In this ruling in May 2002, Judge Whyte denied Elcomsoft's motion to dismiss the lawsuit against them.
One of Elcomsoft's arguments was that DMCA tried an impossible hedge: the DMCA says it's still legal to circumvent protection (e.g. for fair use), but it nevertheless forbids the trafficking of devices whose primary purpose is circumventing protection. If there's no way to do the copying (because there are no devices, because they're forbidden), the fact that the copying is still legal is just a cheesy dodge on the fact that it's effectively not legal.
Elcomsoft's product was a device for bypassing protection mechanisms on PDFs, which are essentially text documents, and the judge points out that "from a technological perspective, the fair user my [sic] find it more difficult to do so—quoting may have to occur the old fashioned way, by hand or by re-typing, rather than by "cutting and pasting" from existing digital media. Nevertheless, the fair use is still available."
Now, if trusted computing succeeds, it will just be flat out impossible to, say, copy a movie stream at any step between the DVD (or newer equivalent thereof) and its display. However, computer scientists are dismissive of the probable success of trusted computing to prevent piracy; there will always be internal sources creating illicit, unprotected copies; and they can't "close the analog hole": media works like this have to be transmitted to our eyes and ears, and we can always recapture them there (with video recorders and audio recorders and such).1
Now, Judge Whyte was only looking at text, but surely his judgements about devices circumventing restrictions must apply to all media. Nobody is going to retype by hand a movie that comes on a 4GB DVD--not merely because it would take too long, but there's simply no comprehensible way of doing it. No, they'll have to point a camera at the screen and go through the analog hole.
A related issue Elcomsoft raised is that works might be prevented from reaching the public domain. Judge Whyte seemingly willfully misinterprets this--obviously things already in the public domain remain in the public domain, you didn't have to point that out. (He literally spends 3 paragraphs addressing the release of already-PD'd works in restricted formats, and none on the other.) But it's clear from the fair use discussion that he has an answer to the PD problem as well: the analog hole. A work that lapses into the public domain can still be copied via the analog hole, even if there's no way to bypass its copyright restrictions.
But hold on a second here: a work whose original format is digital is a string of bits. For example, a movie is an MPEG-4 stream (or whatever). On DVD this is encrypted using DeCSS, producing a different string of bits. We can see those bits all we want; but those aren't the bits of the movie, they're the bits of the encrypted movie. As well, the analog hole does not give us access to the original string of bits (although it might let us generate a totally new one).
Now, any string of bits can be interpreted as a single very very large number, and I can ask questions about that number, like: "is it prime?" Any data I could collect through the analog hole won't help me decipher this.
The Elcomsoft ruling, as far as I can tell, says to us: "yes, it is perfectly acceptable that the DMCA results in a situation where, even after Terminator IV2 goes into the public domain, nobody can answer the question 'is it3 prime?'"
Is that an important thing to be able to ask? I have no idea. Maybe there's some legal standing where it's not the "principal expression" of the work or some legal jargon (the idea being principal expression being not the bits but the 'fully decoded' movie--a stream of changing light and sound). But it seems to me there's some potential significance to this limitation.
1 On the other hand, things like watermarks on currency that prevent xeroxing suggest that maybe it really IS possible to close the analog hole, too.
2 That is, some hypothetical work released primarily in a digital format.
3 That is, the actual work encoded as bits; not the encrypted version of it.