What is most interesting about this thread, which deals with a very important subject, is that there has only been one comment from a legal expert, and that brief and restricted.
I don't think the issues can be solved. I don't know whether they can be resolved, and even if they can be, I don't know whether a resolution can be established for the duration or if it has to be continually renegotiated.
PJ2 is quite right that the "discourse of safety/accident investigation" is different from the "discourse of responsibility" (I avoid the perjorative connotations of the "b" word). Both are human negotiative processes, and both can proceed apprropriately or be abused, and are, regularly.
Anyone who has been involved in civil court proceedings knows, as a colleague of mine memorably put it in an address to the International System Safety Conference in Ottawa in 2004, that such proceedings are not primarily about finding out what went on or what is going on, but simply about reaching a conclusion by some means or other. My take is that the point of view "if I/you do everything properly, then I/you have nothing to hide, and therefore I/you can have no significant objection to data in which I have an interest being made public" does not survive any personal experience with antagonist proceedings, legal or administrative.
Let us put some flesh on the bones of PJ's contention that these are different "discourses" (or "negotiations", or whatever your preferred technical word is).
The fact is that pilots, and controllers, work a safety-critical job in an environment that is almost never optimal and often involves contradictory requirements. They have chosen to do that job, and each and every one will have some method of resolving those contradictory requirements, by adhering to some and violating others. And when a person ends up in antagonist proceedings, hisher antagonist is going to emphasise the requirements that heshe did not fulfil (and there are inevitably some).
For examples, think of the constellation of issues around fatigue, the commute to base, low pay, and family stability obligations, which almost every "commuter airline" pilot has to try to resolve somehow. (I am sure many if not most would go for a relaxed commute the day before, a stay overnight at the Airport Hilton, and a decent breakfast before starting the day's - or night's -work, but that has traditionally not been on offer. Does anyone seriously think Colgan's accident pilots rode jump seats to work in the middle of the night and slept on couches because they preferred to?) And if that doesn't grab you, then think about how pilots and controllers cope with malfunctioning equipment, such as the controllers in the Amazonas involved in the Legacy/GOL mid-air collision that were supposed to be working in the environment of (according to the authorities) "full radar coverage" which clearly wasn't. And if that doesn't grab you, then maybe it is time to familiarise yourself with Jens Rasmussen's work on "migration to the boundary", which is his word for adapting to contradictory constraints.
So, a "safety" discourse will attempt to consider this environment, how an operator adapts, and whether this adaptation is appropriate or not. In reference to Colgan, it would consider, for example, making employers responsible for providing adequate rest accomodation (a hotel room) to pilots before a shift, at no cost to the pilot, and introducing measures to ensure a pilot uses it.
A "responsibility/blame" discourse will point out that the pilots broke company regulation concerning rest time, and may well attempt to argue that the pilots thereby brought themselves into a state of fatigue. If they had "deep pockets", there would follow the argument that they are wholly responsible and therefore liable for the damage they caused. Since they did not have "deep pockets", it will be interesting to see how the argument then progresses. It might try to put responsibility on the employer, who will then argue that it has clear working practices in place to which their employees contractually obligated themselves, that the working practices were adequate, and that the employees violated their obligations.
It is obvious to me, and, I would argue, should be to others, that these discourses are different.
PJ now wants to argue from this, and I think we should all be able to see why, that certain kinds of data, which everybody agrees is essential for the safety discourse, should be protected from the responsibility discourse. It is, I hope, easy to see on the basis of the foregoing considerations why this is desirable.
However, and this is a very big but, the discourse of responsibiity has three thousand years of history behind it, some two thousand nine hundred more than the discourse of aviation safety. And has developed certain principles. One of these principles is that which I will crudely call disclosure (our legal colleagues can correct me if this is the wrong word). I put it crudely as follows: if there exists evidence pertaining to a case, then parties to the case have a prima facie right to examine that evidence. Further, they have equal right to the evidence; if one has access to it, then all.
Now, the rules regarding what is "evidence" and what is not are pretty complicated, and by no means the same in all jurisdictions. Hence disputes about "admissible evidence" (that phrase really just means "evidence").
Now, it is all well and good for a legislative body (Parliament, Congress) to decide, and enact a law, that no one shall have access to a particular document (by which I include all manner of stored record of past events) such as a CVR recording. But the legal principles above are much older and more well-established, and the judiciary can well decide that plaintiff's or defendant's fundamental rights trump any recent legislative decision to keep some things secret, especially if it can be shown that that secrecy is somewhat threadbare.
That is the nub. That is my take on what PJ and others are worried about. It is not clear to me that this conflict has any solution.
PBL