Friday, May 20, 2011

Let's Kill The Intelligence Cycle (Original Research)

I mean it.

The "intelligence cycle", as a depiction of how the intelligence process works, is a WWII era relic that is way past its sell-by date.  It has become toxic.  It no longer informs as much as it infects.  It is less a cycle than a cyclops -- ancient, ugly and destructive.

I want it dead and gone, crushed, eliminated.

I don't care, frankly, what we have to do.  Remove it from every training manual, delete it from every slide, erase it from every website.

Shoot it with a silver bullet, drive a wooden stake through its heart, burn the remains without ceremony and scatter the ashes.

(Geez, Kris, why don't you tell us how you really feel...)

OK, OK, so, yes, I am being intentionally provocative but I have been doing quite a bit of research on the intelligence process over the last several years and have come to the conclusion -- as have others before me -- that our current best depiction of this process, the so-called "intelligence cycle" is fatally flawed.  Moreover, I believe these flaws have become so severe, so grievous, that continued adherence to and promotion of the cycle is actually counterproductive.

My intent, beginning on Monday and over the next several weeks, is to lay out the evidence I have gathered about the cycle itself, about attempts to save it from its worst flaws, about attempts to replace it altogether and let you decide for yourselves. 

In the end, I intend to recommend (with no hubris intended and well aware of the possibility of hamartia) my own generalized version of the intelligence process; one which I think is more appropriate for the intelligence tasks of the 21st Century and which works, in both theory and practice, across all three major sub-disciplines of intelligence -- national security, business and law enforcement.

Next:  The Disconnect Between Theory And Practice


Anonymous said...

Many of us are looking forward to reading your findings! Draw out corruption so that we may move forward into the future productively

Anonymous said...

If you don't mind, I'll add a thought to this.

As an analyst models that we develop are do so for a variety of reasons. They are also prepared for different levels of detail depending on the nature of the analysis or explanation of a something.

It is unwise to attempt to create a model of a process that is holistic, all-encompassing, and of practical value at all levels of detail. Such models are incredibly complex, the linkages and description of the linkages (which for the intelligence cycle would number in the tens of thousands or more) are difficult if not impossible.

Therefore, a thoughtful analyst will develop and use a model that reflects only the nature of a system that the analyst needs to deal with.

If you wanted to explain to the general public what intelligence was and how it works the current model is exceptionally simple. It is partly why that model has become so popular and widespread. While it serves a purpose those of us who actually engage in intelligence operations (of any kind) realize that it is inadequate to support our needs. Therefore we attempt to describe the process in a more "realistic" way. We are naturally conceited enough to believe that only our more involved and detailed description is correct. Take a look at Robert Clark's Target Centric Analysis. He redefines the intelligence cycle and translates the high-level, public-consumption model down to the level that a technical analyst would be able to understand. It is a usable model at that level for that type of analyst. It is good. In fact, if I were bringing on technical analysts and training them I would start with Clark's model. Is it perfect? No. Of course not. Models are a representation of reality--not a complete description of reality. It has its flaws and holes like all models. But it has a purpose, it serves a specific group of analysts, and therefore has considerable value.

Paul Cooke
Professor of Intelligence

Anonymous said...

I don't think it is necessary to gather evidence about the intelligence cycle to prove or disprove it, to bring its flaws to light. That is a destructive activity rather than a constructive activity. There is no question that the current model at the highest level has value, appeal, and meaning. That value might not be to you, your peers, or many in the intelligence community who live a real-life intelligence process everyday. But for those outside, looking for a way of grasping onto a framework it is meaningful.

I also believe one more point is necessary on this topic: models are not processes. A work process is a highly detailed, lowest level description of actual functions, flows, and decisions. It includes specifically described inputs and outputs. At a car manufacturing plant a work process is physically and mentally what workers do to manufacture cars. It isn't a model. Work process or any kind of process (manufacturing, service, intellectual, governmental, etc.) must by its definition be complete in order to have value. Processes at a nuclear power plant are highly detailed precisely because the tasks and activities of the work must be understood and controlled. This is the same with the processes that take a satellite to Vandenburg, marry it with a booster, and put it into orbit. It is also the same with HUMINT collectors, covert action operators, etc. In fact, the "intelligence process" is so vastly large and detailed and intricate that it is impossible to describe each piece. HUMINT collectors do not sit at their desk and document exactly what they do.

This is where models come in. A recruit is sent to the Farm and is given both work process training (which is detailed) and they are given models that represent an intended reality (and the recruits move through exercises to carry out the intent of the model while never bothering with specific work processes).

A model, as I mentioned, is a reduced representation of reality. These models are typically developed to support the needs of an individual or group of people who are responsible for some element of the contents of the models.

Therefore, I think it would be a good idea to conceptually break the notion of an intelligence process from a model of how intelligence functions. These are two different things and have two different purposes. I don't think it is wise to try to mix them together--in fact it would be impossible. The second recommendation is to not focus on why the intelligence cycle, as described, is wrong but rather to focus on what models make sense for what functions and disciplines in intelligence or outside intelligence. A HUMINT operator will have a different, very different, model than an intelligence analyst at a desk at Langley. These differences in needs and levels of models abound from those who work MASINT sensors to those who are briefers to those who examine imagery. A user of intelligence in Congress has a very different need than a O-2 in Iraq.

(continued from previous)

Paul Cooke

GirlSpy said...

Looking forward to this one. I am most definite that it will be interesting.

Kristan J. Wheaton said...


A quick response. There is much we agree upon and I will touch on many of your points throughout the series.

For now, I agree wholeheartedly that how we describe the process needs to be simple but, I would maintain, it also needs to be as accurate as possible -- it must not only inform outsiders but also resonate with insiders. I will begin to talk about the dangers of the current disconnect in my next post and will carry that theme throughout the series.

In addition, I admire Clarke as well and his book is required reading here. I think the target-centric approach is an improvement over the classic intelligence cycle but I think that it, in turn, can be improved upon.

I am not sure I understand your point about "not gathering evidence". If intelligence is a process then it can be improved upon. In order to decide how and where to improve the process, you must have evidence to evaluate. I am not sure how you can evaluate a process in a vacuum but perhaps I do not understand your point.

I categorically disagree with your assertion that "There is no question that the current model at the highest level has value, appeal, and meaning." The whole purpose of this series is precisely to question the value of this model at the highest level.

You have always been a thoughtful and insightful writer. I look forward to your additional comments as the series progresses.


Anonymous said...

please guys, I thought we agreed long time ago that intelligence cycle is an old fashioned, tax-money waster, generally for daft blocks that prefer hierarchy over accomplishments. Changes are inevitable, accuracy (especially in "our" world) non-existing, so "let's kill the intelligence cycle". Mr. Wheaton, cheers to your teachings!!!

Unknown said...

Kill the intelligence cycle? All the mid-level process drones in the Army will have to develop new doctrine to help clear up their thinking.

Anonymous said...


There is a more basic concern than whether there is accuracy in a model. ALL models lack accuracy. By definition they are reductions of complex reality to a usable form. Pursuit of a perfectly correct model is, as a result, only relative to the purpose of the model and the audience that uses that model.

For those just learning about intelligence, the intelligence cycle gives a construct for thinking that has value. It indicates flows and activities. But it is hardly accurate in the truest sense because it is a stratospheric distillation of an extremely complicated system. But it has value relative to the audience.

You are certainly welcome to question the value of the intelligence cycle as it is described. Perhaps there is a better form that will reach and have meaning to the same audience.

The point about not gathering evidence was related to the first task in model building: knowing what the purpose of the model is and who the audience is. I would challenge you that this comes before any sort of data gathering or proof.

What sort of model would be ideal (or at least useful) for those who spend their days reading message traffic? What about those who work with contractors to design and develop satellite platforms and sensors? For each job function in the IC there is a separate most optimal model that addresses their specific need.

What need are you addressing with a relook at the intelligence cycle? That is what is unclear. Who are you trying to assist by improving the model?

Generally you need to classify the audience first by some sort of similar need for information. But you have to remember that "Analyst" is not enough detail for any reasonable classification. The function of analysts is incredibly broad.

As for disagreeing that the intelligence cycle has value, I think it can easily be stated that it does merely by its use in so many introduction to government classes when the topic moves to intelligence. It is used, therefore has value.

Think about this dispassionately. Hypothesize that the intelligence cycle model is a poor representation of intelligence activities for some category of persons. Then set out to collect evidence that is balanced and objective. You might be surprised by what you find. You might even consider a survey of political science professors who teach Freshman and Sophomore courses on government and ask them if the model suits their needs for depth, complexity, etc. That would be appropriate because these models are typically used in instruction of neophytes. As long as you are careful about what population you want to gather on and the sample that you will use you will have some interesting results.

Paul Cooke

Anonymous said...

I have just finished an interesting book "Hidden Empire" by Orson Scott Card. It is written in an interesting format where there are prologues to each chapter. Many are highly prescient to today's events while also being appropriate to this discussion.

For example:
Pg 143: People know many things, and half of them are wrong. If only we knew which half, we'd have reason to be proud of our intelligence.
What is Knowledge? A belief that is shared by all rhe respectable people in a community, whether there is any real evidence for it or not.
What is Faith? A belief we hold so strongly that we act as if it is true even though we know there are many wo do not believe it [and often ignoring significant evidence against the belief - HIN]
What is opinion? A belief that we expect other people to argue with.
What is scientific fact? An oxymoron. Science does not deal with in facts. It deals in hypotheses, which are never fully and finally correct.

Pg 162: Human beings are not [biologically] designed to keep seccrets. Every aspect of our being is shaped for the sharing of information -- through speech, gesture, facial expression, posture, and every other deliberate or inadvertent sign of emotion and intent.
Thus, it should not surprise us that every would-be dictator, tyrant, conqueror, prophet, colonizer, politician, artist, and dog-catcher in history clearly signaled his intentions long before he acted, and in plenty of time for others to prevent them. Neither Hitler nor Churchill, neither Pol Pot nor Abraham Lincoln, ever did anything they hadn't told us and shown us they would do.
That they are rarely prevented has more to do with out inattention, cowardice, or ambition to ride his coattails than with his particular skill. dogs might run from the dogcatcher as soon as they see the net, but they rarely tear out his throat and kill him, which is, of course, the only rational course of action for the dog that values its life, liberty, or happiness.

Harry Nimon PhD

Anonymous said...

The perceived environment dictates the matrix

Eric Denécé said...

Hi Kris
I really like your comment and I totally agree.
The intelligence cycle is a only a very general tool that can help beginners to understand how intelligence globally works.
But it's not an operational tool.
The problem is that most of the people who don't know what really is intelligence are addict to this imperfect method because they don't have any other reference or experience.
In France also, is is almost an ideological tool.
So let's fight on both sides of the Atlantic
Eric Denécé

Arthur Weiss said...

I'm really looking forward to your conclusions. I've also felt for a long time that the current intelligence cycle doesn't work. In fact I also find it dangerous for business decision making - even in the terms used.

As an example, dissemination is a one-way process. In order to refine the process and ensure timeliness and relevance there needs to be a communication process - with the decision makers providing feedback to the analysts. So change the word "dissemination" to "communication" and you've improved the process.

There is also no mention of the stochastic collection of serendipitous intelligence. That always happens - and should be incorporated into any intelligence process. The current cycle doesn't allow for such information.

My biggest gripe is that there is no feedback and also there is a suggestion that analysts, collectors and decision makers are all together. This is nonsense - and they should be seen as separate but overlapping functions rather than linked as suggested by the cycle.

Intelligencier said...

Hello Gents,

I am a little late in the debate. However, I am at odds with the views on the Int cycle. It is designed to be a guiding framework - doctrine. For example, Intelligenciers are not bound specifically by the sequence of events of the cycle, although almost every piece of information that enters the cycle or every direction given will be subject at the very least to a cursory glance and mental int cycle process. The cycle itself, particularly in the UK version is fully encompassed by a circle of continous review. The cycle is doctrine, designed not to constrain but to enable analysts and other int personnel to process and manage intelligence. The bottom line is that the cycle is and has to be simple enough to both teach, understand and implement as a doctrinal enabler. as such it has stood the test of time (when implemented properly) and is a key anchor in busy int environments. I also offer a final thought, and challenge anyone to define an intelligence process or activity that does not fall into one of the elements of the DCPD model, encompassed by continous review (the gauntlett is down)! This is where the value and framework of the cycle has its value and long should it remain. I look forward to your rebutles.

Arthur Weiss said...

Looking forward to your reasons but I've been saying for several years that the traditional intelligence cycle should not be used and is not fit for purpose. At the same time people like a process and expect one - so I wouldn't kill it completely. Instead look at what is wrong and replace it with something better.

The main problem is that it is a cycle - it goes one way. So yes - intelligence analysts say you need feedback, but it's not in the cycle, so where and how! However there's more. As you say it's left over from WW2 military intelligence and that was over 70 years ago. Do most of us still live in a command and control environment which is what the current model implies. I use a model called the 4Cs (published in a number of articles I've written - most recently in the Palgrave Handbook of Security, Risk & Intelligence in the Corporate Intelligence chapter.
1) Planning & Direction. This is pure "Decision maker telling analyst what to do". This leads to errors if the decision maker lacks objectivity or doesn't fully understand the issue and so requests something that won't achieve the required results. Replace this with Conversation - where the decision maker / end user discusses the requirements and needs and comes up with an action plan to obtain the intelligence. Note that this is a 2-way process and interaction rather then 1-way process planning & direction implies.
2) Collection - the second C. A mix of OSINT and HUMINT depending on what is needed. Sometimes mostly OSINT. Sometimes HUMINT is needed.
3) Conversion - take the collected data and analyse it, catalogue and tag it for future use, interpret it and confirm and check it - ensuring that the sources used are valid and the information makes sense. (This last step is crucial. Too often intelligence is gathered but the sources are not checked / confirmed - so you end up with something that may be disinformation but as it looks believable it ends up being used with sometimes disastrous results. If at any stage there are gaps, then go back to the collection stage (i.e. feedback on what you have) and if it raises questions on the objectives, go back and clarify with the end users. The objectives can change as a result - so another feedback step. This step replaces Processing & Exploitation and Analysis & Production but is essentially the same but covers this and much more as the aim is to convert the intelligence into something that the end user can own and use. To do that it needs to be communicated - the next C
4) Communication. Communication is a 2-way process checking that what you are putting across is understood. It also is part of checking as if the end user(s) do not understand they can ask for more or clarification. In contrast Dissemination is a one way process implying all is done and dusted and time for the next task. Instead the result of Communication involves both sides - the user and the analyst.