Re: Corp. Knowledge Repository LO555

mbayers@mmm.com
Mon, 27 Mar 1995 09:42:52 -0600

Replying to LO492 --

Alexia asks in her note about the makeup of the designers who use
our Design Log to pass on the rationale for their design decisions:
Are they in the IT group? Are the eventual users of the results of
their efforts involved in the design process? How are the 'rules'
captured? How are they used later? Did we ever define critical
success factors?

All excellent questions, some without good answers. Let me offer a little
more explanation . . . We have an on-going problem with information
systems constructed in the past for which we have no real understanding of
how they were constructed. In the case of older systems, the staff
responsible for their construction have typically moved on to other
project and new systems. In the case of more recent systems, the staff
who made those decisions were frequently contractors / consultants who
left after the project ended -- and took all that knowledge with them --
perhaps even to competing companies -- who knows?

To _try_ to retain that knowledge, we _encourage_ designers to create
the Log. We cannot (from our support position on the periphery)
mandate that they create the Log. We try to make it as easy as
possible, and try to foster the notion that documenting the
decision is part of making the decision.

We use a Lotus Notes database for our log -- but it remains largely
a prototype. In today's environment, a designer can search for
such things as, 'Has anyone made a decision about storing
data on such-and-such a database at severals sites versus storing
the same data at one central site? What did that designer
consider in making that decision?'

My hope is that someday, after more of these decisions have
been logged and enough time has passed to somehow evaluate the
effectiveness of those decisions, that we can cull the decisions
and move them into some sort of case-based reasoning system. At that
point, a new designer could look for the same question, but now
get not only what was considered, but did it actually produce the
intended results? That clearly requires a retrospective view, and
the application of some sort of evaluation scheme.

As for what questions we ask: they fall into several categories.
We want a relatively _objective description- of the decision in a couple
paragraphs -- what did the designer decide? We want also a relatively
_subjective prescription_ of the decision -- what would the designer
prescribe as preferrable in this situation? Which audiences /
stakeholders held greatest sway in the decision? Which segment of
the overall system had priority? Which aspects of overall systems
quality had priority? What prior decisions had significant influence
on this decision? Are there specific scenarios where this decision will
have some (subtle) impact? Are there ethical implcations for
this decision? How much latitude did the designer have?

We just now begin to see a few people who think use of such a log
may offer some value. The payoff, if any, will occur at the point
when we can make new decisions based on what we learned from old
decisions. The transmission of learning from one generation of
designers to the next requires a forward-orientation that must
struggle against the prevailing 'get it done now' mindset.

Michael Ayers
mbayers@mmm.com (612) 733-5690 FAX (612) 737-7718
3M Center Bldg 224-2NE-02 PO Box 33224 St Paul MN 55133-3224
All ideas expressed in this note represent the author's thinking
and do not represent the positions of any organizations.