ROBERT STEELE: I am doing fact-checking as I create the first textbook on intelligence that is useful to all eight tribes of intelligence and breaks away from the mis-directed focus on expensive useless secrets for policy alone. Below from Colonel GI Wilson, USMC (Ret)
Yes, sir. This is how I recall it and Zinni confirms it.
Zinni asked his J-2 at CENTCOM how much of their usable intell was open source. His J-2 said 80%. Zinni asked his J-2 how much of the 20% classified could be found or expanded in OSINT after was it was initially indicated by classified sources. It was determined to be 80% of that.
in my personal experience working with Zinni, it was very evident Zinni recognized the role of OSINT plays in the intelligence community concordant with classified intel. Zinni like Gen Al Gray never had a predilection or penchant for an over-reliance on classified intel just because it was classified.
GI, writing a new book, INTELLIGENCE WITH INTEGRITY: Decision Support in the Public Interest. I'd like to identify you as the source if I may, of this statement from Zinni:
 As related to me by Col G. I. Wilson, USMC (Ret.), one of a tiny handful of asymmetric warfare pioneers in the US Marine Corps, as he heard it from General Zinni. This was first published, without attribution, in Robert Steele, “Open Source Intelligence,” in Johnson (2006), Chapter 6, pp. 95-122. Subsequently I created Graphic: Tony Zinni on 4% “At Best”.
ROBERT STEELE: Recollection vary on how the 4% was communicated earlier. 20% of 20% is 4%. I am reminded of the OSS.Net Quotes Page, where the follow appears:
Henry Stiller, Director General of Histen Riller, said at the French Information Congress – IDT '93, that, 95% de l'information dont une enterprise a besoin peut s'acquiris par des moyens honorables. “95% of the information an enterprise needs can be acquired through open means. 
 Observed personally by Robert Steele and included at the beginning of “ACCESS: Theory and Practice of Intelligence in the Age of Information.”
I continue to believe that the best intelligence (decision-support) is that intelligence that can be shared, subject to scrutiny by all, and leveraged by all to harmonize their varied as well as their coincident endeavors. Secrecy is cancer. In my view, counterintelligence — ruthless pervasive counterintelligence — is the ONLY domain within intelligence that merits absolute long-term secrecy and extrajudicial action (one man, one bullet) but ONLY if intelligence as a whole has integrity.