Just today I see that Hamas used a fake American front to sucker Israeli Jews into picking up and forwarding their packages. I hear that recently captured documents are allowing the IDF to unveil many PIJ/HAMAS deception operations.
Another wonderful SpyWeek. Keep them coming! The words "connecting the dots." however, reared their ugly head in the AI piece. Machines will not replace narrowly focused analysts and I must remind as I have written elsewhere: Five deadly sins of “connecting the dots”
(Lessons that should have been learned from Iraq WMD)
(1) Never enough dots
(2) Can just as easily connect bad or
misleading dots as good ones
(3) Easy to miss good dots on the margins
(4) Some good dots are considered irrelevant
(5) Some irrelevant dots are considered good
The key to producing better quality current reporting can never be achieved by just connecting the easily identifiable dots (what I call the obvious bits and pieces) that come streaming in every day. Far too many dots go by unnoticed (what I call the less obvious ones). Others stream by completely unrecognized (what I call the hidden dots). Thus, our guesstimates, even by the most skilled current reporters, appear to policymakers as mere opinion not evidence and as a result it rarely causes them to challenge their own strongly held beliefs. Intelligence officers should also never forget that policymakers often read or are briefed on the same obvious dots directly from collection agencies even before they receive the current analyst’s written report.
The key lies in identifying the less obvious and hidden data lying fallow in the gigantic haystack of information that grows bigger with each new collection cycle. The solution is not rocket science, nor is it AI. It can help, but not replace the human input. Analysts need to apply research methodologies to harvest new understanding from the haystack. In doing so they create new knowledge (more good dots) for the current analysts to use in their reports. Don’t forget intelligence analysis is no different than any other field of study such as history, medicine, biology, or physics. CNN’s Dr. Gupta does a remarkable job of explaining current medical events, but no one would ever expect him to find a cure for cancer. Only teams of researchers have a chance of pulling off such a miracle. For intelligence analysis we need both teams of researchers and current reporters.
Unfortunately, the senior managers of the Intelligence Community (IC) long ago decided to put all their eggs in the current intelligence basket. Research offices were dismantled, and their analysts let go or were put to work doing current reporting. Although we have many smart, capable analysts with all of them doing current reports, a tremendous amount of duplication of effort exists. More importantly “connecting the dots” can’t answer most of the policymakers’ principal questions. Until we recognize this and rebuild our research capabilities, we will continue to make mistakes both small and large like India going nuclear, the WMD fiasco, and the more recent misjudgments of Russian military capabilities and Ukraine’s willingness to resist. These are the very reasons we invest so much money in intelligence – no more Pearl Harbors or 9/11s. Instead, we do the easy things and completely ignore addressing the hard questions and mysteries that abound.
Thanks Carl for your thoughtful comments. Can you help me understand something? If humans are missing the hidden or less obvious data points too often then shouldn't AI be able to help us see them? AI could theoretically point the way to new research methodologies we hadn't even thought of before.
I don’t have a quarrel with AI. I think it could make a real difference, particularly in the right hands – a narrowly focused research analyst. What I object to is the almost total focus on current reporting and actionable intelligence of today’s IC. To improve the quality of analytical production the IC needs to recreate the research offices that it disbanded almost 40 years ago. “Connecting the dots” absence a sound base of research has proven unsuccessfull time and time again.
How many deep fakes have been used to condemn Israeli behavior in Gaza?
Hello John. I don’t have an answer to that—do you? I’m sure it’s going on.
Just today I see that Hamas used a fake American front to sucker Israeli Jews into picking up and forwarding their packages. I hear that recently captured documents are allowing the IDF to unveil many PIJ/HAMAS deception operations.
Thx. We’ll look into it. Lots going on.
Glad I found this substack, the writing and info presented is the best and suitable available. Thankyou. Happy subscriber.
Thanks so much, Chuck.
Another wonderful SpyWeek. Keep them coming! The words "connecting the dots." however, reared their ugly head in the AI piece. Machines will not replace narrowly focused analysts and I must remind as I have written elsewhere: Five deadly sins of “connecting the dots”
(Lessons that should have been learned from Iraq WMD)
(1) Never enough dots
(2) Can just as easily connect bad or
misleading dots as good ones
(3) Easy to miss good dots on the margins
(4) Some good dots are considered irrelevant
(5) Some irrelevant dots are considered good
The key to producing better quality current reporting can never be achieved by just connecting the easily identifiable dots (what I call the obvious bits and pieces) that come streaming in every day. Far too many dots go by unnoticed (what I call the less obvious ones). Others stream by completely unrecognized (what I call the hidden dots). Thus, our guesstimates, even by the most skilled current reporters, appear to policymakers as mere opinion not evidence and as a result it rarely causes them to challenge their own strongly held beliefs. Intelligence officers should also never forget that policymakers often read or are briefed on the same obvious dots directly from collection agencies even before they receive the current analyst’s written report.
The key lies in identifying the less obvious and hidden data lying fallow in the gigantic haystack of information that grows bigger with each new collection cycle. The solution is not rocket science, nor is it AI. It can help, but not replace the human input. Analysts need to apply research methodologies to harvest new understanding from the haystack. In doing so they create new knowledge (more good dots) for the current analysts to use in their reports. Don’t forget intelligence analysis is no different than any other field of study such as history, medicine, biology, or physics. CNN’s Dr. Gupta does a remarkable job of explaining current medical events, but no one would ever expect him to find a cure for cancer. Only teams of researchers have a chance of pulling off such a miracle. For intelligence analysis we need both teams of researchers and current reporters.
Unfortunately, the senior managers of the Intelligence Community (IC) long ago decided to put all their eggs in the current intelligence basket. Research offices were dismantled, and their analysts let go or were put to work doing current reporting. Although we have many smart, capable analysts with all of them doing current reports, a tremendous amount of duplication of effort exists. More importantly “connecting the dots” can’t answer most of the policymakers’ principal questions. Until we recognize this and rebuild our research capabilities, we will continue to make mistakes both small and large like India going nuclear, the WMD fiasco, and the more recent misjudgments of Russian military capabilities and Ukraine’s willingness to resist. These are the very reasons we invest so much money in intelligence – no more Pearl Harbors or 9/11s. Instead, we do the easy things and completely ignore addressing the hard questions and mysteries that abound.
Thanks Carl for your thoughtful comments. Can you help me understand something? If humans are missing the hidden or less obvious data points too often then shouldn't AI be able to help us see them? AI could theoretically point the way to new research methodologies we hadn't even thought of before.
Seth,
I don’t have a quarrel with AI. I think it could make a real difference, particularly in the right hands – a narrowly focused research analyst. What I object to is the almost total focus on current reporting and actionable intelligence of today’s IC. To improve the quality of analytical production the IC needs to recreate the research offices that it disbanded almost 40 years ago. “Connecting the dots” absence a sound base of research has proven unsuccessfull time and time again.
Thanks for this.