Bottom Line Up Front: This is a rebuttal to “Fixing Intel: A Blueprint for Making Intelligence Relevant in Afghanistan”(Flynn 2010), and includes my own assertions regarding military intelligence. This post is a series of observations based on my own experience within the community (spanning more than a decade) and a response to navel gazers within the community that have determined that specific emphasis on intelligence du jour will solve our nation’s problems.
I do not yet have a comprehensive solution. I am still working on it, however, even correcting symptomatic issues will seem unpalatable to most policy makers. Having said that, my job has never been one of pleasing decision makers, but rather, tactfully spoon-feeding unpalatable information to colicky adults. I harbor no resentment toward this because I’d like to think it has worked out well to this point in time. This one might be long folks, so I hope you have some coffee. Also…from here on out, fuck tact.
The Value of a Distributed Network of Small Analytic Cells
JICs (Joint Intelligence Centers) suck. Battlefield Surveillance Brigades suck. Stability Information Operations Center (Flynn 2010, 5) suck.
Military intelligence is a horrible hydra or chimera. Pick your beast. It seeks to stay beholden to the TO&E of traditional (and sometimes non-traditional) Army infrastructure. Little intelligence shops to big intelligence shops. TATs (Tactical Analysis Teams) to ridiculously over-staffed JIOC (Joint Intelligence and Operations Centers) or JICs. Policy and decision maker input or interface is at the JIC level.
Forget battlefield rotation. That’s like suggesting a night in the back of your Chevy with Suzie Rottencrotch is like the enduring relationship with your wife. Not the same at all. If someone were to suggest (outside of operational battlefield rotations, and even that’s questionable) that a battlefield rotation by the SecDef to each JIOC or JIC or S2 shop on the battlefield gave them a “real” understanding of what was going on I’d snort Rice Krispy’s through my nose in between chuckles of jaded laughter. What works? Starfish.
A sea of starfish actually. An effective analytic cell that is composed of no less than five analysts and no more than twelve. In the cell, no SIO (senior intelligence officers) or SME (subject matter experts) are required. In fact, there should be none in the form of policy. Instead, a team lead for a particular problem set is chosen. This team lead will drive the problem from inception to conclusion. Team leads may have multiple problem sets. Team leads will also be analysts in other problem sets.
Analyst selection is extremely important. Selection in this case is similar to a type of Special Forces selection. A brief outline of analyst selection might be based on Malcolm Gladwell’s “Outliers” (Gladwell 2008, 69-72). Analysts should be sought on the basis of divergence, not convergence. Questions as part of this process might be: “Why is a manhole cover round?” “If I give you a brick and a blanket, what are all things you could do with it?” Our recruiters need to be looking for Liam Hudson (Gladwell 2008, 70), not John Rambo or a would-be David Petraeus. Oh yeah…young analysts are retarded…old analysts are like the old Biff (Hello McFly!) in Back to the Future 2…annoying. You need maturity, but you don’t need gerontologists.
The kind of people you want as analysts do not generally abide by the rules or do not necessarily work well in groups. They may be introverts. They may be hippies. The analysts may or may not have had a military background. They may or may not have a criminal record. If you are lucky, the analysts have spent a quarter of their lives in some third world war zone as part of their maturation process.
Recruiters should also look for hybrid analysts. “I like to do intelligence analysis AND graphic design.” “I like to do intelligence analysis AND code in Python.” “I like to do intelligence analysis AND I like to shoot and collect weapons from around the world.” “I like to do intelligence analysis AND spend ALL of my off time penetration testing.”
I know a good analyst five to ten minutes after talking to any random person. I gravitate toward them. From my own experience you’ll be lucky if you get a good one. Very lucky. Try getting one out of my kung-fu grip. You need many of these cells augmented by a long tail of software and…
Value Added Small Analytic Cells: Coders
You need coders. Computer scientists. NOT IT dudes. Those are J6 guys. (Love the help desk dudes! (Had to say that to keep my shiznit running.) Uber-nerds supporting nerds. Guys and gals who prefer hexadecimal and binary over the spoken word.
They are part of the cell of the analysts. Two to four coders per analytic cell. They bring the necessary computational expertise to the cell. They can code pre-processors. They can code scripts and API’s to force open source software to work with each other. They can disassemble a CPU at the logical and physical level. In addition to their expertise in this area, their insights into analysis from a computational standpoint have greatly benefited work in the past.
In particular, from a workflow management perspective; because coders… Uhh… are inherently lazy. Smart…but lazy. Put another way, any coder worth a damn will find the simplest and potentially most elegant way to write a code block because it will save time in the end.
To finalize this, you now have a fourteen-armed starfish. You need hundreds of these. Maybe thousands. Some of these starfish are working in groups. Some are working individually to support a specific unit. No JIC’s. If you get a VIP or DV in, you corral the cells into a briefing area and have them brief their stuff. The key to this is the “long tail” (Anderson 2006). You need industrial strength software. You need Pacific Northwest National Labs IN-SPIRE, FuturePoint’s Starlight, Semantica Pro, Palantir, Rapid Miner, Tableau, and Google Refine.
Some of these are actually free. IN-SPIRE is generally free to government customers. Rapid Miner is free. Google Refine is free. In fact, you can make the entire tail (if needed)…free. Gephi and or NodeXL are free social network analysis programs. Ontotext is an awesome semantic relationship engine.
Why spend thousands of dollars on Palantir when Gephi will do? I have no idea. The key is to let your analysts and coders decide. You have to trust and respect them to make the best decisions (in this aspect) for you. A good analyst will not seek every program under the sun to do the work you need done. They will seek the cost/benefit, maximizing utility and minimizing cost. That’s their part of their job anyway.
“Fixing” implies that something is broken. There is no fixing intel. Our intelligence community is focused almost exclusively on terrorism. Fail. To paraphrase Kanye West, imma let you finish an all…but empty chessboards is what needs to happen. We need to push the pieces off and start with a fresh game constructed around a 2013 reality. Not 1945 reality.
John Robb is probably the most prescient voice in our community. I can’t recommend his blog or book enough. Right now…Charlie or Tango is envisioning a UAV bot-net that supplies on-demand DDOS of a specific topographic area. It’s fucking cheap, and its real.
On the other end of the spectrum, in an interesting bit of irony, the PRC is attacking the US using US economics. Supply and demand is the weapon. This can be done if you have a command economy to fight a free market, particularly if you mobilize an entire nation toward the effort of attaining R&D at any cost. The PRC is undermining us at the one thing the US better than anyone else in: Joint operations. Kill joint C4, and you kill the your enemy’s greatest strength. How do you kill it? Steal the secrets to command and control. Networks. Military networks.
That war is being waged right now in a virtual environment and starts with a “sync ack” packet. Companies don’t want to tell the public this. The government doesn’t want to tell the public this. Remember that your economy is based on “fiat” or, loosely paraphrased, “a proposition.” The stability of our economy and government give value to our markets. If you are losing 40% of your R&D investment to another nation-state due to larceny, I’d say you need to start worrying. They like to refer to this entire process as the “Shashou Jian” strategy. Oh yeah…that’s right..they are telling you how they plan on ripping you off. All the way back in 1991.
Succinctly put, we have a national system to manage, identify, and reciprocate against violent extremist organizations and terrorism. Your chance of dying by a terrorist related event is less than being attacked by a shark. Yet there is a 50% chance that a 9/11 style event will occur again within the decade.
Given those figures you might assume our community is looking ahead. You would assume incorrectly. The intelligence community is not looking at the analytic future anymore. We are looking at the past and the present. Afghanistan and maybe some Africa. So much for a strategic pivot.
More importantly, every day more human behavior is added under the auspices of national security and intelligence. Anthropology, nuclear science, epidemiology, machining, engineering, computer science, etc…etc. Note, I said ‘human behavior.’ Intelligence isn’t and shouldn’t be “human intent guessers.”
That said…there is nothing that humans do, that is not their behavior. War? Human behavior. Drugs? Human behavior. Theft? Human behavior. Etc..etc..
From a positivist standpoint, observable and repeatable matters. As an analyst, I got news for you. I can’t process that aforementioned cognitive load. I need computers like bats need echolocation. I need Linux and Windows boxes and about fifteen computer screens. I also need decision makers to make some decisions.
Actions officers…take some action.
I’m an analyst…I will give you COA’s (courses of action). You gotta trust me to die/sleep at my desk/spend days without sleep/shit in a desk drawer/do anything to get you the best possible picture of your battlefield (which is the globe BTW), but for goodness sake…
Make. A. Decision.
Even if its not a COA. General officers…make strategy. Not “Operation Ivegotabigwang.” That’s not strategy. “Strategery” usually looks something like a joint campaign plan…only better. Do like the Chinese, and copy theirs! Rumor has it operational art is dead and so is strategy. Bad time time to die. We need it. Badly. Check out how the combined chiefs of staff used to manage their commanders. One page. Yet they had a strategy.
Deep Blue versus Garry Kasparov
From a fresh board, start with your knight…not your pawn. This means understanding the laws of thermodynamics and closed systems. Based on the second law of thermodynamics, closed systems will increase to maximal entropy.
In data and computational theory, entropy is defined as “a measure of the loss of information in a transmitted signal or message.” (Lexico Publishing, LLC, 2013) Our system of classification as overseen by CAPCO (Controlled Access Program Coordination Office) does not function effectively and is outdated. It is a closed system.
Today’s analyst is confined in a T-SCIF (Tactical-Secure Compartmented Information Facility) or SCIF. Information is fed to the analyst over government circuits. The circuits are classified. The analyst will occasionally have an unclassified circuit controlled by the USG.
Use the starfish network. Start with a misattribution system, force analysts to start on the unrestricted internet, and then move up the chain of circuits from SECRET to TOP SECRET. Use sites like Data.gov, media, social media, open source data repositories. These are your first weapons of choice.
Then, when you run out of valid data…hack. Hack, hack, hack, hack, hax0r, hax0ring. Just do it. Do it now. Back Orifice, E-Bomb, DDoS, bot-nets, malwarez, warez, keygens, brute force, overclock CPU’s and shutdown the fan via killer poke.
If you can’t get it, shut it down.
Use the internet to scrape all your open source data and consider hax0ring as NTM (national technical means). Then when the scrape cycle has completed once, push it up. Push it up the classification ladder and cross-cue the data using the aforementioned tools with national technical means.
Think of the internet as the ocean, then the next level as your first catch, and then the last level the best of the catch. Then push it down. Take that and refine the scrape cycle. Use analytic methodologies.
Consider eliminating the classification system and using obfuscation on the low side. Create 50 different variations of the same message broadcast and then re-broadcast. Tell your receiver the correct variant. 20 ICBMs? Or five?
How can your enemy tell the difference? Which message is “real”? Use the media to hype up different variations of the right message. Yes, I mean actual traffic derived from different sources. Tell the NYT we have 20. Tell the Post we have eight. If you have 52 cards and all of them are aces with each containing a minor change, which ace is the true ace? How long would it take you to tell?
As it stands it may not be any different from the leak-ridden culture of one million cleared personnel you have now. Use the massive information barrage to your advantage and beat your enemy’s OODA loop. Open source warfare is not for the timid, and its encroaching on an entropic system that is archaic. Information always seeks to be free. More rebuttal to come?
End of Line.
- Anderson, Chris. 2006. The Long Tail?: Why the future of Business is Selling Less of More. New York: Hyperion.
- Flynn, Michael T. 2010. Fixing Intel: A Blueprint for Making Intelligence Relevant in Afghanistan. Voices From the Field. Center for a New American Security.
- Gladwell, Malcolm. 2008. Outliers: The Story of Success. New York: Little, Brown and Co.
- Lexico Publishing, LLC. 2013. “Entropy.” Dictionary.com. Accessed January 22.
- Zhongwen, Huo, and Wang Zongxiao. “Sources and Techniques of Obtaining National Defense Science and Technology Intelligence.” Kexue Jishu Wenxuan Publishing Co, 1991.
Some images courtesy of
- Joint Warfighting Center, and Joint Concept Development and Experimentation Directorate. 2006. Commander’s Handbook for an Effects-Based Approach to Joint Operations. Handbook. Suffolk, Virginia: Joint Forces Command.