evidence about coordinated evidence (e.g., perpetrator's record of mail sent matches the mail logs of two intervening ISP's which also matches mail received on victim's computer).

I raise the question because the risk of falsification or modification is present with all sorts of other scientific or technical evidence, even with techniques which have won widespread acceptance - like fingerprints, tool/weapon marks, DNA or blood type matches, etc - the evidence itself is ephemeral or difficult/impossible to comprehend by a lay factfinder, so they're forced to rely on second-hand reports about facts and conclusions from those facts .. but legal systems haven't evolved a complicated escrow system for storing other evidence types - sometimes material is available for a confirming second test (or for testing by opposing experts), sometimes not - but that doesn't result in the evidence being excluded. It allows the side opposing its introduction to offer arguments to the factfinder(s) about the possibility of technical error or mistake, and the possibility that the judgement or truthfulness of the witness/ technician have been impaired by some sort of bias.

I agree that a hypothetical perfect world where evidence was held in the custody of a neutral body and made available on an equal basis to prosecutors and defendants (and perhaps civil claimants, too) is an attractive, and interesting one - but we are operating within systems which fall far short of that idea today, and for the most part don't seem to have big problems with falsified evidence.

Is cryptographically strong hashing as important as its proponents would like to argue? I think it always makes sense to use the best methods available - and I'd certainly recommend the use of SHA-1 over MD5, and MD5 over a CRC, if someone asked me - but in the absence of something better, do we need to throw out digital or electronic evidence merely because it's not strongly protected against fabrication, where we don't apply a similarly stringent test to photographic or paper-based or forensic/technical evidence?

--
Greg Broiles, J.D.
gbroiles@netbox.com
gbroiles@parrhesia.com

-----------------------------------------------------------------
This list is provided by the SecurityFocus ARIS analyzer service.
For more information on this free incident handling, management 
and tracking system please see: 
http://aris.securityfocus.com