Alas, the Felt case isn't nearly as cut-and-dry as that, is it? Most of the discussion floating around the past few days has focused on the politics of it, with a bureaucratic turf war being the more likely source of motivation.

> I think it is pretty clear that all our institutions, and also our
> models of financial cryptography support the concept and presence of
> whistle blowers. It may be hell when he's not on your team, but that's
> a different issue.

Is this the case in the big picture? While covert channels to press from insiders are great, do many systems allow plausible deniability of misdeeds? Or internal access audit trails, allowing an annoyed boss to determine which high-level underling grabbed an anomolous file that ended up in the press?

I think that the problem runs deeper. The [limited liability employer] ~ [guaranteed salary employee] relationship is fundamentally flawed, IMHO. On a moral/ethical level, it leaves noone truly responsible for corporate actions. Managers are following their fiduciary duty to the shareholders, shareholders are just investing their money, employees are just following executive orders. Everyone has a near-perfect excuse, clearly limiting their responsibility. At the same time, there's nothing to limit the damage from corporate misdeeds. And its a double-edged sword, too: employees can (and do) cause unlimited damage to the company and its reputation/security/bottom line and get away with it -- they only risk their salary, after all.
Sure, there's criminal responsibility too, but that doesn't apply in many cases. One can cause a lot of damage without committing a crime by simply being inefficient. Also, the costs of criminal proceedings are prohibitively high in many cases.
John Kornai has developed a neat economic theory to explain the flaws of planned economies. It has two crucial elements: the soft budget constraint and the plan bargain. It is the second one that is of interest here: basically, those who are expected to follow some plan have huge economic incentives to misrepresent (make the planmakers overestimate) the resources that they require to fulfill the plan. They might even collude with those up and down the technological chain. Now, I am more and more convinced that large capitalist corporations are no different from the USSR (or was it the other way around?). Inner budget constraints are soft, people are coordinated by bureucratic orders (rather than market incentives) and the whole system is riddled with purposeful disinformation.
Whistleblowers often find themselves between two (or many) fires, since the lies are agreed upon and in the best interest of large numbers of people, often in the position of power: planned economies do not tolerate whistleblowers, even though they are beneficial for the system as a whole. Every planned economy has systemic problems with incompatible incentives.
I think that as capital stock (all the stuff that makes people more productive) is getting smaller and cheaper, the XXth century models of organizing cooperation need a rethink: maybe the economies of scale are not worth the loss of efficiency due to incompatible incentives and insufficient information anymore?

The whistleblower concept is so simple and clean in concept, but disgusting in practice. When I say that our ideas and practices are aligned towards whistleblowing, I mean in the sense of audit trails needing people to follow them. If auditors aren't doing that job, then the only one left might be the whistleblower.

In practice it's nowhere near as easy as that. Most whistleblowers will have their lives destroyed and never work again in that field. It is by no means a costless action, they are really putting everything on the line, so when we say that at the high level we support them, what we are really saying is that if things have got bad enough that someone is going to throw everything away to get the truth out, then the truth must be important.

Any criticism like "bureaucratic turf war" must be seen in that context. The whistleblower will be faced with a barrage of attacks designed to utterly destroy them. These attacks might be fair and just, even, but they are only a response to the truth that is put out there. If the whistleblower kept his mouth shut and didn't reveal the truth, the attacks wouldn't happen.

(The actual specifics of the Felt case I'm unaware of - and perhaps deliberately so. The principle of whistleblowing I find very interesting because it challenges our notions of reliable systems. How can we build secure systems if they keep getting blown up by whistleblowers? Are we really so far away from the mark?)