Unfinished paper by Gunter Heim, December 2000:
The neural company
Abstract
Pointing out similarities between various forms of human society
and human bodies is nothing new. From John Hobbes Leviathan of
1651 down to the still vague metaphor of the global
brain there have been more or less serious minded attempts
at establishing functional similarities between highly
individualized organisms such as humans and large conglomerates
of individuals like human societies. This paper will be in the
same strain. It will argue that large enough organizations do
already resemble the information processing of human brains in
some salient aspects and that they will increasingly do so as
many groupware features, organizational features and workflow
engines emulate isolated aspects of neural intelligence.
A foreword to the scientifically minded reader
It is not the aim of this paper to produce a fully elaborated
testable theory. This paper aims at providing some inspirational
stimuli to those engaged in shaping the way large groups of
people will do knowledge work in the future.
In style, it will try to persuade with a profusion of detailed
examples and metaphors rather than convince by means of
inescapable logical reasoning.
And although we as the authors are quite convinced of the
validity of many of the propositions made in this paper, the only
credit we have a right to ask for is the benefit of the doubt.
So, as a reader, you should adapt your mode of reading
accordingly. Suppress your critical capacities, your rational
impulse to reason and your demands for authoritative evidence.
Instead, lend us, for a while, what may best be called your
intellectual generosity.
We do not claim to be first in formulating the ideas presented
here. In fact, we would be surprised had not someone else laid
them out already. For the time is ripe for them. But we have
certainly not stolen anything from sources without due reference.
References have been included where they serve to illustrate our
suggestions. They have been omitted where they would have given
the wrong impression of scientific ambitions.
The central idea: synaptic weighing of information
Although the brains working cannot be reduced to any one
principle alone, the connecting points between different nerve
cells, called synapses, seem to play a major role in the
brains operating system. The basic information processing
units of the brain are single nerve cells, called neurons. Each
neuron has one outgoing nerve fibre called the axon. This may
bifurcate a number of times. It is along the axon alone that the
neuron can send signals to other neurons. Another structure
stretching out from the neuron`s body is called dendrites.
Dendrites usually have a tree-like shape and this is where
ingoing signals from the axons of other neurons are received
through. Now, each signal that passes from the axon of one neuron
to the dendrite of another neuron has to cross a physically
visible structure called the synapse. The likelihood, however,
with which an ingoing signal will be transmitted across the
synapse is influenced by the synapses momentary internal
structure. The synapses readiness to pass on signals is called
synaptic conductivity. Inside the neurons body, called the soma,
the ingoing signals are processed in a way unknown. But most
scientists would agree that the outgoing signal of a neuron is
significantly influenced by its input.
If you appreciate the fact that an ordinary neuron is connected
to about 10.000 other neurons, this fact can be formulated as
follows: In any given situation, a neurons output is at
least partly determined by the synaptic conductivity of its
synapses. Weighing ingoing signals according to the current
situation seems to be one key to intelligent information
processing. We must, therefore, suspect some intelligent
mechanism regulating synaptic conductivity.
The strength of this principle is proven by the widespread
success of neural networks. These computer programmes
exploit the mechanism of altering synaptic conductivity. By the
use of more or less sophisticated learning algorithms, synaptic
conductivity within a network of neurons is so manipulated that
the networks overall output increases in quality during a
session of learning sets.
Let us now turn to human information workers within an ordinary
organization like a company and see the similarities.
The picture above shows quite an ordinary working place in the
control centre of an industrial plant. Via the monitors and
through the telephones and notes lying about on the desk, the
operator is linked to vast ressources of information. Long lists
of telephone numbers, shift plans, process data showing
temperatures inside tanks, rotational speeds of enginges,
electrical currents and voltages, CH4-contents of analyzed coal,
maps of the plants premises, plans of electrical
intallations, flow charts etc.
In case of an accident, the speed and quality of the
operators decisions and instructions given to other people
on the spot will largely depend on how fast he can
access relevant information. This is true of any situation where
human attention and time for information retrieval are limited
ressources.
Technologies for user- and case-sensitive data-presentation are
one key-focus of groupware and knowledge-management systems being
developed at present. They all aim, more or less, at sensibly
modifying the attention a certain piece of information can
attract from a user in a given situation. To revert to the model
of neurons, they modify synaptic conductivity between providers
and potential recipients of information. If you place an
important note right on the desk of a colleague instead of
sending it through the company snail-mail system you are
increasing the synaptic conductivity between the notes
content and your colleagues attention. And the same thing happens
when you are surfing the internet using a commercially oriented
browser. By constantly monitoring your interests, the browser may
be able to suggest links, banners and tickers that are most
relevant to you in such a way as to increase the likelihood that
the linked contents will pass through the monitor or loudspeaker
straight into your brain. Even the simple rules governing the
right to speak between different people in a company fall in this
case: the higher up on the hierachical ladder you are, the
greater is your privilege to interrupt lowers in their
conversation to put in what you have to say. This, obviously,
increases the likelihood with which superiors can get their
messages across. And the true value of a good secretariat is to
filter and weigh ingoing information in such a way, as to best
suit the needs of the superior. For an unimportant and unpopular
colleague, the synaptic conductivity of the secretariat may be
near nil whereas an important customer may gain direct access to
the boss in the same circumstances.
So here we have one (no more!) basic mechanism of neural
intelligence to build on: the weighing of different sources of
information. And although the weighing of information inputs can
occur in many ways without the aid of computers, this paper will
concentrate on digitalized communication and knowledge
management. Thus, the classical model of communication including
sender, recipient, content and medium of information transmission
is slightly modified to bring it nearer to the neuron model of
real brains:
figure: model of organizational neuron
But the broader perspective of this paper cannot be attained
unless we allow statistics to enter into the concept.
A case study: a large mining company
Small companies are not large enough to be treated statistically
in the sense we want to. Average nations, however, are too big to
be treated as an individual with some sort of unifying groupware
making any sense. So something inbetween like a large company
with at least some few thousand employees will do better. A
differentiated organizational structure, too, is needed. And the
complexity of the whole thing should be well beyond the grasp of
even the most intelligent human individual.
So, for the sake of illustration, imagine a large mining company
with many divisional branches spread across the globe.
Open-pit copper mines on the southern hemisphere, newly developed
underground coal mines in the former Soviet Union, the
exploration and exploitation of off-shore gas and oil fields and
the mining of complex polymetallic ore bodies at various
locations are only some of the companys key-engagements.
Besides, the company runs two global selling organizations. It is
doing active research into milling, coking, geophysics and
metallurgy. And the company entertains a worldwide network of
lobbyism, public relations and widely appreciated sponsoring.
Through some of its products, its name is explicitly known to
most consumers of even the most basic constituents of the western
life-style.
There is a headquarter somewhere in Europe, but many activities
are managed on the spot without much interference from HQ.
However, many issues need central coordination:
By taking a closer look at the last example, we will try to
identify some helpful mechanism to reduce the complexity of the
issue in order to reach an optimum answer. These, of course, will
correspond to some phenomena in the human brain, as will be
shown.
Digression: the internet mode of thinking
But let us first swerve a little from the main course of this
essay. We will pose you a question and ask you to turn your
attention inside-out and look into yourself: Monitor your
thoughts for a while before reading on. Here is the question:
Where will your next shopping tour take you to?
What happened? When we do experiments of this kind,
there usually crop up a number of rather diffuse images, ideas or
words competing for full attention against each other for a
while. In the case of the shopping tour these may be, for
example: a vage image of the contents of the refrigerator to
suggest what is needed, elusive connotations of various
supermarkets or small shops, snaphshot views of congested streets
and some fleeting sensory reminiscenses of dishes that are on the
cooking plan for the coming days.
Sooner or later, one of the images will dominate over the others.
As it gains in clarity, new images come within the ken of your
attention. The image of a certain supermarket may suggest the
image of a friend of yours you met there last week and who
suggested a barbecue evening. From here, your thoughts may wander
off to still other images and the chances are high, that you may
lose your initial question of of sight.
Also bear in mind,that as each image brings with it a number of
further associations, the direction of your thinking and the
quality of its result heavily depend on the strength of those
associationsl
This mode of thinking corresponds in some striking ways to a
typical internet session. Suppose you were consulting the
internet to make a list of possible holiday locations for a local
youth group. You`d probably start by entering some search words
like holiday, children, youth. As a result you are
presented with a number of hits. You will open on of them which
brings a new load of facts and further links to your attention.
The quality of your session will heavily depend on how well links
correspond to the content of the information sources thus
addressed. And a good internet surfer will soon develop some sort
of instinct that gives him some clue as to the content of
internet sites. However, as in thinking, there is always a danger
of getting lost by an uncontrolled following of links and
associations.
We do not wish to overstrain the likeness between human thinking
and knowledge work at a PC yet. We must, however, lay
considerable stress on the similarities between the flow of
ideas, images, words i. e. notions as existing in human brains on
the one hand and information ressources in information networks
such as the internet or companies and information working in both
cases seems to involve hopping from one image to the next. We
can, therefore, expect the nature of these notions to be of high
importance.
Organizational reports as standardized ideas
Human thinking would not work the way it does without the
existence of some reliably addressable informations ressources.
Admittedly, each one of us may have completely different contents
in his working space of consiousness when presented with the word
channel. The important thing is, that within one
brain, the information addressed by such a word is constant
enough to play a defined role in complex information processes.
And the same is increasingly becoming true of organizational
sources of information. Financial data is already made available
in a highly standardized form to meet the rigid demands of
professional accountancy. Software systems offer specified
reports. But controlling, monitoring and other
technical data are also increasingly shaped and moulded into
reliable reports. The driving forces behind this tendency are
manifold. The wish to employ standardized software, the demands
for transparency and clear definitions of processes as laid down
in various norms such as ISO 9000, the growing need to provide
electronic data in an interchangeable format and, last but not
least, the growth of intra- and internet structures where the
exploding amount of information can only be met with the creation
of reliable methods of retrieval.
In a number of years, the by far largest proportion of
organizational knowledge will be stored in computer systems.
Knowledge workers will rely heavily on standardized sources of
information.
So, let us now return to our internationally operating mining
company. Suppose it is your task to make well founded suggestions
for future mining engagement anywhere on the globe. And let us
further take it for granted that most of the information you may
need is available electronically.
Starting from scratch, you load a work-flow sheet (just another
report) labelled new mining engagements onto your
screen and switch into the internet mode of knowledge
work. The work-flow is a very good guideline and offers many
sources of information that you should consult to get a first
impression of viable potentials for further elaboration:
Title | Type | Description |
mineral markets of the world | intranet site | long term prognostications of mineral markets |
mineral deposits | internet site | maps, geology, mineralogy and geography |
world mining report | internet site | legal, financial and entrepeneurial issues worldwide |
strategic finances | intranet site | the company`s long term financial planning |
comany mining expertise | intranet site | mining techniques currently in use |
company R & D potentials | intranet site | current, past and planned R & D activities |
competitors review | intranet site | portal of the company`s secret service |
sleeping tigers of the world | internet site | somewhat exotic futurology on politcs and economis |
global miners company newsgroup | discussion forum | company issues of international relevance |
Scanning over the titles and the short descriptions of the
ressources can be compared to evaluating the first images and
words that come to ones mind when working out the tomorrows
shopping tour. The deeper contents of the ressources are yet
vague and you have to opt for one of them. In the case of the
mining company, you may chose to start with a general overview of
the mineral deposits. You open the site and you are then
confronted with maps to click on and with further links to follow
up. Depending on your disposition and circumstances, you may then
stick to your work-flow or you may plunge into the fascinating
world of geology and mineralogy of the world...
A report in this sense need not be identical with a physically
compact storage of information. It can better be described as
some standardized access to a more or less constant themes and
topics. In practise, it is more like a personalized portal to
some specific topic than a static list of facts.
However, don`t forget that your time is limited. Your telephone
may ring any time to call you to some meeting, or your boss may
give you some other most urgent work, or you may get ill one of
the coming days. So the order of working through the knowledge
ressources is most important. Missing some hint on a
competitors activities, for instance, may invalidate
completely the the result of your work.
So once again we are confronted with the importance of priorising
links. This leads us back to the concept of synaptic weighing and
it is left up to the reader to acknowledge the importance of this
for any individual worker by analysing his own workaday office
life.
In consequence, the tendency towards structuring knowledge in
organizations will be flanked by a growing focus on producing
various forms of reports for various situations and various
people. The techniques for this are already being developed,
albeit against different backgrounds. And they will only have to
be merged into what my fittingly be called synaptic
browsers to emulate the internet mode of thinking that
takes place in our mental life.
A common model of organizational and mental
knowledge work
Having thus sufficiently established some functionial
similarities between organizational reports, intenet-sites,
documents etc. on the one hand and mental entities such as
images, ideas, notions on the other hand we now want to set up a
model that can account for many phenomena related to knowledge
work in both fields of experience.
The model is restricted to knowledge work only. All knowledge,
information and data is addressed as a report.
Anything that can change reports on the basis of other reports in
some predictable way is called a knowledge worker. All knowledge
work takes place in the form of tasks. Tasks are so
defined as to update a specific report on the basis of certain
initial input knowledge processed by clearly defined knowledge
workers. Various mechanisms subsumed under the titels of
task-manager and emotion-manager
constantly reassess the allocation of ressources and the weights
of links within reports as to optimize the whole systems
performance.
Reports must have an unequivocal title corresponding to a URL or
its alias in the internet or a symbol within human mentality. It
can contain some information like images, sounds etc. It also
contains references to other reports.
In organizations knowledge workers can be identified with human
individuals or computer programmes doing a certain job. They must
only fulfil one criteria: If they are fed with certain pieces of
information, their output must at least show some statistical
dependance on the input.
Tasks are identified by the report they are meant to update. All
information work finally results in the update of a report. By
rallying together an initial set of reports and knowledge
workers, a task is activated.
The task manager is like some background routine continously
assessing the right distribution of knowledge workers to certain
tasks. It stand for a mechanism that both monitors and asseses as
well redistributes the the use of these ressources.
Using the metaphor of computer data processing the elements of
our model can be characterized in the following way:
reports = data
knowledge workers = programmes
task = working space of a computer
task manager = operating system continuously managing the system
ressources
--------------------
--------------------
This piece of work was left unfinished in December 2000 due to a
lack of time to further elaborate the model. The chief difficulty
of the model seems to be as to characterize a neuron. Would it be
a single worker? A group of people? A virtual team? A project?
All of these?