Knowledge Specialization #taxonomy #expertise


Stan Garfield
 

A friend of mine who writes a leading economics blog posed the question below to me.  Lee Romero provided his thoughts (see below).  What other insights do you have?

Regards,
Stan

---------

From: Arnold Kling <akling4378@...>
Sent: Sunday, November 23, 2008 5:45 PM

I am interested (a bit) in the phenomenon of knowledge specialization.  For example, in medicine, there are many more specialties and sub-specialties than there were 30 years ago.  My guess is that if libraries are still using classification systems, there should be a lot more categories.  My guess is that major universities have many more departments than they did 30 years ago.

I think this is important in economics because I think that businesses and economic systems have become harder to manage as a result.  In short, the leaders tend to know less about the specialized information that is further down in the organization, because the amount of the latter is increasing (I conjecture).

I would like some quantitative indicators of the rate at which new knowledge categories or sub-categories are being developed.  Do you know how to even go about searching for such indicators?

Arnold Kling
http://arnoldkling.com
http://econlog.econlib.org

---------

From: Lee Romero <pekadad@...>
Date: Tue, Nov 25, 2008 at 8:45 PM

Here are some thoughts:

* First -I think would be useful is to split the idea into its parts -
there are really two theories in there that would need to be validated
and then correlated - First, that knowledge specialization is
increasing in a meaningful, measurable way and second, that this
increased specialization is somehow impacting the management of
businesses and economic systems.

* For the first question - My first reaction when reading through
Arnold's email was to look at SIC codes and (if possible - I don't
know SIC codes and their management process in detail) try to get a
picture of how they've changed through the years (specifically, how
much more refined they've become).  SIC codes don't really correlate
to "knowledge domains" but they could provide an indicator in terms of
how companies describe what they do.

* However, even though SIC codes get quite detailed at a macro level,
I don't think they get quite detailed enough to really shine much
light on this particular question.

* Another thought that might provide some hints would be to try to get
some insight on the keywords applied to research papers through the
years (using one of the many on-line systems that provide access to
abstracts for example) and try to determine if the range of keywords
used to describe research papers is increasing.  Again, this would
seem a bit indirect but it could support the argument.

* That idea is pretty similar to what Arnold suggests with regard to
examining classification systems used by libraries - that could be
another angle on this same idea, though - have the primary ways to do
that (Dewey Decimal System or the Library of Congress Classification
System really gotten "larger" or is the continued detailing of those
systems more a reflection of a refinement of understanding? Just
because someone added a new label does not necessarily imply that it
covers a "quantum of knowledge" that would be equivalent to other
labels, but, instead, just a recognition that a new label is needed.

* Similarly (and obvious, though perhaps a bit challenging to do),
following up on one of Arnold's theories - it could be useful to do a
review of departmental organizations within major universities and try
to measure some kind of increase in specialization.  To reduce the
effort of this, perhaps identify the top schools in particular domains
and focus on those (the idea being that the top universities in, say,
medicine or engineering would likely have the most need for
specialization).  Of course, that could then miss the "long tail"
where deep increases in specialization might be more apparent?

Some thoughts on the second question:

* One anecdotal thought that could be interpreted as supporting
Arnold's theory of the challenge of managing could be the rise of
communities of practice in organizations - the idea being that the
more strictly hierarchical organization that was (perhaps) more
prevalent has broken down because it is (perhaps) common now that a
manager does not have the specialized knowledge to usefully mentor
his/her employees with the broad variety of specializations they might
have - whereas the informal organization of a community of practice
enables people to cross organizational boundaries but still be bound
by the specialization in which they work.

* (On the flip side, the "flattening" of many organizations could
argue for that same thing as well - as organizations get flatter,
people have a larger management responsibility and so less of an
ability to really be specialized enough to know everything to mentor
team members...)

* I guess the challenge with regard to the second question is really
providing some kind of evidence (neither of above do so) that
businesses have become harder to manage - in what sense is the
management getting any harder?  Does that mean that the work of
individual managers is getting more challenging?  Or is it a question
more targeted at management in the bigger sense of company failures
(i.e., we believe management is getting harder because the # of
companies going under is increasing)?

Well, there are a whole lot of words and thoughts, though I'm not sure
any of them are all that useful in helping Arnold.

Regards
Lee Romero


Matt Moore <laalgadger@...>
 

Stan,

That's very interesting set of emails. I wonder about the species metaphor from biology. From my understanding, different species can no longer breed and produce fertile offspring (biologists correct me please). The specialisation of expertise is only an issue when experts can no longer communicate with each other - or understand each other's research.

I wonder if a network mapping approach might be appropriate here (hello Valdis). If you take the academic world, then if authors are publishing with each other then presumably there is some interchange. If they aren't then presumably there isn't. So say you take a particular field (e.g. economics, electrical engineering) and you identify the journals in your field. Looking at networks based on co-authorship over time, has the network (or networks) become more fragmented, have more cliques and sub-groups emerged?

Other approaches (that spring to mind for the academic world):
- Total number of books published in a discipline and the average number of copies bought (very crude)
- No of journal titles in a discipline (also crude)
- No of management related professions / professional associations (yip, crude)

As Lee notes, there are 2 parts to Arnold's question:
- Is there greater knowledge specialisation / fragmentation? The gut-feel answer is yes but it's difficult to quantify (and economists love quantification).
- Is this a bad thing? Well yes and no. Adam Smith talks about the advantages of a division of labour. Knowledge specialisation is not necessarily a bad thing. And there are many other issues that prevent senior managers from understanding what is going on in the trenches (status/arrogance on their part and deceit/information filtering on the part of their subordinates being just two).

Perhaps the issue is whether managers can make sense of conflicting information in complex situations or whether they can delegate effectively to their subordinates.

Cheers,

Matt


Peter Marshall <peter.marshall@...>
 


Valdis Krebs <valdis@...>
 

Thanks for the heads-up Peter & Matt!

I have done quite a bit of work lately in "co-authorship" networks -- who writes/publishes with whom? This type of analysis is especially hot with pharmaceutical firms -- they want to know who the thought leaders/connectors are in various medical fields. I can't share client data/analysis, but here is a simple map of co-authors in the field of "social network analysis" -- a network of the network gurus.

http://www.orgnet.com/SN.html

Double-click on any node(author) to see more publications by author via Google Scholar. This map does not show metrics, but that is not hard to do -- we often find "mavens" and "connectors" in these maps using various network metrics. In a large organization we could look at node attributes such as location, department, division, etc. and see how much cross-functional (cross-silo) interaction/publication there is. We have a specific metric for this that can be tracked over time to see if interchange/intersection is inc/dec.

Valdis Krebs
http://www.orgnet.com
http://www.thenetworkthinker.com

On Nov 26, 2008, at 9:29 AM, Peter Marshall wrote:

Matt -- I really like your suggestion of analyzing whether this is a real problem by looking at the extent to which experts from separate specializations are publishing work together.. Looked at over time, one would hope that cross-fertilization growth counterbalanced the growth in specialization (speciation).

I was thinking about a response suggesting that "emergence" of new disciplines and synthetic categories was the natural counterbalance to specialization, and that this balanced evolution has always been happening. Altho info technology has no doubt sped up specialization, it certainly could also speed up synthesis.

Even aside from academic research, there are tons of current books that bring together insights and patterns of thinking across disciplines, Gladwell's as one example. My sense is that bemoaning specialization is like bemoaning how kids these days are (fill in something bad)... It's only one side of a story which has been going on forever, and whose other side (kids grow up and all that youthful energy and creativity fuels the cultural evolution) is the "answer".

I think your suggestion would allow us to measure this. Valdis?

Peter


On Wed, Nov 26, 2008 at 12:00 AM, Matt Moore <laalgadger@yahoo.co.uk> wrote:
Stan,

That's very interesting set of emails. I wonder about the species metaphor from biology. From my understanding, different species can no longer breed and produce fertile offspring (biologists correct me please). The specialisation of expertise is only an issue when experts can no longer communicate with each other - or understand each other's research.

I wonder if a network mapping approach might be appropriate here (hello Valdis). If you take the academic world, then if authors are publishing with each other then presumably there is some interchange. If they aren't then presumably there isn't. So say you take a particular field (e.g. economics, electrical engineering) and you identify the journals in your field. Looking at networks based on co-authorship over time, has the network (or networks) become more fragmented, have more cliques and sub-groups emerged?

Other approaches (that spring to mind for the academic world):
- Total number of books published in a discipline and the average number of copies bought (very crude)
- No of journal titles in a discipline (also crude)
- No of management related professions / professional associations (yip, crude)

As Lee notes, there are 2 parts to Arnold's question:
- Is there greater knowledge specialisation / fragmentation? The gut- feel answer is yes but it's difficult to quantify (and economists love quantification).
- Is this a bad thing? Well yes and no. Adam Smith talks about the advantages of a division of labour. Knowledge specialisation is not necessarily a bad thing. And there are many other issues that prevent senior managers from understanding what is going on in the trenches (status/arrogance on their part and deceit/information filtering on the part of their subordinates being just two).

Perhaps the issue is whether managers can make sense of conflicting information in complex situations or whether they can delegate effectively to their subordinates.

Cheers,

Matt





--
Peter Marshall
-------------------------------------------------------------------------
Email: peter.marshall@gmail.com
Mobile: (949) 689-7000
Skype: ideasware
GTalk: peter.marshall