« Clapper will probably be DNI | Main | Shaping the Iran story - Giraldi »

22 May 2010

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Castellio

Interesting, a desire to base intelligence on the capabilities of machines versus intelligence based on the intentions of those who control the machines (the national context).

The idea that if you squeeze out the human your understanding will be better...

Cloned Poster

No NO NO,

Database upon which of the three Gods are followed sounds so much easier.

Fred

Castellio, you hit the nail on the head with that comment. Works the same way in any organization too. We humans just don't tend to be mathematically predictable (Harry Seldon's not here yet).

Jose

Castellio, nice way to make himself popular with the Neocons by providing the intelligence in a useful way to continue the agenda.

All those people pushed out were probably replaced by more dependable or amendable types, IMHO.

walrus

When a manager does not understand the human dimensions of their task, their immediate action is almost always to look for, and fasten on to, a technological solution to their problem.

Machines are simpler to understand, and much more predictable, than people.

At least it appears Clapper is/was capable of correcting his own errors. There are many who cannot and will not.

Adam L Silverman

The tech over human approach is a major issue for supporting all phases of operations and policy making. When I got back from my Iraq deployment I was assigned to be a subject matter resource for the tech folks who were (and I think still are) trying to design and build the 2nd generation hardware and software tool kit for HTS, as well as for a good chunk of Civil Affairs. These people are really quality folks, but the problem is in the assumptions that underlie what field and analytical personnel should be doing.

The post 9-11 responses to the inability to get information from where it was residing to where it needed to be, led to several over reactions. Not the least of which were the calls for more data (intel or information) and applied tools to analyze it. Remember, with 9-11, just as with the underwear bomber that Blair is getting grief for, the problem wasn't that the data or information or intel wasn't in a system or database or known to the appropriate folks at one unit, office, section, or agency. Rather it was that it wasn't put into a useable (what I call mentally digestible) format and distributed to the correct decision makers and responders. Now we know that a lot of that was because of interagency infighting, which while better is still not properly fixed. But the response of simply going to a tech solution of vacuuming up everything "knowable" actually makes the problem worse in two ways: 1) too much data to ever be analyzed (my understanding from reading newspaper reports back in 2003-2005 is that we have so few quality linguists that the backlog to do translations just gets bigger and bigger) and 2) so much data that what the more statistically minded of us would refer to as error or noise is creating false positives - this is what my Dad, who taught stats for a long time, would refer to as "if you throw enough shit against the wall, some of it will stick in a recognizable pattern!" problem.

These two data issues are compounded in two further ways. The first is through the belief that there is a magic technical process that will take the data and turn it from its base components into intelligence gold - essentially Intelligence Alchemy. This includes over reliance on social network analysis, link analysis (basically the graphical display of the nodes and links in a network or SNA light), geo-spatial mapping, agent based modeling, etc. All of these analytical methods, and a number of others, are useful tools, but they need to be treated as just that tools; they and their output are aids to assist in understanding, not the understanding itself. There is a tremendous difference between being hired by a corporation or organization to map and analyze its social networks, where one is given the official org chart and all personnel are told to speak candidly about who they interact with and rolling off of a FOB or COP or out of an embassy somewhere, talking to elites, notables, non-elites, etc and trying to map a network and draw meaningful conclusions about its dynamics. In the first case the data has high validity - it is known to be good and accurate. In the second case, even when one has confidence in their subjects or sources or informants, it is much, much harder to validate the data that is being captured. As such any SNA or link analysis built off of data collected in the field in informal circumstances has to be seriously caveated because of questions of how good the data is (some will likely be excellent, some good, and some horrible).

The second follow on problem is what happens with the analytical product, let alone the data? MG Flynn and two co-authors issued a report about this issue for Afghan Operations back in January of this year. I did a response/reaction to it for Tom Ricks, which can be found here:
http://ricks.foreignpolicy.com/posts/2010/01/18/the_flynn_report_v_how_to_feed_the_beast
The bottom line, so you all don't have to click through, is that the data is often in a database or repository somewhere. And if its not its often in someone, or multiple someone's heads. The key is to get it out, get it properly archived, then get it into an easily digestible and distributable format, and out to the appropriate decision makers. The solution to the problem of decision makers not having the information or data or intel that they need isn't the creation of more databases or levels of analytical teams, rather its a change in the way we handle our data, information, and intelligence. Just as was the case prior to 9-11 and prior to the attempted underwear bombing, there is data, information, and intelligence (even if its not 100% complete) in the various repositories spread out across a variety of teams, units, elements, sections, and agencies. The solution isn't new analytical tools or even more data (better data is a different story...). The solution is creating a dynamic where the information gets to the people who need it to do the analyses, where the analytical output is packaged in an easily digestible way for busy decision makers, and then distributed to them in a timely manner to service the decision making process.

This, of course, runs contrary to the two standing principles in the Intel world: 1) Knowledge is power, so if I know something and you don't I have power and 2) If you needed to know it, you would already. Until these dynamics are broken down, it doesn't matter who is the DNI or the DCI or the DDIA or what information is sitting on what hard drive. If we want to see past failures not continually repeat, whoever is put in charge has to be given the clear authority to change the dynamic. And that person has to be strong willed and independent enough to take that power and make sure that our decision makers are told what they need to know, not what they want to know or even what they want to hear.

Russ Wagenfeld

Hi Pat,
I too hold Jeff in very high regard. I think I met Clapper once before I decamped for Europe. I did hear much grousing about his reorganization from former colleagues. Being only a tank expert hardly qualifies one to estimate a nations capabilities or intention. Working in such an environment would have given new meaning to the phrase "a bad day of fishing is better than a good day in the office. As I recall, Tony Cordesman and the Net Assessment tried to force DIA into a bean counting outfit as well.
Regards,
Russ

William R. Cumming

For the last 40 years almost every President, leaders in the Executive Branchy and Congress have sought to fix all administrative problems by reorganization. This is the result of ego, hubris, and ignorance. It almost never "fixes" any administrative or regulatory failure but it certainly allows those doing the design to have a fair chance to garner the spoils of the reorganization.

The Twisted Genius

Castellio, your three lines have elegantly captured a critical fault that is widespread across the Intelligence Community. I'm sure I'll have the opportunity to use your wisdom in a professional setting. I will, of course, cite the source of this gem.

Herb

I was at NGIC thru 1996. We had interagency groups specializing in the processes by which the Soviets developed and deployed new weapons. We knew that a program to develop a weapon would not be a threat until the weapon could be fielded. If a similar country based analysis had been done on Iraq, we could have known that Iraqi WMD programs were years a way from becoming an actual threat.

Peter

Adam L Silverman:

Your view into this problem, using machine data, to garner useful data to act upon, needs to be dealt with rapidly. I hope the person chosen for the DNI, who understands the dynamics, emerges with enough trim tab effect to break the tidal wave of digital data into a tidal rhythm.

One of the deeper thinkers on the phenomenon of human computer interaction is Jaron Lanier, he authored You Are Not a Gadget release last January. Jaron hits upon many themes about how software design drives problems that you are describing.

I am currently reading and digesting his work.

Patrick Lang

All

IMO there are no machine analysis answers to understanding human behavior. "Hari Seldon" will only appear in the time of the 1st Galactic Empire. The US Army tried to eliminate human judgment from the tactical intelligence analysis problem The result proved a total failure in these 3rd World Wars. Now, they have hired contractors who have "re-discovered" human thought processes to teach enlisted men and warrant officers to think for themselves. pl

The Twisted Genius

Adam Silverman addresses two main problems in our current intelligence process. The first deals with the difficulty of conducting good anthropological research and presenting this research in an understandable and useable form. The artistry and craftmanship necessary to gain the "information necessary to understand the socio-cultural context, socio-cultural dynamic, and socio-cultural location of the white layer elements" in not something learned in a three day seminar. The lack of sufficient linguistic skills is, by itself, a daunting hurdle to overcome. Having sufficient time to conduct this research will always be a problem. I doubt if Margret Meade could walk into an Afghan village and fully understand the culture if only given a few days to do so. Adam Silverman's work with the HTT system has been a huge step in addressing this shortcoming. It is unfortunate and extremely shortsighted that the Congress does not not want to expand this program even though the Army has said it wants it. Expanding the FAO program and expanding SF will also help in putting more keen observers of culture in the field.

The second problem is deeper and harder to fix. The current intelligence and decision making systems (on the whole) do not value the artistry and craftmanship necessary to reach real cultural understanding. As our host has pointed out, the artists and craftsman are not the ones making the management decisions. That is the realm of the impatient, overly ambitious self-servers. Too many of them can't think deeper than a PowerPoint slide. They may still be literate, but they've apparently lost the knowledge of how to turn a page. They would rather have an automated analytical machine that would spit out a numerical answer, even if the code for this machine was written in pure snakeoil. They also seem to value information in direct proportion to its security classification. MG Flynn and I'm sure others are recognizing these problems in our own national security culture, but as Margret Meade once said it would be easier to create an entirely new culture than to predictively change an existing culture.

Adam L Silverman

Twisted Genius: Thank you for the kind remarks. Unfortunately more and better HTTs, let alone sufficient FAOs, Civil Affairs, SF, and a real civilian agents cadre available (and my view is that HTS needs to eventually become the latter. The TRADOC G2 deserves great credit in taking a risk and giving the program a home for its funding line, but HTS and the HTTs aren't intended to be intel units. They need to be under the Ops and/or Civil Military Affairs and in my view should be moved to the Joint level and out of TRADOC. Once that happens then every COCOM can be sourced with a group of rear Deck personnel as part of the COCOM Commander's Initiative Group increasing effectiveness and integration.) isn't enough to fix the problems. These assets need to be active and utilized from beginning to end, as well as in Phase 0 through Phase 5 Operations. At the point where HTS was created, which is something some of its harshest critics never seem to think about, it was created and made operational to assist in triage - in fixing what had been done wrong and/or gone wrong in Afghanistan and Iraq. In those situations there is only so much an HTT or a CA company or an ODA or a PRT can do!

Part of this comes down to the fact that 8 years into OEF in 5 years into OIF we still don't seem to have a common operating picture, let alone concept of operations, for exactly how to best utilize these various and important assets. Various elements are working on that, for instance TRADOC has a very interesting initiative that they're spooling up that will help, but how far that will penetrate is a question that won't be answered for a while. Additionally, what these assets can do can be leveraged for things outside of COIN or stability operations. The core of human terrain work - the socio-cultural research, analysis, and mapping which ties together the overlap of the three socio-cultural domains that you allude to above and that I've written and lectured about - can and should be leveraged to assist with disaster management and emergency response in places like Haiti, for dealing with public health issues such as H1N1, even for facilitating conflict resolution. It doesn't all have to be done by one organization, and certainly HTS as currently envisioned and configured isn't going to be able to do it - its not designed to, but that doesn't mean that FAOs can't be trained on it or first responders or community policing specialists or conflict/crisis negotiators. This too is outside the scope of anything currently being done, but the base concepts are sound, and the adaptation is something I've been working on for months.

Patrick Lang

All

Walrus was right. we can not afford any of this imperial nonsense. The COIN fad will die on that shoal. Failure in the COIN campaign will speed the process in the context of Obama's political needs. pl

David Habakkuk

Adam Silverman,

There is a tremendous difference between being hired by a corporation or organization to map and analyze its social networks, where one is given the official org chart and all personnel are told to speak candidly about who they interact with and rolling off of a FOB or COP or out of an embassy somewhere, talking to elites, notables, non-elites, etc and trying to map a network and draw meaningful conclusions about its dynamics.

Without doubt there is a very great difference. But it may be not quite so radical as you suggest.

I and my wife had quite an extensive familiarity with British broadcasting institutions -- and also quite a diverse one, as I came in to television current affairs from journalism, and she began as a production assistant in drama, and between us we worked over the years in a variety of roles in a variety of organisations.

We have watched the decline of the industry -- and although it is only a small element in this sorry story, the influence of outside consultants on the BBC is by no means a negligible one.

Part of the problem, I came to suspect, has to do quite precisely with the fact that such consultants need to operate like (good) anthropologists studying tribes -- and do not know how to do so.

Like a lot of television current affairs producers, their whole education and training conditions them to put direct questions to people of similar education and training, analyse the results, and present them in forms comprehensible to people like themselves.

It is people in this mould who tend to dominate the upper echelons of an organisation like the BBC, and they are certainly capable of giving you a fluent account of how they think the organisation works.

But if you look at how, for example, the kind of programmes which were mentioned appreciatively in a recent thread -- such as the BBC series 'Dad's Army', or the ITV series 'Rumpole of the Bailey' -- used to be made, the whole process depended upon dense networks of highly skilled people, whose training and experience was in doing things, not analysing them.

Behind the actors, directors and producers whose names got noticed were the production assistants, production managers, costume designers, set designers, lighting cameramen, film editors, etc etc. And such people had an enormous amount of accumulated expertise, not just about their own specific areas of competence, but about the whole art of making television drama.

Actually to understand how programmes were made -- and make constructive suggestions for how they could be better made at less cost -- you would have needed to talk to such people.

If however, as an outsider equipped with an 'official org chart', you had come in and asked them direct questions requiring them to 'speak candidly about whom they interact with' you would have been liable to end up with a very misleading impression of what went on.

This is not because they were stupid, or did not analyse what they did. Such people, commonly, used to inhabit very dense networks, whose members had worked together for years, and spent much of their leisure time together: particularly as their partners and friends were commonly people met at or through work.

Accordingly, a great deal of their time, both on and off work, was spent talking about what they did. But precisely what they did not have any occasion to talk about were the basics of the activity in which they were engaged -- because to everyone involved in these networks these were second nature, and taken for granted.

So their normal style of talking about what they did was highly context-dependent, and the problem for an outside analyst was to recover the context of what was taken for granted so he or she could understand what was being said, and done: which is not perhaps altogether different from the problem of someone trying to work out how Iraqi or Afghan tribes operate.

A further point is that the fact that people had been told to talk candidly with an outside observer by those running the organisation would by no means have guaranteed that they did so -- indeed, in many circumstances it would have produced a horse laugh.

How far people would have opened up with such an observer would have depended on the kind of organisation for which they worked, and their position in it, but there were certainly plenty of situations where your prospects of being told the truth would probably have been markedly worse than if you were researching an Iraqi or Afghan tribe.

Up until the Eighties, whether in the BBC or ITV, programme-makers worked in large bureaucratic organisations with what in many ways was excessive security of employment.

Certainly, people who were itching to get up the greasy pole often were not only prepared not to say critical things about their superiors, but even to think them. But the combination of job security and the fact that those superiors actually needed to have subordinates who could do make programmes meant that people who did what they did well and had no ambition to rise could afford to be candid.

The destruction of those old bureaucratic structures, as a result of the introduction of independent production and short contracts, whatever its other advantages, has tended to militate against candour.

In television at least, it is liable to be highly imprudent for an outside contractor to be candid about how a contracting system actually works, either to those running it, or to anyone working for them. One is dealing with patron/client relationships, in which fawning servility is commonly an eminently rational strategy: it all depends on the quality of the patron.

Likewise, people on short contracts are commonly in terror of losing their jobs -- and with very good reason. And all too easily, they can lose the habit of being candid with anyone, at which point the whole organisation can become dysfunctional.

My suspicion would be that similar difficulties face outsiders trying to understand how all kinds of organisations actually work -- as distinct from how they are supposed to work. And I would certainly expect this to be the case with military organisations.

Tapping the immense reserves of knowledge of experienced sergeants, for example, would seem likely to involve very real difficulties for an outsider without military background. (But then, these probably do have practice in explaining the basics to people who know very little, as they have to deal with wet-behind-the-ears young officers.)

As to how relationships with contractors actually work, I suspect that -- in military matters as in television -- those who are in a position to know have every reason not to be candid, and those who are in a position to be candid all too often do not really understand what is going on.

Adam L Silverman

Mr. Habakkuk: Thanks for the thoughtful and useful information. As I've never actually tried to do SNA on a "formal" organization, all I have to go on is what I've either read or been told by people who do this sort of thing. I think your example just further demonstrates the difficulty of trying to leverage some of these very interesting, and in many cases very powerful, technologically dependent methodologies. One of the most important things for doing the type of work I did was that if you can't do it with a pen, a notebook, and maybe a digital camera, then you're doing it wrong.

The comments to this entry are closed.

My Photo

August 2020

Sun Mon Tue Wed Thu Fri Sat
            1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          
Blog powered by Typepad