Healthy Competition LO18180 -Naval Aviation

Michael A. Beedle (
Sun, 24 May 1998 03:41:45 -0500

Replying to LO18158 --

Don Dwiggins wrote:
> This relates to something happening in my company; we're reevaluating our
> compensation strategy (with assistance from an outside consultant). We've
> always had a peer review system, where the person that evaluates you today
> may the one you evaluate tomorrow. I've recommended that we go to a more
> frequent performance review, somewhat decoupled from compensations, so
> that the feedback loop can be tighter and more sharply focused (like the
> LSO's). Also, it means getting the feedback from those who know your
> recent work best. Reading your message, it occurs to me that to make this
> most effective, we need to do some learning in the area of team support
> and evaluation, in order to minimize the potentially immergent results of
> reviews.


(How are you? I hope everything is going great in CA.)

Your posting reminds me of the "mentoring networks" used by some
successful consulting companies (ours being one of them), or even the
hypertext organization described by Nonaka and Takeuchi in "The Knowledge
Creating Company". (I won't bother with the exact reference since this
should be one of the basic texts read my most here.)

These "tight feedback loops" you describe held among a selected number of
reviewers, form a "hypertext org" pattern of relationships. And they are
very "structural" things.

For example, Anne Maria Bell from Vanderbilt University, has actually
drawn pictures of some of these hyper-text networks coming out of
empirical data in market exchanges. So they are very real.

However, the "Reviewer Hypertext Org" you describe, differs from the
knowledge based "Hypertext Org" from Nonaka and Takeuchi, because its
relationships are not based on "knowledge exchanges" or on "mentoring
exchanges", but on "review exchanges".

So, where do we go from here? (I was about to say: "so, where do you want
to go today?", but I might be infringing a trademark used by an all too
familiar Seattle based company... :-)

Well, one direction perhaps worth following, is a "unified network"
concept, where one superimposes all of the hypertext networks into a
"integrated multi-faceted hyper-text network" which links capture:

- knowledge exchanges
- peer review exchanges
- mentoring exchanges
- service exchanges
- <add your favorite resource exchange here...>
etc. etc.

and at a larger scale (teams, organizations, division, companies):

- economic and workflow exchanges (e-commerce, EDI,
value-chain integration, BPR, virtual partnering, etc.)

A self-interested agent of this organization (i.e. a person, integrated
team, or company, etc.); "fits well", or more bluntly "survives", in this
environment if it is capable to define:

a) who are its networked connections
b) what are the "messages" it responds to (i.e. its sensors)
c) what are the rules that define the exchanges
(i.e. its internal models)
d) what actions does it take (i.e. its effectors)

n) And most importantly, how does it best satisfy its utility
functions while it LEARNS from its environment.
(i.e. the hard part here is to define what is important to your
"self" interests == what is key to your survival...)

In summary, and agent survives, only if it has a well crafted "agent
architecture". Similarly a multi-agent survives ONLY IF every agent or
multi-agent that composes it, has a well defined and compatible "agent
architecture" with some shared utility functions among each of its agents
(or multi-agents).

(HINT: Here I am going to give you a small hint of what we are attempting
to do at Framework Technologies:

People and Teams can be instructed precisely on
how to define effective "agent architectures", and how
to operate according to some patterns, so that these
architectures are properly installed, and result in
intelligent and adaptive multi-agents.

If at every level of scale, organizations can be taught
on how to operate as multi-agents (forming coalitions and
sharing utility functions), the resulting organizations
yes "emerge from self-organization" but in a
very directed way (according to a pattern language).

So, we are talking about "controlled emergence", with
the purpose of re-creating patterns of hyper-productivity and
comfort to the employees of a company.

At the recent Agents98 conference I proposed to capture these emergent
behaviors using pattern languages, because the "aggregates" and even the
multi-agents, resulting out of "emergent behaviors" are not predictable,
thus the resulting "structures" need to be tabulated using hyperlinks
among them using their initial and resulting contexts.

OTOH, from the "patterns" perspective, this is really nothing new, because
after all I am just proposing capturing structures that work
synergistically with each other.

The difference is in the realization that these patterns have to be
installed both from the outside-in, and from the inside-out i.e. that they
must be generated from properly defined "agent architectures" but that
they need to be supported by proper management directives.

- Mike

Michael A. Beedle Ph. D.

Framework Technologies Inc.
1901 N. Roselle Rd., Suite 800
Schaumburg, Illinois.

(312) 218-6562 cellular
(773) 693-0876 direct line
(847) 490-7110 office

Management and Technical Consultants specializing in:
Knowledge Management, BPR, Applied Complexity Theory,
Enterprise Architectures, INTERNET applications,
Object/Agent Technology, Frameworks, (UML, Java, CORBA, C++)


"Michael A. Beedle" <>

Learning-org -- Hosted by Rick Karash <> Public Dialog on Learning Organizations -- <>