[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Video & cryptography...




Remember what your target is here.  (Focus, Pinky!)

The ultimate goal is that you're trying to convict a suspect of some wrongdoing in a courtroom, and you would like to use as evidence a piece of video footage.  But now, in addition to the old standard of "chain of possession" whereby a police officer logs when s/he receives the evidence, who has accessed it in the lab, storage, etc., you may now be required to establish the "truth" of the video -- was it forged, staged, or altered?  Can the defense convince a jury that a video is not what it appears to be?

Look at the O.J. case where they convinced people that a piece of DNA might not have been his.  What a load of malarkey they heaped upon those people, and shame on the prosecution for not going back and matching up another 50 markers, dropping the chances of an error from 1 in 3*10^12 to 1 in 3*10^60.  (Ultimately, it wouldn't have mattered of course because they pointed out other "flaws" in the chain of possession, specifically the FBI crime lab.  Why they might have had a *different* case with O.J.'s blood is beyond me...  They could have retested the DNA from every cell in evidence on that trial, found only Nicole's, Goldman's and Simpson's DNA, and STILL they would have gotten it thrown out, because jurors are incapable of understanding math.)

What this means to us is that, in a courtroom, a signed piece of digital tape will probably be accepted the same as an unsigned piece of analog tape.  The courts have already established precedent that says that the scientific method has less value than an overpriced lawyer.

Think about it:  what we strive for as a cryptographic proof means less than the unsworn audio tape from a 21 year old intern.  If some lawyer wants to challenge a piece of video, and can throw enough money at it, s/he's going to win.  You're better off spending money training the persons operating the VCR, because that's the defense's next point of attack.

The idea behind the digitally signed video frames is that the video "consumer" (the persons who shot and recorded the video) probably didn't edit the video to falsify some evidence, and that the person who built the camera is willing to go to court to testify that their video is not tampered with.  As a video "consumer", you're willing to pay for a witness from the video equipment supplier to come testify that you didn't edit their video.  It theoretically (in a certain, very limited subset of theory) eliminates the "Rising Sun" plotline where someone edited the digital video.

Now, we all know that since you can falsify the video by all kinds of other means, the only thing it really accomplishes is it adds a float to the parade going past the jury.  And whoever has the prettiest parade wins "justice", right?

At 05:02 PM 1/26/98 -0500, Adam Shostack wrote:
>Jim Choate wrote:
>| > From: Adam Shostack <[email protected]>
>| > Subject: Re: Video & cryptography...
>| > Date: Mon, 26 Jan 1998 13:56:57 -0500 (EST)
>| 
>| > Schneier, Wagner and Kelsey have done some work on an authenticating
>| > camera.
>
>Dave points out that this was Schneier, Hall and Kelsey.
>
>| > One issue to be concerned with is that what the camera sees is not
>| > always the truth.  Putting a film set together to film bigfoot is
>| > easy.  The fact that the film is authenticated as having come from the
>| > camera doesn't mean a whole lot in some cases.
>| 
>| Doesn't this same sort of issue arise from any other digital signature
>| process then? There should be nothing fundamentaly different between the
>| characteristics of a video camera signing a frame than a person signing
>| email.
>
>	It arises in a different context; with a signature on paper,
>you're generally indicating that you've read and consented to whats on
>the paper, not that you created it.  The meaning of a camera signing a
>video still is not obvious to me.  Is it intended to be 'this is what
>we saw through the lens?' or 'this is what really happened?'
>
>Adam