SPEECH TO AFIO CONVENTION: "THE STATE OF ANALYSIS"
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP90T00435R000100100014-6
Release Decision:
RIFPUB
Original Classification:
U
Document Page Count:
15
Document Creation Date:
December 27, 2016
Document Release Date:
May 15, 2013
Sequence Number:
14
Case Number:
Publication Date:
October 14, 1988
Content Type:
MISC
File:
Attachment | Size |
---|---|
![]() | 587.11 KB |
Body:
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
*5 UNCLASSIFIED
SPEECH TO AFIO CONVENTION: "THE STATE OF ANALYSIS"
14 OCTOBER 1988
It is very constructive that this convention is focused on the
state of analysis in our intelligence community. And I am
pleased to be able to make a presentation to you on this
important topic, one that has received a lot of attention in
the academic world and within the family, so to speak, but too
little from the general public.
Let me give you my bottom line right off: The state of
analysis is, on the whole, very healthy. It has improved in
recent years, and it promises to get even better in the
future. There are problems, some very old, some quite new; but
they can be managed.
I reach this judgment neither casually nor as some kind of
cheer leader. I have spent the most of a 25 year professional
life in and around intelligence analysis, as an analyst and a
manager of analysts, as a critic and assessor of the \
performance of analytic organizations, as a National
Intelligence Officer and, now, as a manager of a body of such
officers, as a consumer of intelligence at the National
Security Council, and finally as a member of a think tank and
even as a consultant.
UNCLASSIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
So I have pondered this animal from many vantage points over a
good number of years in which much has changed. The views I
shall offer you have been screened for propriety and security
by colleagues. But I want to make clear that these views are
my own, and should not be attributed to the DCI or other
Community managers.
The bulk of my remarks will dwell on the two factors which are
dominating the development of intelligence analysis: people
and advanced data processing. But first I want to offer some
more philosophical thoughts on the role and nature of analysis.
The Role and Nature of Analysis
Most students of intelligence know analysis as one of the links
in the chain of activities that form the entire intelligence
process, from defining requirements, to collection, to
processing, analysis, production, impact on policy, and then
--back to defining the requirements that guide collection. Yet
analysis is as much a glue that holds the whole process
together as it is a specific link: Without analysis at every
step -- namely, thoughtful and systematic assessment of the
meaning of data for answering the questions of customers -- no
step of the intelligence process can work. Exotic and
sensitive collection is what makes official intelligence
unique, that is, what distinguishes it from scholarship or
UNCLAAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
journalism. But analysis is what makes it really intelligence,
that is, focused on real problems and intelligible to a
customer.
Standing against this ubiquitous necessity for analysis is one
of the less healthy phenomena of our business, namely the
barrier between the two cultures of analysis and operations. I
sense a positive trend toward lowering this barrier, but it is
painfully slow.
Although it started learning from others, primarily the
British, the United States has developed its own intelligence
style. I haven't done a rigorous study of this, but I suspect
our intelligence style puts more emphasis on analysis than do
other advanced intelligence systems. I have the impression
that we put a higher proportion of our overall intelligence
resources into processing and analysis. We also, I believe,
push intelligence analysis deeper into what in other cultures
is regarded as the province of the statesman. Thus, American
intelligence analysts spend a lot of time trying to discover
the answers to what I call mysteries, as opposed to secrets, a
distinction I find extremely useful but which is not often
consciously made by analysts. A secret is the answer to a
question which somebody knows or which can be found in some
reservoir of information; it is a collectible. A mystery, on
the other hand, can only be resolved by God or, if you wish, by
history. A hard fact like the radar cross section of a new
UNCLAAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
Soviet aircraft, or even a soft fact like the real intentions
of Gorbachev about economic reform...these are secrets.
Whither Iran after Khomeini, on the other hand, or the impact
of AIDS on underdeveloped countries, these are mysteries.
Cracking secrets can sometimes help resolve mysteries, but in
the main we must rely on the tools of science to provide
inevitably uncertain answers. It is fundamentally America's
respect for science that has made analysis so important a part
of our intelligence culture, And why analysts have to spend
time on both secrets and mysteries. Both add zest, drama, and
gravity to our work, But the combination over a growing
panorama of topics is what makes it impossible to reduce the
business of analysis to some rules or codes comparable, let us
say, to tradecraft or orbital mechanics in the collection
disciplines. The only rules which apply across the board are
reason, logic, and intellectual honesty.
The one issue that has drawn intermittent public attention to
intelligence analysis is that of honesty, or "objectivity",
usually in connection with some political controversy. Our
national security requires objective intelligence analysis;
this is truly the measure of our work. But outside critics as
well as inside participants must understand in a mature way the
conditions in which the measure must be met. Rarely in our
business are there provably true answers to the really big
questions, We cannot do our work without getting close to the
policymaker and the policy process. Providing intelligence
4
UNCLASSIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
support to policy is like being a parson in a bordello. There
are temptations and risks; but without running them, no sinners
can be saved. The more serious dangers arise when intelligence
analysis is too distant from policy, and corrective views don't
even have a chance to form.
In this context, charges that policymakers or intelligence
managers dictate analytical conclusions and thereby, as the
expression has it, "politicize' intelligence analysis are
usually overdrawn and often just plain wrong. During the
1970s, I spent many years fighting city hall on a major topic.
In my case it was assessing the Soviet appetite for strategic
nuclear superiority. My views did not prevail until the
evidence became so convincing as to make it hardly an
intelligence problem any longer. Now even Soviet spokesmen are
admitting that the hawkish views I held were right all along,
At no time, however, did I fail to get my views heard or even
to get them, appropriately identified, to policymakers, Nor
was I shunned on other topics because my views were rejected on
this one.
In my opinion, the most important threats to objectivity come
not from the chain of command or the policymaker. Rather they
come from fashion and general intellectual conventions, on one
hand, and from smug attachment by individuals to their own
opinions, on the other. The antidotes to these ills are always
at hand, curiosity and openmindedness, the liberal use of which
is what makes intelligence analysis fun as well as productive.
UNCLAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSII-ILD
The Human Cadre
When I returned to CIA headquarters in the mid-1980s after a
six-year absence in other government and private employment, I
was struck by the omnipresence of computer terminals and the
large number of very bright, very new, and very young faces.
These two impressions arise from the most dynamic and
influential trends in intelligence analysis: new and better
people, and new and better ways of handling data.
The observations I want to offer about both these topics arise
primarily from CIA's experience because the Agency's statistics
were most accessible and my personal impressions have been
formed there. But many of the trends I'll mention are also
evident at DIA and probably other Community components. Where
this is not the case, perhaps other speakers will offer
corrective points. Out of security considerations, I shall not
present any hard quantitative data.
With the growth of intelligence budgets in the early 1980s, the
analytical complement of the Intelligence Community grew
substantially. Other activities also grew so the size of
analytic components, such as the CIA's Directorate of
Intelligence, stayed about the same relative to other
components and the proportion of analysts versus other types of
personnel also stayed about the same. But the total number of
analysts grew significantly.
UNCLASSIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
This new cadre of recruits showed some qualitative improvements
over earlier cohorts.
Their test scores were higher than their counterparts' in
the 1970s on analytical aptitudes, interpersonal skills,
and work attitudes.
Larger percentages of the newcomers had advanced degrees
and other specialized academic preparation.
More had had overseas travel experience through tourism or
study, although interestingly, there was a statistically
significant drop in the foreign language capabilties of the
newcomers. This appears to have resulted from the
unfortunate relaxation of language requirements in our
universities.
On the whole, however, we saw both a quantitative and a
qualitative boost to the analytical cadre. Why did this
happen, beyond mere budget growth? Several factors were
involved. The economic slump of the early 1980s made
government employment more attractive to university graduates
at that time; it also noticeably reduced the attrition rate
among employees. In addition, the Agency adopted special
programs to attract quality recruits, such as the Graduate
Fellows Program, and focused recruiting efforts on specialities
in high demand, such as engineers and economists.
UNCLAAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
As a result of this, the analytical workforce at CIA is
unusually young, half being under 35 years old, By and large
they came to the Agency without prior analytic experience,
This has made for an inexperienced workforce, but one which, as
it acquires training and experience, offers great promise for
the future, Because of the all-volunteer military, a much
higher proportion of these recruits than in the past has no
experience with military service, a problem to be compensated
for in some areas where prior service gave a head start to
military analysts, In DIA, however, the median age is about a
decade older and more military experience is represented in the
analytical cadre; so, from a community point of view, this
problem is not acute,
Our managers and personnel people have noted some interesting
attitude shifts among these newcomers. They are more
performance oriented than their predecessors, but also more
advancement oriented, and less committed to a given job or
organization. In a word -- if I can use it non-pejoratively --
we too have acquired our yuppies, This, it seems to me, adds
to the challenge of management in the years ahead.
In the past, the typical career development pattern for
analysts was to get a fairly large number of different
journeyman-level assignements clustered in a general
discipline, like economics or military analysis, or perhaps in
UNCLASSIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
varied fields, and then to move into management. Herein has
lain a management dilemma as old as our various services. How
do you strike the right balance between the generalist and the
specialist? Generalists are, by definition, the most broadly
capable people, especially for management. So your more able
people want to be or are encouraged to be generalists. But the
generalist is not the one who knows an obscure but suddenly
vital topic inside and out; he's not a real expert in
anything, Intelligence needs real experts. Further, how do
you strike the balance between management and advanced analytic
achievement as paths to promotion? Much talked about in the
past, both CIA and DIA have recently made real progress in
setting aside supergrade positions for senior analysts as a
means of easing this dilemma.
To make the most of the new influx of talented young people we
got in the early 1980s, I think Community managers are going to
have to work harder and show more imagination in addressing
these problems in the future. Budget prospects bode for
clogged headroom, thus limited advancement potential; the
relative financial attractiveness of federal employment in
general seems unlikely to grow. We could face a situation
where the best and brightest of our new recruits get
pessimistic about their prospects just when they are getting
the experience they need to perform at peak, but before they
are really shackled by the golden handcuffs of retirement
benefits. If we don't rise to this challenge, we could lose
UNCLAAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
the promise inherent in the new cohorts, If we do, however,
maturation of this new generation of analysts could make it the
best ever in production and management,
Advanced Data Processing and Electronic Tools
Let me now turn to the other dynamic factor on the analysis
scene: Advanced Data Processing, I need to warn you that I am
a real amateur in this field, a friendly and curious user, by
no means an expert. But this makes it all the more obvious to
me that we are witnessing a revolution in intelligence analysis.
Now, US intelligence grew up with and on computers. Indeed we
have been pathfinders and inventors of all manner of advances
in the service of needs which were very demanding and put us on
the technological frontier of computer science. On the whole,
I believe, the needs that drove our innovation in ADP over most
of the past 40 years were, while technologically very
demanding, also highly specialized, e.g., in cryptography, or
weapons analysis. Innovation of this sort continues at what
seems to be an ever accelerating pace: worldwide crop and
agriculture analysis, for example; or assembly of large
specialized data bases on such new topics as terrorism,
narcotics, or AIDS. All these topics demand new computer-based
tools.
UNCLAPIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
The subject matter of importance to US intelligence is
continuing to expand, constantly adding new topics while the
old ones stay on the agenda. Specialized, large-scale ADP
tools for analysis are helping cope with the expanding work
load.
I am fascinated, however, by another phenomenon, no less
remarkable for the fact that it is becoming commonplace: The
role of advanced electronic data handling and messaging in
shaping the day-to-day work of all analysts, no matter what
their specialty or need for special ADP tools. Another word
for it is the electronic office.
This is an area where the Agency has also been a pioneer. This
revolution was just beginning when I was the manager of a
so-called pilot branch for a new electronic mail and filing
system in the mid-1970s. My colleagues and I were able to
watch the unfolding of a crisis of that time faster than our
Operations Center. Now this revolution is in full swing.
Analysts can read electronically a large volume of mail
assigned to them by profiles they design themselves. They can
rummage around at will in the total current data base of
incoming or recent mail to explore ad hoc topics. They can
select out of this blizzard of material the reports they want
to file, quickly assigning them to multiple files. Then they
UNCLAiIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
can rummage back through what they've decided to save and
manipulate it in various ways. They can send a copy to a
friend, with a comment or an extended dissertation, and in due
course get a comment back, all electronically.
This is the electronic mailbox, shoebox, and filedrawer
combined. Of course, there are always problems: Mainframes
crash, my hardware or software won't do exactly what I want,
some specialized data or mailflow isn't on the system, no
keyword search profile will guarantee that a pertinent report
wasn't missed.
But, nevertheless, this stuff is extremely powerful. It allows
the analyst to scan or read his routine mail with much greater
speed than hardcopy allows. It allows for more elegant and
varied storage regimes. It allows for poking speedily into
topics not routinely examined, And, I think most important, it
allows rapid communication with colleagues about data or
reports. Nothing in intelligence analysis is as creative as
the colleague who bursts into your office with a report
shouting "Get a load of this, will you:" Now it can be done
remotely, rapidly, and, in my experience, with no less verbal
impact.
This work environment is now typical; it will soon become
universal. The volume of incoming information has more than
tripled in the last decade and promises to grow even faster in
UNCLAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
the next. Without the capabilities of electronic mail, we
simply could not cope. The impact has already been
considerable even in very soft areas, such as political
analysis, which are usually thought to be unfriendly to
advanced data processing. From the time of Guttenberg the key
to understanding an unfamiliar society has been to read
intelligently. ADP allows the analyst to read more
intelligently and faster, however old fashioned his analytic
methods. It will take a lot of money and management to make
sure our systems keep Up with the demand of information flows
in the future.
I am on the lookout for some developments that will amplify the
already considerable impact of the ADP revolution on
intelligence. Let me mention three that I think important:
I think we are moving toward a situation where the analyst,
perhaps with the help of an ADP specialist at his elbow,
can quickly design a specialized ADP software tool to work
a problem unique to him, and perhaps needed only
temporarily. This will bring the full power of ADP to the
service of the whole family of analysts.
I suspect the pace and efficiency of innovation will
accelerate as people who have grown up with ADP move into
management positions replacing those who now know, but do
not really feel, how important it is.
UNCLAIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
UNCLASSIFIED
Finally, electronic communications and mail make it much
easier for the manager and even the policymaking customer
to plug into the analyst's data environment. It is not
widely appreciated how much raw intelligence reporting is
now available electronically to the policymakers who
constitute the customer community. This should permit new
kinds of communication between analysts and customers.
In a related area, the graphics capability of modern ADP
systems has had a considerable impact on physical production of
final reports, helping to make the production of richly
illustrated and therefore more communicative products easier.
How About the Quality and Imoact of the Product?
You may well now ask: With more and better people and more
powerful capabilities to move and absorb data, are we producing
better analysis?
This is a hard question to answer convincingly. But I am
convinced the answer is yes.
First, the professional standards of inteligence products
have measurably improved in recent years. There is more
interdisciplinary analysis. Clearer distinction is made
between fact and judgment, and better sourcing is offered.
There is more attention to audience needs and more use of
outside experts to review products.
UNCLAt?IFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6
? UNCLASSIFIED
Second, the very fact that US intelligence has managed to
keep up with the information explosion is testimony to an
improved overall performance.
Third, I'll just give you my personal view as one who has
sat on the receiving end, I think the product from all the
major agencies is better than it was a decade ago, The
money and the attention given to US intelligence during the
1980s have brought dividends.
There are going to be problems in the years ahead bound up one
way or another with the austere budgets that we all expect.
But I suspect they will not be the most severe in areas of
analysis, but rather in other areas. The American formula for
organizing intelligence analysis has, I think, proved itself.
A family of agencies of different but overlapping, and hence,
somewhat competing concerns, of differing sizes, and
differently subordinated: This makes for a rich and
informative intelligence product. If the US government
sometimes acts dumb in the world, it's usually for want of a
kind of intelligence other than we are discussing here.
15
UNCLASSIFIED
Declassified and Approved For Release 2013/05/16: CIA-RDP90T00435R000100100014-6