Peter,
First, I'll take my cookie now. 
Some good musings here but I think there is an aspect of intelligence that you're missing and that is the way in which it works to inform one's own clandestine policies--which aren't always limited to espionage.
Imagine, for a moment, a covert effort aimed at thwarting the Iranian nuclear program years ago. Early intelligence about nuclear efforts might have been used to inform countervailing efforts that were similarly "secret." Whether "carrots" or "sticks" these sorts of efforts conducted out of the limelight of public attention might have been tremendously more effective than the blunt policy tools of military force, obfuscated multilateral negotiations, and economic sanctions that American hegemony, er . . . I mean, the international community is left with today.
--
The claim I see here is that secret humint is not very reliable. It tends to be double agents. (If it's, say, 2/3 double agents and the double agents present a story that fits together and makes sense while the other 1/3 tell a different story that probably doesn't fit together as well, not being tailored to do that, which will we believe?) It tends to be people who don't actually know what we want them to tell us. Sometimes it's people who lie to us because for one reason or another they want to be important to us.
If our data is not very good, why would our covert operations be good? Say for example we start assassinating iranian physicists and technical workers. You can only do so much of that sort of thing before they start noticing. But the fact that our data was bad and we'd be killing a bunch of the wrong people might actually help us -- they'd be confused about our purposes because we were so clumsy.
Having worked both as a political science professor and as an analyst (with a TS/SCI), I have to say that that I disagree on several levels.
First, HUMINT is simply part of the intelligence puzzle, together with COMINT, OSINT, and classified diplomatic reporting (assuming here that we're talking here about political intelligence--if it is a more technical matter, like nuclear capacity, it gets even more complicated). I'm not sure that the Weisberg piece really makes this clear.
(The press, I must say, does a pretty good job, and there's a lot to be found out there in the OSINT world. However, a diplomatic mission, with good FSOs and an ambassador who knows how to encourage his/her team to produce value-added, can produce diplomatic reporting that can be much better than anything in the press, and--on current issues--often much better than anything the academic community is producing. )
Second, you're right about the need to protect means and sources. No matter how hard the HUMINT or COMINT reporting often tries to disguise the origins from the analyst, it is often not hard to work out (and potentially compromise) the data source from the data that they are generating.
Third, material has to be classified not only to protect methods and sources, but to avoid antagonizing allies and other actors, and so as to not to compromise national interests.
Can some intelligence assessment be declassified so as to contribute to public debate and inform the democratic process? Yes, they can (as the Iran NIE shows). Can this always be done, or even done most of the time? No. Indeed, its noteworthy that not all the main findings of the NIE could be declassified, nor could the actual text of the report itself.
Finally, I don't quite understand where Weisberg is getting the "myth of the super-spy" from-- high-level sources are nice when you can get them, they hardly represent the primary target of HUMINT collection efforts.
Commenting by HaloScan.com
