is a computer lawyer and legal technology consultant
based in St. Louis, Missouri. He speaks and writes frequently
on legal, technology and Internet topics and was named
the 2001 TechnoLawyer of the Year by TechnoLawyer.com.
His highly regarded web site at www.denniskennedy.com
collects many of his articles and is the home of his
is the president of Prism Legal Consulting, which helps
law firms with the strategic use of technology and legal
market software companies with marketing and strategy.
He is a lawyer by training and has held senior management
positions at two large law firms and two legal software
companies. His articles and blog are at http://www.prismlegal.com.
A Mini Roundtable on KM Strategies That Can Work
Dennis Kennedy (DK): I recently participated
in a discussion with some CIOs of large law firms about
current knowledge management (“KM”) developments.
I was surprised to learn that many KM projects seemed
to have as their goals the results that have been associated
for many years with document management (“DM”)
systems—full-text search of documents, descriptive
fields, taxonomies, and the like. I played devil’s
advocate and asked, “Is large firm document management
broken?” I was not satisfied with the answer so
I’ve been asking the question of other experts.
Ron Friedmann is one of the leading legal technology
experts in the business. In the following conversation,
we discuss the lessons we’ve learned from document
management and the implications for KM projects.
Ron Friedmann (RF): I don’t
think DM is broken, though it is true that full-text
retrieval has not worked so well (a problem the vendors
are fixing). Rather, the problem is with the premise
of what the “results” of DM are supposed
to be. In the early 1990s, many of us thought that DM
would solve an early KM problem – work product
retrieval. DM did not, even when full-text worked, for
several reasons: too many duplicates (even when only
searching on the last version of any document), meaningless
titles, no enforcement of correct selection of document
type, lack of contextual descriptions, among others.
These were, and are, problems of process and use, not
technology. One interesting aside re full-text: larger
firms are shifting from just searching DM to “federated”
searches across multiple data sources.
DK: I see all of this from the individual
lawyer’s point of view. Any system breaks down
whenever a lawyer feels that he or she has to fight
with the system just to do simple things. In my case,
it drives me crazy when I have to input information
many times in many places. The goal of good computing
is “enter once, use many times.” Use every
available way to reuse information you already have
entered at least once (such as client/matter numbers
and other information) to automatically put metadata
on documents using simple business rules. I can’t
emphasize this enough: making lawyers enter information
multiple times is a recipe for failure.
RF: Agreed. We have to make KM as
easy as possible for lawyers. An emerging trend along
these lines is that work product "inherits"
descriptive information (“metadata”) from
the matter – an element of the hot “matter
centric” trend. This means, of course, that you
need to have the discipline to create good data about
the matter. But many lawyers lack this discipline.
DK: I think that lawyers (or, realistically,
their secretaries) will adjust to providing more data
in the client intake process, if that information is
used to populate DM, KM, or CRM (customer relationship)
fields. The key is eliminating duplicate data entry.
Just as important is limiting the number of documents
lawyers have to review to create the system. I believe
in taking a simple approach by reducing the dataspace
by excluding documents older than a certain date as
irrelevant unless specifically placed into the system.
It’s somewhat arbitrary, but it’s a reasonable
way to reduce "noise" and project scope—sort
of like the 80/20 rule. It helps get things done.
RF: Absolutely. While the “half-life”
of any given document is unknown, ignoring older documents
is a reasonable cut to make, especially if you are just
getting started. And, as you say, you can always include
an "old" document known to be valuable still.
But you have to instill a discipline for new matters
and documents to capture some additional information.
DK: Right. The sooner you get a usable
system, the sooner lawyers will see its value and “remember”
older documents that might be good additions. In most
cases, though, it will turn out that there are few,
if any, old documents that “need” to be
included. I also learned an important lesson from Dan
Felean at Pensera, a KM company: consider having secretaries
gathering what “KM” info they can as part
of "closing" files. Accept whatever you can
get through this process and do not keep badgering the
lawyers. The info you get will almost always be better
than what you get by having lawyers fill out forms.
Similarly, I’ve had the idea of using college
interns to interview, by audio and video, lawyers as
a way to capture information, especially the institutional
stories and wisdom of older attorneys. It’s also
a good way to gather biographical and practice information.
I'm sure that you know how hard it is to find out anything
useful about someone in your firm who is in another
RF: I agree that secretaries can and
should be part of the process. My client, ii3 has software,
AdvanceKnowledge, that incorporates secretaries in the
KM workflow. Canada's largest law firm, McCarthy Tetrault,
uses AdvanceKnowledge and secretaries play an important
role there. I also agree that conducting matter post-mortems
is a great practice (and some firms do it); I disagree,
however, about using interns. Interviews need to be
conducted by someone knowledgeable, which typically
means non-practicing lawyers, experienced legal assistants,
DK: My next point may well be heretical,
but I’m coming from the lawyer’s point of
view. Lose the categories and taxonomies. As William
Blake might have said about this, “I must create
my own categories or be enslav’d by those of another.”
I'm about to completely change my subject matter folders
in my email, bookmarks, and newsreader for the umpteenth
time. It's too difficult to determine categories and
assess relevance early in the process. The relevance,
importance and categories may come much later. There’s
a notion of “fluid” or “liquid”
categories that I’ve seen lately that makes more
sense to me. If you must use taxonomies, keep them simple,
base them on something that is familiar, and make it
easy to assign content categories. I’d rather
see an auto-assignment of categories with the ability
to add or override rather than be forced to find a place
among nested lists of categories.
RF: Original DM did not support taxonomies,
so that concept is relatively new. I think we need to
keep taxonomies. First, many firms are adopting them,
so the market is saying they have value. Second, all
my legal tech experience, which started with developing
work product databases and hypertext research systems
in 1989, suggests that a combination of browsing via
categories and search is the best way to find material.
Third, tying this back to entering data only once, a
taxonomy facilitates automatically assigning meta-data
to documents. Fourth, the jury is still out on the accuracy
and value of software auto-classification (though I
am optimistic about this approach long-term). And fifth,
the problem of re-categorizing can be managed, though
it can be a bit tricky; you may need special taxonomy
management software and you have to make sure changes
are properly propagated throughout all systems. All
that said, I personally lean toward relatively simple
taxonomies, though that is a point of contention among
DK: My point would be that you can
do away with the reliance on elaborate taxonomies and
categories, if you have something to replace them that
works better. My suggestion is that modern search tools
may be that better tool. It’s certainly better
than a coercive or punitive system that tries to force
lawyers to comply.
RF: I’ll go back again to old
experience because I think it’s still relevant,
in spite of technology advances. In 1989, I used what
was arguably then the best full-text engine (PLS) that
offered relevancy ranking and statistical algorithms
to find related documents (even those not including
search words). I have not seen evidence that the current
crop of full-text and semantic software engines fix
the problem of under and over inclusive search. Also,
I am not persuaded auto-categorization works yet; for
example, Vivisimo, a web-based categorization search
tool, is a great start but I find it’s categorization
weak. And finally, you still face the "garbage
in, garbage out" problem. Don’t get me wrong,
I do think full-text is very valuable. But you also
need categories. If you are trying to find, say research
on statute of limitations or a model asset purchase
agreement, you will just find way too many hits with
full text or auto-categorization.
DK: Let me suggest an even more radical
approach. Why not blow it all away and consider throwing
everything that you have into a big database and focus
on developing custom reports and saved searches using
regular database reporting tools, such as Crystal Reports.
KM for lawyers does not need to be rocket science—a
few slices of the information, as long as they are the
right ones, will give lawyers most of the information
RF: An interesting idea. If the database
were to include descriptive and contextual information,
I would be all for it. But I agree with you that lawyers
will not enter that extra data. Without the fielded
information, I doubt the reporting would work. With
the dramatic growth in electronic discovery and a new
crop of full-text vendors it is spawning, perhaps we
will see some dramatic new software. For now, however,
all my experience says that magic bullets fail—the
solution requires real work, at least some of it by
DK: You’re right, of course.
I’m not sure that even I would go as far as I
suggest. However, I do think about it and wonder if
we are missing the boat by looking for “perfect”
solutions when the “good-enough” might be
all we really need.
RF: Certainly the current high interest
in full-text searching (especially federated searches)
speaks to the “good enough” approach. It
certainly makes sense to have full-text in comparison
to nothing. I think that explains the current interest
in products such as Westkm, LexisNexis Total Search,
Recommind, or Autonomy. The question is whether it makes
sense to go further. That usually requires some amount
of human effort, typically from “practice support
lawyers” or “knowledge managers.”
Many firms have such staff, but it’s an expensive
proposition and quantifying the return is hard. In my
view, firms should at least experiment with more systematic
ways to identify documents and capture context. Take
some small steps and see if they pay off. And compare
that pay-off to other non-billable activities (such
as writing articles or sending clients newsletters)
to assess it fairly.
DK: And we’re now looking at
the tip of the coming data iceberg. What happens when
IM, RSS feeds, audio and videoconferences, collaborative
workspaces, Webex, and other things all become part
of the documentspace? Some firms are simply banning
instant messaging. That’s not a good idea if your
business clients routinely use IM.
RF: Good question. Perhaps by the
time that happens, we will have a better handle on existing
document types. Especially with respect to videoconferences
and Webinars, however, use of a taxonomy would help
as it avoids need to generate text transcripts. Answers
may also emerge from what’s happening in e-discovery.
DK: Lawyers have proven that they will
not change the way that they work. For better or worse,
finding ways to work around them or taking them as you
find them have more likelihood of success than anything
does that involves a behavior change. In a new firm,
though, there might not be the same story.
RF: That may be true but it's bad.
Would you go to a doctor who refuses to change how she
works? How about if, relative to a now questionable
diagnostic or therapeutic regime, your doctor said "Well,
this is what I learned in my residency and it's always
worked fine for me. Why should I care what other doctors
do—I'm a professional and I know what I'm doing."
Isn't that the moral equivalent of what lawyers are
doing? U.S. business has spent the last 20 years re-structuring
and developing new and more efficient and effective
processes. Why do lawyers think they are different and
why does the market insulate lawyers from these pressures?
DK: Those questions, and some similar
ones, are the $64 billion dollar questions for the legal
profession. It comes down to this, “Is the legal
profession different?” What makes us so sure that
we are immune to the changes that technologies and the
Internet have brought to every other type of business?
For me, I don’t see how we can conclude that we
are immune. It is only prudent to make plans based on
the changes we see elsewhere. What Ron and I have discussed
are some practical ways to take advantage of some of
the lessons we’ve learned from document management
experience to enhance the likelihood of success in KM
projects. We’ve learned a lot and we need to apply
it. If you can start with the goal of making the system
work with lawyers rather than against them, you increase
your chances of getting great results.