Critique of Action Technologies' "Business Process Methodology"
April 1994
This write-up on ATI was composed in a larger context in which the workflow
market was examined in both qualitative and quantitative ways for a client.
Some paragraphs specific to the client have been removed, but the value
of surfacing these arguments remains. All information about the product
stems from April of 1994, and no attempt has been made to revise the perspective
in light of any changes in direction or marketing of ATI. The fundamental
points, especially regarding their patent claims and overall approach, stand
in any event. The text was composed by Paul Pangaro after collaboration
with Walter Lee.
Scope of this critique
The full depth and breadth of discussion regarding the value and structures
of a business process methodology are beyond the scope of this critique.
What follows is a summary of initial findings based only on a review of
ATI's marketing materials, presentations, demonstrations, software documentation,
and publications (see references for a complete list).
Although we would not expect any contradictions to our critique to arise
from actual use of the software or discussions with ATI users, it must be
quite clear to readers that such inputs were outside the scope. Deeper inquiry
would benefit from such inputs.
This section presumes basic familiarity with ATI and their approach.
Reasons for concern over ATI
Software which embodies a "Business Process Methodology" (BPM)
receives wide respect in the workflow industry. General wisdom holds it
useful to have some methodology to guide the application workflow software,
because it structures the process, allegedly leads to better understanding,
gets at the basis of human discourse in business, etc.
ATI has a stronghold on BPMs because
- ATI has an approach that appears encompassing and powerful
- their approach appears to have a philosophy behind it
- ATI had been issued 2 patents covering communication elements of their
BPM "loop of agreement", consisting of preparation, negotiation,
performance, and acceptance
ATI receives a great deal of media attention, based, it would appear, on
their own marketing. It may seem difficult to fault ATI's approach, given
its apparent scope and power. Without other commercial contenders, criticisms
may not be obvious. However, review of their own product literature exposes
a variety of inconsistencies.
Debate on contribution of "a methodology" for workflow
The issue of the potential contribution of "a methodology" for
workflow persistently arises and must be handled from first principles.
The appeal of a methodology (or, as we prefer to call it, a philosophy of
business organization or culture) is probably a mixture of "getting
religion" and an acknowledgment of some advantages to an analytical
tool to simplify, categorize, structure and give coherence to a complex
problem space. The given wisdom (from ATI's marketing literature, and industry
experts such as Marshak and IDC) is that having a philosophy is good - although
most if not all (clearly ATI, QDM and IDC) make the conclusion based solely
on ATI's approach.
IDC argues that it is necessary to consider business processes in the abstract,
at a level above "moving the information around." We agree, although
there are many possible modes of abstraction "above" the data:
pathways, states, transitions, roles, purposes.
Ironically ATI's "loop" leaves out anything explicit about the
data itself. The methodology is pitched at the level of interpersonal transactions.
ATI's rhetoric about promises and commitments and the loop of agreement
provides a psychological "warm fuzzy" that is intuitively graspable
and relates to everyday interactions (promises, commitments, etc.).
Note that QDM acknowledges ATI's contribution but then declares that most
business processes, especially those in the Ad Hoc workflow market segment,
cannot be analyzed a priori and hence a BPM is far less useful. We feel
that there is still utility in a BPM for workflow including Ad Hoc that
is pitched above the level of interpersonal transactions and routes.
This type of BPM would operate at the level of the purpose for the
processes of the organization that manifest in artifacts such as routes.
ATI gets tremendous mileage out of having a philosophy. Our impression is
that the feature set of their products does not have full overlap with that
philosophy.
The sources and scope of ATI's "loop"
ATI claims that their approach stems from important intellectual work, implicating
cybernetics, phenomenology and speech act theory [Winograd and Flores, 1986].
In practice, only speech act theory is directly involved in the design of
ATI's products. They have significantly watered-down the impositions made
by The Coordinator software (since sold to Da Vinci) in the equivalent aspects
of their new Analyst software - although remnants (promise, agreement to....,
etc.) of what many found objectionable in The Coordinator are the basis
of the Analyst.
Stripped of their rhetoric, ATI's Analyst contains an implicit model for
appropriate personal interaction contained in an explicit model of what
they call "agreement." This is not a model of business
in any conventional sense. Note too that this is not the meaning of agreement
in the sense of "holding opinions that are aligned", a sense which
is more relevant to the organizational forms emerging in today's businesses.
ATI's notion of agreement is closer to the contractual one of covenant,
with the connotation of service to another's needs. However there is little
in practice in their methodology that requires or even could involve
explicit participation in another's values. Involvement by the performer
in the customer's values contributes enormously to the potential for success
of an enterprise. This is because without an understanding of the customer's
values, it is far more difficult for the performer to produce a result that
will conform to the full context of the customer's "conditions of satisfaction."
We are not suggesting that everything known to the customer must be known
to the performer, but we maintain that understanding a customer requires
a much larger sense of context than ATI's BPM encourages.
ATI's components of "how agreements are formed" are preparation,
negotiation, performance, and acceptance. In the social context of a corporation,
these components (especially the last 3) become what we have called "value-laden";
that is, they are interpreted by individuals in the context of values that
vary depending on the relationships of the specific individuals involved
(for example, boss to worker vs. peer to peer). It is critical to
note that these values are not expressed in the model. If a promise
(or other transaction) is made, the value or purpose of that promise is
assumed; this might be defended by ATI with the claim of the "universal"
nature of these transactions:
"The loop structure is universal in that it is independent of
any culture, language, or communication medium in which it is conducted."
[Medina-Mora et al, 1992]
(It appears that no one has called the bluff of this insupportable claim.)
Many interpretations can be made of the implicit purpose of the promise
made by the "performer" in their workflow loop. This could be
something like "Mature behavior requires promises" (surely an
assumption and not a universal) or "If an individual doesn't promise
he is unreliable" or "Without promises the boss has less control."
The relevance of these implied but persistent values must be taken in light
of the current evolution of control in organizations, from hierarchical
and authoritarian to distributed and mutually-defined.
The conclusion must be that ATI's model declares that business must/does/should
operate under certain interpersonal transactions to be effective.
ATI's methodology in respect to "satisfaction"
ATI emphasizes time-based and cost-based criteria for satisfaction, which
are simple cases. The Analyst software provides an opportunity to specify
a text description of conditions of satisfaction, but the notion of metrics,
so critical to re-engineering, TQM and the like, is not introduced explicitly
(see next section).
Furthermore ATI's approach is weak to the extent that "conditions of
satisfaction" for a given exchange cannot be known a priori.
The "customer" is able to specify conditions of satisfaction only
to a limited degree in advance of using what the "performer" produces.
In practice evaluation of the quality of the result can be made only long
after the negotiation is made, and somewhat after the performer delivers.
Their "loop" is not construed to be an iterative process to refine
products as a consequence of "completed" performance and subsequent
evaluation.
Thus there is no "primitive" or even "convention of use"
within ATI's software for this fundamental aspect of business process. In
response to this issue being raised, ATI's marketing people offered that
it was sufficient and appropriate simply to "add another loop to the
business process", which misses the point that their "methodology"
misses the point.
Business processes and ATI
Some initial common sense questions in the context of TQM and "re-inventing
the corporation":
- Does ATI's methodology require metrics for business performance? encourage
them? allow them to be incorporated into the analysis beyond cost and time
to complete?
- Does ATI's methodology provide for explicit modeling of information
collection?
- Does ATI's methodology provide for modification and maintenance of models
after a business process re-engineering effort? Does their commitment to
"universal" primitives of discourse actually achieve this?
- Does ATI's methodology gracefully allow for an evolving measure of customer
satisfaction?
If the answer to the last question is no, then ATI does not possess a strategy
for the evolution of a business. The more the answers to the above questions
are in the negative, the less ATI provides a tool for the current wave of
business process management tools, including TQM.
ATI's rhetoric implies that they have solved the "principal-agent"
problem, though they haven't. The principal-agent problem arises in the
common example of car repair. You bring your car to be fixed, and the mechanic
gives you a bill for $1,000. The good news is that your car is fixed, the
bad news is you cannot tell whether you got a fair deal or not. The problem
is more than one of information/observation and competancy. It arises in
many situations: shareholders vs. management; management vs.
employees.
ATI is about the fulfillment of promises and commitments. How can the agent
know whether a principal is acting in the agent's or the principal's interest?
The role of workflow in increasing productivity conjures various cases:
- "Fixing" the process by merely incorporating workflow as automation
of existing process, vs. process redesign
- if incorporation of workflow is merely automating of task, this constitutes
ossifying management control in a "batch" process
- automating tasks provides limited productivity increase compared to
providing real-time feedback to individuals to monitor, add value, and/or
fine-tune the process (the real meaning of Japan's kaizen).
- Re-engineering, no matter how well performed, is basically static. Even
the best process redesign implicitly casts the new process in concrete.
Kaizen, by contrast, assumes that processes are always broken - i.e.,
they can always be improved. To make sure this is so, the Japanese deliberately
stress processes (by withdrawing resources) until they do break. They then
figure out how to "fix" it so it runs with less resources; and
then they stress it again. The workers involved in the process are given
the training and the real-time feedback to continually improve it.
Where is the business value in methodologies? For example, what is the relative
value of providing efficient processes (the usual claim of BPMs) vs.
establishing shared mental models? Shared mental models are widely considered
to be a requirement for coordination in the "new corporation"
that is no longer controlled by hierarchy (e.g. Senge 1990), but
are not spoken of explicitly in the workflow market. If we accept the importance
of shared mental models, then consider that
- shared mental models must be logically prior: efficiency cannot come
without shared mental models about the purposes and processes of an enterprise
- shared mental models are also required to provide a measure/sense of
what is effective
- processes must be evaluated not only for efficiency but also for effectiveness,
which requires shared mental models to evaluate (effective against what
criteria?)
- shared mental models are therefore more important than process definition
"Value-free" approaches to systems/business modeling
In regard to any modeling approach two questions must be asked:
- to what degree does the system impose its own philosophy?
- to what degree does it "mirror" the existing values and processes
of a business?
Each question is really the converse of the other. Imposing values may have
the advantage of bringing new values and processes to a business, or the
disadvantage of preventing valid, pre-existing aspects to be captured and
maintained.
An alternative approach to ATI's would be a "value-free" philosophy.
Of course every structure or philosophy implies some values; the
issue is, where do the values lie: in the domain of interpersonal relations
or in the domain of effective action relative to explicit goals (these are
different). What is the importance of this issue where the ultimate values
to be respected are those of the customer?
We must remember that any tool that is applied by some individuals to design
business processes, and whose results are used (and not modified) by others,
contains very strong implied values: top-down dictation of process; hierarchical
control; business as a predictable production line.
One test of relative amounts of value implied in any model is the capability
of embedding one model into another. (For research work that is moving into
the commercial arena, see the OVAL system by Malone et al 1992.)
References
ATI Product Literature (as of mid-1994):
- ActionWorkflow Analyst
- ActionWorkflow Manager for SQL Server
- ActionWorkflow Manager for Lotus Notes
- ActionWorkflow Builder
Barney, Doug: "Action's Workflow guides users of SQL and Lotus through
tasks", InfoWorld, Oct. 25, 1993, v15, n43, p22 (1).
Marshak, Ronni T.: "Action Technologies' Workflow products: Coordinating
the activities of people as they work together", Patricia Seybold Group,
The Workgroup Computing Report, May, 1993, v16, n5, p4 (8).
Medina-Mora, Raul; Winograd, Terry; Flores, Roderigo and Flores Fernando:
"The Action workflow approach to workflow management technology",
CSCW Nov. 1992, p281 (8).
Medina-Mora, Raul; Wong, K. T. and Flores, Pablo: "ActionWorkflow as
the enterprise integration technology", Data Engineering, June, 1993,
v16, n2, p49 (4).
Malone, Thomas W.; Lal, Kum-Yew and Fry, Christopher: "Experiments
with Oval: A radically tailorable tool for cooperative work", CSCW,
Nov. 1992 p289 (9).
Radding, Alan: "Stopping the bottlenecks with workflow analysis",
InfoWorld, Jan. 31, 1994, v16, n5, reprint.
Rodriquez, Karen: "Action helps user design, modify document management
systems", InfoWorld, Nov. 1, 1993, v15, n44, p54 (1).
Rooney, Paula: "Action aims workflow modules at Notes, SQL", PC
Week, Sept. 13, 1993, v10, n36, p75 (2).
Schaidt, Patricia: "Tom White: President and CEO, Action Technologies",
LAN Magazine, March, 1993, v8, n3, p30 (3).
Senge, Peter M.: The Fifth Discipline, Doubleday Currency, New York,
1990.
|
