Managing improvement amongst autonomous actors
with OMCA: the case of the Chilean Educational Reform
Martin Schaffernicht S.
Departamento de Informatica de Gestion - Facultad de Ciencias Empresariales
Universidad de Talca
Avenida Lircay s/n - Talca - Chile
Phone: (56 - 71) 200 253
e-mail: martin @ pehuenche.utalca.cl
ABSTRACT:
This paper proposes a methodology for action-learning that is currently under
configuration. It is based on a model of the human actor that is inspired by biological
and linguistical investigations. This model describes how organism and observer co-
develop and shows how and why acting and explaining interact and how different
levels of coherence can be obtained.
The necessary discipline in acting and explaining is proposed in form of the
OMCA approach, which organizes cycles of observing, modeling, constructing and
acting. Inside this methodology, one specific method is developed, drawing from
congitive mapping, systems dynamics simulations and information systems
development techniques. The method is currently used in a project of the chilean
ministry of education, from which the examples or this papers are taken.
This first experience suggests that following OMCA hels producing validable
knowledge and improved action, allowing for parallel search amongst autonomous
actors.
1 Introduction
This work is rooted in the observation that the chilean educational reform is a
“wicked mess”: there is a multitude of autonomous actors intervening in its
evolution, and education as a “production” is a complex process wich allows for
divergent points of view and policies, and also one with many interrelated sources of
influence. ENLACES as part of the reform introduces computer networks into the
schools, trying to foster innovations and support autonomy. In it, there are three
levels of autonomous actors, defining different aspects of the relevant “world”.
In this type of context, OMCA (observing - modeling - constructing - acting -)
as a scientifically oriented action-learning approach can help expliciting and
improving otherwise tacit choices in the actor’s policies. We have used it to draw up
a conceptualization of this world and configure action in one of the actors, seeking to
attract the other levels, and the effect of OMCA on the models and consequences of
action are visible.
This paper introduces a model of the human actor that fundaments our
iniciative, in section 2. In section 3, OMCA is introduced, in methodological terms
and with a specific method. Section 4 illustrates its use in a selected subdomain of
ENLACES.
2 Action and Explanation as Human Forms of Knowledge
During the history of thought about human affairs, many images of the human
being have been proposed. Most of them had philosophic reflection and direct
observation as fundament. Nowadays, advances in our scientific knowledge about our
biology allow us to propose a model grounded on a scientific base. This has the
advantage of making the why? of our model of the human actor critiquable, and with
it our approach. The biological explanations in this section draw mainly on
Maturana‘s work (1997), and coincide with views held by evolutionary psychologists
(Barkow et al., 1992). The explanatins about coherence are inspired by the work on
decicion-aid of Roy (1985).
2.1 A model of the human actor
Just like any mammal, we have an organism that consists of a nervous system
on one hand and muscles, other organs, bones and so on on the other hand. Each of
these complexes is an operationally closed system: in each moment of their existence,
they find themselves in a specific structural arrangement that defines the possible
transitions towards other structural arrangements.
Thus each of the two systems is self-referring and closed. However, they
intersect in the sensors and effectors. Our muscles move (we act) when they are
triggered by nervous impulses; also, any muscle movement triggers new nervous
impulses. Additionally, many of the cascades of transition in the nervous system do
not terminate in effectors (they stay internal to the brain).
Thus it becomes possible that, as part of its self-centered transition cascades, the
nervous system correlates stimuli in the sensors with responses in the effectors. As
our sensors are in-formed by a change in their medium, they push a cascade of inner
transitions along one possible way of inner changes. The organism progressively
perceives or constructs an image out of the multiple impulses sensed, by
discrimination and classification. Thus the organism distinguishes between groups of
conditions or changes in its medium. As we follow the cascade of transitions, we may
finish in effectors that make the organism move as a whole, together with its medium.
An observer of the whole organism will call this behavior or action.
All the way down the cascade, many decisions have been taken, and at each
point new cascades may have been triggered. Each of these may result in qualitative
structural transitions (one of the possible ones inside the former structural
configuration) that we will observe as learning. Additionally, the changed medium
(in which there may be other organisms) will terminate triggering sensors of our
organism anew. Life is a continuous cascade of such transitions, and in each moment,
the structural configuration means a particular predisposition (decision flow), that we
usually call emotion.
Up to here, all learning takes place automatically, without conscious effort, in
fact even without awareness. Much of our learning works like this, and as long as we
live, we cannot avoid it. But there is more to it:
When there are several such organisms in the reach of one another, repetitive
interactions can trigger streams of learing that result in coordinated behavior. The
interactions that coordinate action are called language. In the case of human
organisms, the particular organization of the brain makes it possible to go even
further. We create objects in language -chunks of inner configurations and
movements that stand for patterns of action we share with other humans- and then
treat them as if they were external objects. In a way, we cannot avoid doing so, since
the nervous system does not inform itself back about these changes as in the case of
muscle movements (this is the lack of proprioception that David Bohm worked
about). We can thus coordinate our coordination of action in language.
In the sphere of language arise awareness and the self (Maturana), and it is there
where we exist as observers of the world and makers of experiences. We generate
observations (objects in language) that refer to distinctions in the organism, and again
we cannot avoid doing so: we cannot not explain to ourselves the experiences we
make, and by default we do this without becoming aware of it.
As with any action, what we observe as experience and how we explain it refers
to something that has realized itself without our observation and explanation.
However, observing and explaining are actions that will trigger particular cascades of
transitions according to how we observe and how we explain (Maturana).
Explanations are recursive, that is, we can explain explanations. Thus an
explanation B can tell us why an explanation A may be valid, and then explanation A
can refer to explanation B to obtain validity.
This model of the human actor is resumed in the following figure:
Explanation Language
validates < ‘) refers to
transforms C explanation: ») re-presents
g
$
E & Distinction
3S §
3 E trans-forms Emotion
= 5
=
Discrimination
Classificationn Decisions
Perception
in-forms
atu Effector
: -
Fig. 1: the model of the human actor
The following statements resume essential features of the model:
Action Explanations
changes the medium are images of the distinctions that can now be explained
changes distinctions can in themselves be explained
does not require explanations can change distinctions
may under specific circumstances change action
1.2. Coherences
Why should explanations deserve our attention? The answer is that if we wish
to escape from competence traps and other courses of superstitious learning, and if we
wish to escape from problems that stem from the abyss between tacit knowing and
explicit knowledge (Polanyi, 1983), then disciplined explaining may be the only way
out towards "theories-in-use" (Argyris, 1993).
To see how explaning may do this, let us begin by stating that the explanation
and what it refers to are two seperate things. We can invent any explanation we wish
or we happen to. We can use it to justify the past (Maturana and Varela, 1984, p.
154), or to design the future. Also, each actor creates his own distinctions (tacit
knowing) and explanations (explecit knowledge), and so there are as many complexes
of knowledge as there are actors, even if they are coordinated by language. Some
explanations may conduce to successful action, and others may not; the only way to
find out for sure is to try them out. However, all we have to check validity is
explanations, which are built on observations (experiences). No one has direct access
to what we like to call ,,external reality“.
Obviously, we wish to explain in a way that sorts out explanations that would
lead into anticipable problems. We distinguish four levels of coherence, two of wich
can indeed be obtained without trying them out.
Once we have a set of explanations (a model), we can try to find inner
contradictions, for example by simulations. If we confront the model to its own
consequences, does it resist? When we have sorted out these problems, we have
explanations in inner coherence. We can also check against other, currently accepted
explanations (as suggested by Popper, 1990), however there is a conceptual problem:
in a world where no explainer acceeds to the one, exclusive, ,,real reality", the ,,truth*
of one model does not imply the ,,falseness“ of a different model.
We can then go on to see if our (simulated) model reproduces historical
experiences (this is suggested in Senge et al., 1995; it also is one of the conditions for
considering explanations as valid according to Maturana). Once our explanations do
so, we can call them historically coherent.
And this is all we can do before acting. If we act according to our explanations,
we will become informed about the consequences, and compare them to those we
would have expected following our explanations. If we take care to specify what we
expect to observe, and how we could observe also what we did not expect, then the
information about our action‘s consequences will reveal our model‘s explanatory
coherence. This is not the same as the historical one, since the experience we had to
compare our model‘s behavior to was based on older —often even tacit- models, and
did probably not come from disciplined observing.
Beyond this comes operational coherence, which refers to the cascades of
transitions resulting from action taken. If they prove to be innocive to the actor, then
the corresponding explanations can be said to be operationally coherent. However,
this is out of the reach of explanations, that only refer to what happens outside the
sphere of language. (To illustrate this, think of a smoker who keeps smoking with a
happy set of explanatory coherent ideas, that seem to confirm his action until the day
he falls sick.)
Resuming, disciplined explaining is what we do to obtain a set of explanations
that is internally, histortically and explanatorily coherent. This means that we need
explanations in order to act and learn (as said above), and we need action to explain
and learn: acting, explaining and learning have to become united. This is what
OMCA is about.
3. OMCA: Observe — Model — Construct - Act
3.1 Linking Action and Explanation in Four Steps
Observing, modeling, constructing and acting are four types of action that,
combined in iterative loops form the OMCA approach. We propose this generic
process as means for joining action, explaining and learning in persons and in
organizations, since it enables single-loop and double-loop learning. Here, we give a
brief definition of each of these actions:
Observe
¢ Function: create raw explanations that refer to distinctions of the organism
e Function: to learn to observe
e Attention: what is not observed cannot be designed
¢ Observing the observing is useful
Model
Col
¢ Function: to generate internally and historically coherent explanations
e Function: to learn to model
¢ Causal mapping and systems dynamics simulation are useful
e For each explanation we have to explain its operational aspects, its relationships
with other explanations, and our possibilities to observe it in action
¢ Take into account implementation issues, in order to be able to distinguish them
from “theory” problems
¢ There has to be an explicit process of using the observations in order to improve
the explanations and correct "errors"
e what has not been modeled, shall not be constructed
mnstruct
¢ Function: to generate artifacts that allow to act out and test the models
e Function: to learn to construct
Act (and Observe)
e Function: to intervene in target-systems such as to obtain desired states
¢ Function: to obtain observations
¢ Function: to learn to act with a higher degree of awareness
OMCA is anchored to our model of the human actor:
Medium
in
Explanation canguage
validates refers to °
Model
co Explanation ——
fansforms re-presenl
g G
8 . .
ba Observe Distinction Construct
S
€ trans-forms Emotion
3
=
crimination
ClaSsijcationn Decisions
Act
Perception
-forms
aati Effector
- i>
Fig. 2: OMCA and the model of the human actor
3.2. An OMCA method for shaping management and information systems
Since the above formulation is general, it may be used in various contexts. We
use it as a possibility to configure and specify management systems together with the
corresponding information systems. (To be accurate, a management system is not the
system it manages; however, we cannot exclude the management system itself from
our modeling without losing the possibility to manage the management system.)
The following method embodies this intention. It divides modeling into three
steps; in the first of them, it borrows from cognitive mapping (Rodhain, 1997) and
definines a set of concepts and links that serve for creating an image of a part of the
world in wich we wish to manage. Then the map is translated into simulations that
leave us with defined policies in the management system. In the third step, a
specification for an information system will be derived from the models. The
following figure shows this schematically, before presenting each of the steps in turn:
Free observing
Conceptual modeling
Arithmetic modeling
Data and process modeling
Acting and structured observing
Constructing actor-oriewnted
information systems
Fig. 3: a specific OMCA method
3.2.1 Observe
The starting point of our work will be a joint inquiry into the Objectives we see
in the system under study. Any technique to do this may be used, but we call
attention onto dialogue and related approaches to suspending tacit knowledge (Bohm,
1996), and listening as worked out by Winograd and Flores (1989).
3.2.2 Conceptual modeling
Now we will build the first “real” model, and for this part we use a particular
notation we have defined, using the "Descision Explorer" software coming from the
field of congitive mapping (Rodhain, 1997):
Symbol Meaning Explanation
Unit
Unit: a ,,thing“ than can be
observed o experienced; takes
the form of a dimension or
attribute. In each moment,
finds itself in one of its possible
states; can transit between
states; if a range or tendency of
states is specified, the Unit is an
Objective.
Our objectives describe desired states of
something in the world; we have to say
what to pay attention to, in order to see ho
we achieve them (or not). Also, we shall
connect our models (theories) to observable
parts of the world in order to discover their
explanatory coherence. Later on, Units will
turn into types of entities and types of
attributes in datamodels.
Actor
Actor: a person, a group or a
role that controls a domain of
activity. This symbol stands for
the idea of an actor; any
concrete element is a Unit.
Things happen because actors do them;
each actor has his wishes and
particularities, and thus our configuration
process will take into acount the actors who
influence any of the Units we are interested
In.
activity: a process that triggers] An activity is a domain in which an actor
E> changes and transitions in one|configures a policy. Inside, he is free to
(ey) or various Units; will be broken| design what he choses to do, but also
down into cycles of observe,|responsible for the validity of explanations
model, construct and act. he uses to refer to his policy.
Table 1: types of concepts
Note that the software objects representing the concepts allow for descriptive
text to be added; this way, many details about Units, say, can be hidden from the
diagram without being omitted. Several types of relationship connect these types of
concept:
Symbol Meaning Explanation
a4 One or more Units can|The contributes/describes relationship
contribute to the constitution | allows to manage a system of objectives
of another Unit. Typically, | over various levels of
U3 U3 Objectives will be cosntructed | generality/concreteness. Additionally,
as hierarchies. Also, the same}we can thus represent _ ,,natural“
Contributes to/ | type of relationship describes; | influences that are not configured by any
describes for example, a Unit can)actor.
Actor describe an Actor.
Each and every activity is|Each time that an Actor has the
configured by one Actor. possibility to influence the evolution of a
Unit, he cannot escape the neccessity to
{ configures chose (even not doint something is a
sacite choice). The way to take this
liberty/obligation reveals his theories.
Usually, Actors observe one or|This enables basic forms of feedback
more Units/Objectives when | from a Unit/Objective, and OMCA could
f = = configuring an activity. not possibly work without it. This is the
observed .
U4 seed for information flow models later
Actor on in modeling.
US us One or more activities trigger|Side effects and other interrelations
changes in one or more|/between activities can be taken into
Units/Objectives. | This can} account by approaching activities from a
triggers occur with or without] global, systemic starting point.
Givity) intention. When several activities trigger the same
A Unit/Objective, it may be hard to
configures discriminate between what causes which
change.
Actor
Table 2: types of relationships
With these symbols, we can express a system. But since we are also interested
in learning and change, we will produce a cascade of models over time. For changing
our mind from one model to the following, other ideas have played a role; so we have
a special place for expressing them, too.
model 1
Model change model 2
Explanations
*purpose
ewhy it works
hypothesis (what we expect to observe)
eobserve (what we have to pay attention to)
Fig. 4: concepts that explain a change of model
These explanations may give rise to changes in the collection of Units. It may
be worth noting that the transition from what we had before starting to the first model
is not explained this way.
After having constituted an image of the Objectives, we can add Actors and
activities. We can start doing so parting from the former or from the latter. There are
a couple of questions that provide orientation:
e is my activity based on valid explanations or not?
e how did I obtain this validation (what are the Units I observe)?
¢ how could I obtain it, become able to see what I do not expect or what I
know “cannot” exist? (What are the Units I might or should observe?)
For each activity, we will work out a submodel that establishes the OMCA cycle
according to which the policy is configured. However, sometimes more than one
activity will be included into one submodel, due to interdependencies between
activities and Units/Objectives. Inside each submodel, there are all the
Units/Objectives that an actor observes, or that an activity influences. However, Units
that appear in more than one submodel or that are influenced by more than one
activity, as well as any Objectives, have to apprear in the general model, too, in order
to clearly show interdependencies.
Each submodel has to be formulated in opertational terms, explicitly
establishing:
e each observed Unit,
¢ how observations are used and new information is created out of them in
order to model action (new policies, consequences, decision criterions),
e how they are used in order to improve logic and historic coherence
(learning)
e how they are used to obtain information about explanatory coherence
(learning).
To be sure, we should not expect our progression to be a linear one!
3.2.3. Arithmetic modeling
Obtaining logical and historic coherence is much easier using the simulation
faculty of computers. Keeping in mind that our models shall be usable for
professionals who are not simulation specialists (Morecroft and Sterman, 1994), we
opt for systems dynamics as implemented by the “iThink” software. The conceptual
symbols translate easily into stock-and-flow symbols (and vice versa):
e Units <— resources
¢ trigger relationships <> information flows that regulate flows
¢ activities < converters
¢ observation relationships <> information flows
e Actors are not explicitly shown in stock-and-flow diagrams.
We can now follow the recommendations given in the tecnichal documentation
of the software. Probably the elaboration of simulation models will imply the creation
of new components (converters and the like); these can be distinguished from the
“real” components by a particular color coding. Once the actors whom we work with
have converged —for now- to a particular policy, we translate back into our conceptual
submodel.
Each particular domain submodel has to consider the possibility that during
action, a person may prefer acting “outside the system”. It matters that this be
supported by our system on all its levels of representation, as long as this exception is
duely registrated inside the system.
Once all the submodels have been frozen, the general model is updated.
3.3. Data and Process Modeling
Now the Units serve for building the data model (entity and attribute types), and
the activities allow process modeling. The resulting specification serves two
purposes: we can establish rules to be followed by persons in the role of an actor, and
we can construct automated artifacts that will support actors.
When going to construct automated support, it is important to take into account
the conceptual and interactional design (Winograd, 1996). We pretend that our
approach to the configuration of managament systems favors well-designed
conceptual models; however, interactional design has to be taken care of explicitly.
The resting steps -construct and act- are not described here in detail; one can
construct following one of the established methods, and acting shall be done
according to the configured rules and supports.
4 Acase of use
4.1 ENLACES
ENLACES is one of the projects in the ministry of education of Chile. It intents
to foster pedagogical and administrative innovations and technological autonomy in
public primary and secondary schools (for a more detailed introduction, see the
appendice). Here, we will present how in one "executing unit" (EU, in charge of
training and techincal support), our approach has been used to learn more about the
work of the technical support function.
At the outset (beginning of 1998), it was clear that tech support would have to
visit each of the EU's attended schools a certain number of times per year, and that it
would have to intervene in technical problems that do not concern PC hardware
(which is under the vendor's warranty).
In the shown extraction, that reproduces a specific part of the whole model (you
can see that there are other activities like "learn", "prepare" and "train"), we can see
that the activity "visit", such as configured by the "tech support”, triggers changes in
the "coordinator's” "activities" and in the "machines" (of a computer lab). The
“coordinator's" “learning” is influenced by these "activities", and this in turn
influences their "capabilities" that are important for their activity "repare", that also
triggers changes in the "machines".
repare
Capabilities
oa
es Capabilities
<>) <> a Technical Support
Fig. 5: technical support model 1
Coordinator.
During 1998, the EU saw that the majority of interventions had had causes that
simple that the question was: "why do they call us instead of just troubleshoot
themselves?" The frequent "firefighting" interventions that come with a lot of
travelling time have been a constant source of disruptions in the other activities of
technical support. On another level, an impression rised as if a lot of problems would
be caused more by human and organizational issues that technical ones.
In response to these observations, the question "what is the raison d'étre of
technical support?" was rised, and the final result of the following re-modeling is
shown in the following figure:
Computer Lab
State
Coordinator
Technical Support
Material
Fig. 6: technical support model 2
We see that there has appeared a superior objective (the "Computer Lab State"),
which in this extraction is constituted by the "machines" (their state) and the "access"
to them. The other change is that the "tech support" activity "visit" has gone, and
"coach" and "repare" have appeared as seperate activities. The change was justified
by the following points:
In order to:
e incentivate self-reparation
¢ stabilyze the planification and realization of jobs
why it will work: if reparation and coaching are separated, then
e obtaining help in form of visit will have a higher cost (of waiting), and
so there are stronger incentives for self-reparation
¢ there will be less time lost for traveling and less re-programming of jobs.
hypothesis: what we should observe as consequences are
¢ less calls for reparation, and less travelling and “firefighting” time
¢ fewer changes to the job-programme of a period
observe: Units that we should pay attention to are
¢ jobs of “firefighting” type
© changes to the job-programme
¢ fewer actions of type “travelling”
List-items in italics indicate Units that become part of the corresponding
submodel:
Ceassian>
//
Ceopare) [mies] Va,
[ Activites |
Tail Coordinator
SNeoy
i, >) >)
[Machines]
— 7: the <2 ane for the changed part of the model
Trainee
In this figure (in which the actor "tech support" has not been shown as designer
of its activities for the sake of lisibility; this is again an extraction of a more complete
model about the management of technical support that we have developped with
"iThink" simulations), shows some operational details about how the "assigning"
activity distinguishes between the new types of "jobs", and how the actions’
consequences are "observed" in order to "model" the "assigning" activity anew.
Besides what we have shown here, we are shaping the assignment of resources
to the diverse activities with the same method; also, there are issues dealing with the
configuration of training-courses and the management of trainers. The shaped
policies are constructed in form of Java applications that connect to a relational
database.
4.2 Current question marks
Each issue at hands demands time to observe, model and construct.
Additionally, there are a lot of issues to be treated. During all the time needed, life
goes on and the actors have to act. This gives rise to the question: "does everything
have to get OMCAed in order to do it right?" A good heuristic to answer this is:
whenever there arises a doubt over if a particular issue might be improved, one can
use the described OMCA method, one issue at a time. As long as there is a general
map that captures interdependencies, each decision opportunity can be trated in
isolation.
A second question may be “are ther other OMCA-compliant methods I could
use?" Yes, we think that there are: especially "decision aid" (Roy, 1985) can be used
in chains of decisions such as to substitute the disciplined construction of alternatives,
consequences, dimensions and criteria for the presented modeling methods.
5 Conclusions
In this paper, we have started by presenting our model of the human actor, that
establishes the distinction of the spheres of existence: the organism and the observer.
According to it, explanations can show various levels of coherence, of which the
highest accessible one needs action in order to get informed on its coherence
(validity). OMCA has been presented as an approach to combining disciplined acting
and explaining in order to improve actions and learn the necessary, congruently with
the model.
One particular way to do the modeling part has then been introduced; it intends
to shape management systems, comprising the corresponding information systems. In
its way to map ideas, this method respects the actors' autonomy and distinguishes
between observable states of the world and the processes that influence changes in
them. This tool stays sufficiently simple to be read by a no-specialist, allowing
simulation-for-learning at the same time.
It is has to be said that this is ongiong work, and thus no strong claims on
validity can be made yet. However, we believe it to be a workable approach to
introduce organizational learning into the everyday work of everyday people. We also
believe that various actors can use OMCA to cultivate their respective policies/models
by responsible experimentation-in-action, using the resulting diversity as kind of a
parallel search system for viable policies.
6 References
Argyris, 1993
Barkow et al., 1992
Bohm, 1996
Brown and Duguid, 1991
Kuhn, 1971
Maturana, 1988
Maturana, 1997
Morecroft and Sterman, 1994
Rodhain,1997
Polanyi, 1983
Popper, 1990
Roy, 1985
Schafiernicht and Pereira, 1998
Schon, 1983
Argyris, C. (1993), Knowledge for Action, Jossey-Bass
Barkow, J., Cosmides, L. and Tooby, J. (1992), The Adapted Mind, Oxford
University Press
Bohm, D. (1996), On Dialogue, Routledge
Brown, J. S., Duduid, P. (1991), Organizational learning and communities-of-
practice: towards a unified view of working, learning and
innovation, Organization Science, Vol. 2, No. 1, 2/1991, p. 40
Kuhn, Th. S. (1971), La Estructura de Revoluciones Cientificas, Universitaria
Maturana, H. (1988), The ontology of observing, Irish Journal of Psychology:
9/1, p. 25-82
Maturana, H. (1997), La Objetividad - un argumento para obligar, DOLMEN
Morecroft and Sterman, (1994), Modeling for Learning Organizations,
Productivity Press
Rodhain, F. (1997), “La construction et la confrontation de réprésentations: le
cas des besoins en information”, Doctoral dissertation, Université de
Montpellier IT
Polanyi, M. (1983), Tacit dimension, Peter Smit, 1983
Popper, K. (1990), La l6gica de la investigacién cientifica, Tecnos
Roy, B. (1985), “Méthodologie Multicritére d’ Aide 4 la Décision”, Economica
Schaffernicht, M. and Pereira, J. (1998), Building a flexible management
system, Working paper of the “Departamento de Informatica de Gestion,
Universidad de Talca”; ask copies at martin @pehuenche.utalca.cl
Schon, D. (1983), The reflective practitioner: how professionals think in action,
Basic Books
Winograd and Flores, 1989 Winograd, T. and Flores, F. (1989), Understanding Computers and Cognition,
Abkex Publishing
Winograd, 1996 Winograd, T. (1996), Bringing Design to Software, Addison Wesley
7 ~ Appendice: a brief introduction to ENLACES
In this additional section, we present the essentials of the ENLACES initiative
in which the presented method is used; also, we explain why this seems a relevant
"laboratory".
7.1_ENLACES as part of the chilean educational reform
Today, three kinds of schools provide basic and secondary education to chilean
children: public, private-subsidized and private-payed. The ministry of education has
responsibility for defining and overseeing minimum general standards, and at the
same time, it is the superior instance for the public schools, that depend directly from
their respective community (municipality). Chile has a high concentration of
revenues, and so it is not surprising to see the vast majority of families send their
children to the free public schools; those who are able to pay a bit use private-
subsidized establishments, and relatively few can afford private-payed schools.
Due (though not only) to the scarcity of finance, the quality of schooling is
roughly equivalent to what families actually pay. In this situation, the ministry of
education has set up “the reform” and a special programme for the improvement of
equity and quality in education (“MECE: Mejoramiento de Equidad y Calidad en la
Educacién”) under the motto “good education for everybody”. The basic idea of this
reform may be called “local autonomy” (decentralization), which is thought to free
schools from bureaucratic burdens and allow for pedagogical and administrative
innovations and progress.
Part of the reform is ENLACES (spanish for “links”), which is an initiative that
searches innovation in pedagogy and administration, as well autonomy by the
massive introduction of computers into schools. Currently, this means installation of
a peer-to-peer PC network with MS Windows 95, the MS Office 95 suite, a shared
printer and a phone-line with modem for inter-school communication, and the “Plaza”
software; together with the installlation, there are two trainig courses and technical
support during a limited period of time. The heart of ENLACES is a software called
“La Plaza”, which is an on-screen “plaza de armas” (the central place in chilean
towns), specially designed to be easily understood by kids. The “Plaza” offers
electronic mail and interest-lists, as well as access to educational software packages.
On the organizational side, the ministry has formed “zonal centers” (ZC) that
are in general universities that assume responsibility for determined geographical
spaces, in matters of technical implementation and trainig provided to schools. Each
zonal center hires “executing units” (EU) which will actually carry out the technical
supply and training. Inside the target space defined by the ministry as “innovation
(pedagogical and administrative) and autonomy by informatics”, each actor is free to
manage and act as he understands; the national coordination (at the ministry) visits
each zonal center in turn, to stay informed about the local advances and difficulties.
7.2 Towards the definition of a problematic situation
The concepts directly mentioned by the objectives-statement (at the level of the
national coordination) are innovation, informatics, pedagogy, administration and
autonomy. This could make one expect the intervention into schools to be designed
such as approaching all of the issues. However, the practice of training and support
has been limited to informatics. This is one possible choice, but we may wonder
wether it has been taken consciuosly, in an informed manner, and wether other
possible choices have been tried out or its validity is being tested.
Inside the informatics training and support, other choices have been made that
are not part of the ministry’s objective statement. For example:
© training is distinguished from technical support; (however experience shows that a
lower level of user skills comes along with a higher demand on technical support,
with a causal relationship from the first to the second.)
© training is divided into two separate courses, the first of which is intended to build
“basic user” culture, and the second proposes educational applications; (however, it
has not been made explicit what exactly has to be understood by “basic” nor by
“user”, and the software distributed with the “course 2” uses the computer as an
encyclopedia rather that a tool for simulation, for instance, which is another
implicit choice; different choices seem to be possible, and we do not know why
things are as they are.)
¢ each executing unit has liberty to design and carry out its own training and support,
as long as it respects the ministry’s objectives and the above mentioned choices.
There is no systematic use of already made experience in order to improve the
quality of the one who made the experience, nor of helping to provide orientation
to starters-up. Recently, a common manual intents to norm down the possibly
existing diversity, which may be seen as one possible choice to un-do the problem
of dynamically “control” (in the cybernetic sense) this diversity; (however,
instruments for rapid accesso to valid information might be an alternative that
would not be anti-diversity.)
¢ quality of output is d by supervision visits that focus on input; evaluation by
mapping the innovativeness of participating schools after participating into the
training against the type of trainig provided, in order to distinguish patterns inside
the multitude of courses provided to about 2,000 schools (up to now), is not part of
current practices. Recently, an ex-post impact appraisal on what has happened in
the first ENLACES sites has been bought from a consulting firm; (however,
Internet and remote database systems may provide a base for configuring
instruments for faster in-process feedback at a lower cost.)
One last choice needs some explanation before being cited.
According to Polanyi (1983), we act basically in a tacit manner (what Argyris,
1993 calls “skillful”), and when we try to make explicit descriptions of these
processes, we first lose our skillful acting by the decomposition, and later on we may
“compile” it back into readily available knowledge for action. In this sense,
innovating is what Schon (1983) calls “art”: it may well be rigorous, but it lacks a
degree of explicitness to be “scientific”. These capabilities are learned as “becoming
part of a community of practice “ (Brown and Duguid, 1991), and Kuhn (1972) writes
that even scientific practice is learned this way.
It should make sense, then, to expect that innovating is best learned by
participating in innovative activities, and in this sense the ENLACES interventions
might be an opportunity to generate such an experience. However, ENLACES looks
for innovations in schools, not in itself, and nothing in the organization between the
ministry, the zonal centers and the executing units seems to aim systematically at
fostering innovation at these levels.
Thus, we distinguish spaces left to explore, in which both single-loop and
double-loop learning would contribute to effectivity, and in which the parallelism
inherent in the system would allow to validate information and share it across the
country-wide organization. We use our observation that such efforts are currently not
made, to bring in our own proposal.
7.3 The global “infotecture”
Putting the parts together, one obtains a global image of a possible
organizational action/learning process inside the domain of ENLACES. In Fig. 8, we
show the three levels of ENLACE's organizations (M for ministry, ZC for zonal
center and EU for executing unit) in their respective OMCA cycles. One can see two
important things. First, the actions and what is constructed at one given level in the
organization are observed by the next lower level, for which they constitute kind of a
frame. For example, the zonal centers take the rules and the perceptible behavior of
the ministry (memos, visits, tools, coordination style and so on) as part of the world in
which they move. Second, the observations made at one given level are available for
observing at the next higher level. For instance, when an executing unit observes a
high rate of learning in one particular content, the zonal center (and thus the ministry)
will observe this, too.
>)
Fig. 8: OMCA at the different levels of ENLACES
@)
@)
5
OOO 000
This way to look at the whole organization reveals that there is a network of
interacting cyclic processes, where each is an autonomous system (operationally
closed in the sense of Maturana, 1988). This allows to stress the importance of
learning processes at each level: if we wish to design a global system of autonomous
systems in structural coupling (co-evolving), there have to be the connections and the
internal processes that are to be triggered by interaction. Any attempt to constitue a
high-performance executing unit is conditioned by the existence of a high-
performance zonal center and vice versa. The same holds between the zonal center
and the ministry.
We observe that it is of great importance to approach ENLACES from a
systemic (uniting or global) perspective, with a special attention to structural coupling
(Maturana, 1988) between the actors at their respective levels.
As for the time-horizon of each of the cycles, it is important to keep in mind that
the illustration oversimplifies the processes; as seen in the paper, there are various
cycles at each level (one for each activity). Accordingly, some of the proceses cycle
faster than others. However, this does not interfer with the general statement of that
the processes interact. We may anticipate that a given process p at level / will have a
time-horizon of not less than the processes at level /-/ that inform process p.
CC BY-NC-SA 4.0