Between Taxonomy and Typology

1   The War Between Taxonomy and Typology

David Weinberger presents a conundrum in a brief piece in KMWorld magazine, “Who Cares About Knowledge?”  Is knowledge distinct or is it indeterminate?  Is it orderly or messy? Is it true belief, or just ideas and opinions?  Is knowledge embedded in content?  Or is it free floating in a web of human relationships? Weinberger concludes that “knowledge is becoming an old-fashioned term.”

Is he implying that we don’t need to talk about knowledge anymore?  That it is a term of art, of use only to specialists (aka knowledge managers)?  Perhaps some other idea, say “memes,” will supplant knowledge?  I’m sorry but I think I’ve heard this argument before.  This is just another skirmish in a long running battle. To use another old-fashioned term, it is the latest round in a war between Taxonomy and Typology.

2   The Knowledge Battleground

The starting point for this thesis is a blog by Dave Snowden titled “Typology or Taxonomy.”  Snowden cites as a reference an excellent paper by Kevin Smith, “Typologies, taxonomies, and the benefits of policy classification” in the September 2002 edition of the Policy Studies Journal.  Snowden describes a situation from his knowledge management consultancy work at IBM where adherence to an ingrained taxonomy led the company awry.  He concludes: 

The message is very simple, rigid boundaries have huge value in static situations so taxonomies work.  But where things are subject to rapid change and the possibility of encountering novelty is high, they [taxonomies] are plain dangerous.  However we do need constructs to make sense of the world and that is where conceptual frameworks, or typologies, come into their own.

It is clear that Snowden favors typology and is exceedingly reluctant to accept taxonomy as a guide.  He consigns taxonomies to the Simple domain, and finds them unsuitable for the rapid change of the Chaotic and the novelty of the Complex.

I’m not sure that I can agree with this proposition, as knowledge forms from the constant interplay between structured information and the amorphous, even formless, mass of data we are immersed in.  If we are to accept that ambiguity is a principle component of fragmented knowledge, then we must take a closer look at taxonomies and typologies and their interactions.  The patterns that are found in one frame of reference are different from patterns that are prevalent in the other.

3   Systems of Meaning

It is tempting to consider that our patterns are degrees of order, and to try and map them through the Cynefin framework. That approach leads us into the "I-space" described by Max Boisot.  However, "I-space" is too confining, and does not account for certain attributes of complex knowledge that are essential to understanding this puzzle.

(3.1)    Human systems are mutable – they are stable for a long while, until they suddenly change.  Stability is a property of the system, which suggests predictable behaviors in the ordered domain (simple and complicated) and probable ones in the unordered complex domain.  We will see that the form of the system is an emergent property based on how much information was available at its origin.

(3.2)    As we construct a taxonomy or a typology, we create a system of meaning.  As a novice, by the time we can talk cogently about a subject we are enmeshed in a taxonomic system of meaning.  The same holds true for experts constructing theories, for all theories are typologies.

(3.3)    So what we face is the continuous interplay of systems of meaning.  Patterns are more apparent when contrasts are strong, that is, when conflicts occur vice agreements.  When conflict happens, it will most often occur at the boundaries.  What we look at and think about human systems, what we see most often are the borders of systems.  That systems are constrained by their boundaries is inherent in the definition of all systems.

(3.4)    This isn’t to say that there are no conflicts over principles, which are the operating rules of a system of meaning.  There are plenty of conflicts over principles!  Still, it is genuinely hard to tell which type of conflict we are seeing: edge or inner.  The disagreements on the edges, the rubbing of boundaries, are conflicts that won’t change the running system inside.  The result is lots of noise and smoke, and infrequent change, which is what we see in all communities and social networks.

(3.5)    Human systems are consistent, although that consistency may or may not be coherent.  Each transaction that emerges from disorder drives the system towards consensus or coherence.  The transition itself forms a dampened oscillation, along with a shift in the community towards an empirical proof, or else towards a satisfaction that is fallible, the new consensus.  See Rotate-45-degrees-and-think-anew for an illustration of this process.

(3.6)    We see the conflicts more clearly because the boundaries of systems are more visible than the rules and principles of behavior.  Thus transgressions of the rules may be more shocking because they occur in unexpected contexts, nearer the heart (center) than the edge.

(3.7)    Fragments slip through these boundaries.  Stories move without difficulty as carriers of fragments and the providers of context.  In this sense of carrying, stories are dis-intermediated.  However, stories are more limited than fragments that are obviously “good” or “evil” or noticeable in some way.  Stories are filtered by culture, and what types of stories are acceptable are dependent on the norms of storytelling and the rituals of listening.  There are many more constraints on stories than on fragments.

4   Taxonomy

Taxonomy starts with scarce information and some observations or empirical evidence. We then search for some alignment of the evidence, producing an ordered set, which is new information.  When we build out taxonomies we aggregate information, and leave behind the scarcity that we started with. 

A taxonomy … classifies things based on clear empirical characteristics and will have rules that allow determination of location. They have clear boundaries … On the downside, once a taxonomy is established if something does not fit, it will be made to fit as the taxonomy itself creates a filtering mechanism through which we filter observable characteristics.  Kevin Smith

(4.1)    We can see the emergence of a system of meaning from praxis of a profession.  Practice is experience – an increase in knowledge - and eventually there is recognition, with or without titles, that you have knowledge and are an expert in some domain of information.  Within an organization, work leads to the formalization of a role, or even an office (organizational unit) as the system.  These patterns persist in the norms and cultures long after the condition of sparse information has disappeared.  What has emerged is a system of meaning where identity and knowledge grow together.  This is how the practice of “knowledge management” formed and how it grew into a recognizable domain of information.

(4.2)    Management is in its essence the making do with limited resources, inadequate time, and other severe constraints.  To talk of managing is to use language and metaphors of making do and accomplishing things when resources are limited.  It makes sense to talk of knowledge management when information is hard to come by, and when knowledge is scarce.  As we depart the condition of scarcity, we can still talk with our companions in the organization with the shared management concepts and terms that we grew up with.  KM makes sense in scarcity.  KM doesn’t work in a framework of abundance because management does not have terms and language for a world without constraints.

(4.3)    An autopoietic system is a system that grows and renews. (wikipedia/wiki/Autopoietic) We have just seen how taxonomy forms such systems from praxis.  Without the impetus of growth, the system destabilizes, turns chaotic, and decays into the confusion of disorder.  The system may also devolve into a simple and stable form that is superstition.  These systems of meaning are self-sustaining, unchanging, and have legitimacy in an information-is-sparse sense.  Conspiracy theories are one example of the devolved form of taxonomy.

5   Typology

Typology is possible when we have ample information. Typology also holds when there is more than enough and often far too much information.  We construct theories of various combinations of information and then filter by some measure of importance.  Filtering discriminates and excludes information, eventually reaching a balance that is the theory. 

In a typology the dimensions represent concepts, they do not necessarily exist in physical reality (although they can). As such typologies generate heuristics which are more adaptive under changing circumstances. On the downside the concepts can be arbitrary, may not be exhaustive and can easily be subject to clashes of interpretation.  Kevin Smith

The typology scenario differs from taxonomy because of the condition of ample information where disruption is far more pronounced.  I’m certain that systems of meaning will form, but the nature of these systems is not as clear.  It is much harder to isolate persistent patterns, and we will find a variety of system forms when there is an abundance of information.

(5.1)    For typology in the extreme, the condition is that there is too much information – a super-abundance.  Discordant information is the norm.  Snowden’s property of coherence is a necessary condition to attaining stable patterns. It isn’t a sufficient condition, as these patterns may be more like islands in the stormy sea than emergent, self-sustaining social systems.

(5.2)    We need look no further than the IT Department to find one example of the organizational form.  The practices of work may change fairly quickly, but the norms and culture of the group serve as memory that evens out the information flows.  When there is too much information, we must filter to reduce the level of discordance.  Filtering avoids disruption by creating compartmentalization or specialization.

(5.3)    Memory fails in larger groups as communications media do not scale effectively.  Diversity is lost because it is outvoted, it is not in the mainstream, and it is not loud enough to be heard in the Echo Chamber of the media.  Even with the best social media technologies, loud voices and hyper partisanship will drown out the diverse and obscure.  Consequently what forms is the notorious silo.

(5.4)    Silos exist to preserve order, because they are social structures that make knowledge simple and complicated (the ordered domain).  The primary characteristic of order is equifinality, therefore silos are open systems. (wikipedia/wiki/Equifinality)

(5.5)     Without silos the knowledge that the organization holds is at best complex and chaotic (the unordered domain).  It is an open question as to whether we can have an organization that is completely disordered and thus holds no knowledge.  I expect that any organization would collapse and disband well before the truly disordered stage, as smart people decamp into other organizations or communities.

(5.6)    An open system can easily devolve into conflict.  The conflict pattern is a situation where opposition becomes the operating principle for the social system.  The focus of the system becomes obstruction in all things, to the detriment of coherence. On the surface the conflict pattern seems similar to the silo and it’s equifinality, in so far as conflict is the emergent property.  Conflict breeds conflict; consider the so-called Law of the Jungle: “kill or be killed.”  On deeper inspection we will find that the language of the conflict pattern is one of provocation and escalation, and the conflict pattern hardens into a conflict system of meaning that is autopoietic, not open.

(5.7)    When there is too much information available, a group can reach consensus without coherence.  They can cherry-pick from the facts and construct the simple theories that we call fantasy.  This is another variation of the open (equifinal) system of meaning.

6   Balancing

To distinguish system forms in the case of ample information we need to examine an organization’s culture and its reaction to incoming (additional) information.  How does it react?  Does it avoid discordant incoming information?  Avoidance shows us a closed system.  Redirection and deflection are likely the hallmarks of a complex adaptive system - we can see that some change or some work (information processing) actually occurs in the handoff.  As noted in the definition of systems of meaning above, it isn’t clear whether any transaction impacts the system at a larger scale.  However, if we consider the interaction of information and constraints that bear on the system of meaning, we have illuminated the field where the patterns can play out.

(6.1)    When there is too much information, we can see in the equifinal characteristic of open systems the motivation to maintain the status quo.  Here we can see the roots of the antagonism of certain topologists who hunt the KM zombies and the other undead fantasists.

(6.2)    There is the reinforcing pattern of “remembering” that Patrick Lambe has described from his taxonomy work (http://wiki.sla.org/download/attachments/54264068/Taxonomy_KM_Lambe.pdf).  We filter naturally and we down-select based on preferences and aptitudes. Consequently, systems of meaning emerge.  Lambe’s remembering suggests that to change the form of the system there is a rigid set of bad habits to overcome.  We will find there is a particular rhythm that must develop to achieve effective remembering.  It is something that is uniquely found in story circles, campfires, and skilled tellers of stories.

(6.3)    Organizations consist of two or more silos.  Disorder occurs between the silos when they cannot agree and create a common meaning.  Where the silos do agree we will find a complex adaptive system where dialogue occurs.  The CAS system of meaning is one where taxonomy and typology are balanced and rapid change is possible.

(6.4)    There is one other case to include: exaptation.  Bricolage is less ordered, and less structured, than an open system; it takes more energy to cope with the abundance of information.  Bricolage is less safe than an open system, as the primary safety net of equifinality is missing.  But safety is not the operative motivation here, for if you want safety then fantasy is the easiest world to reach and inhabit.  Remember fantasy is the devolution of typology, and superstition is what devolves from taxonomy.  Bricolage is of course more familiar as pragmatism, the seeking of small gains from the system that one resides within.  Dave Snowden suggests that diversity may be key characteristic of the adbuctive reasoning of exaptation, see http://cognitive-edge.com/blog/entry/5575/exaptation-managed-serendipity-ii.  Also see Yiannis Gabriel’s discussion of bricolage at http://www.yiannisgabriel.com/2012/08/on-paragrammes-theory-of-organizations.html.

Emergent System of Meaning Types
Taxonomy	Typology
Praxis and learning (growth, autopoietic)	Silo (equifinal, open system)
Confusion (disordered, not a stable pattern)	Conflict (growth, autopoietic)
Superstition (equifinal, open system)	Fantasy (equifinal, open system)
Dialogue (complex adaptive)	Bricolage (complex adaptive)

7   Knowledge Redux

By the time we can talk cogently about searching or filtering, we are already embedded in an information system.  It may be a system of learning or remembering, or one of collaboration; these are stable patterns of information as we can see from Patrick Lambe’s taxonomy practice.  Other patterns of information which we can discover from following Dave Snowden’s principles of coherence are systems of power, of privilege, and of ignorance, both willful and blind.  These too are coherent systems of meaning.

(7.1)    What lies between taxonomy and typology is not a difference of information or even ideas.  What we are really seeing is the clash of systems of meaning in opposition.  Systems have boundaries and that is where the conflict occurs.  We do not often see the conflicts at the center, the heart, which is the ground from which the system forms

(7.2)    In the center of information space is a perpetual collision of systems. Concepts compete at the boundaries, but fragments easily slip through. 

(7.3)    It is only after we have recovered a perspective of the scarcity or abundance that once was in force, that we can finally talk about understanding and knowledge.  This perspective is what I believe Dave Snowden describes as cognition.

(7.4)    Other systems – of economics or of power – may have stronger effect than systems of meaning, and pure examples of systems of meaning may be hard to find.

(7.5)    We do not have a language for superabundance, nor do we have much language for low constraints – when there is little to hold us back from hurting others as we tell “our truths.”  I suspect this is the basis of David Weinberger’s complaint.

(7.6)    Principles may be in conflict, but if so, how can we tell them apart from the conflict over boundaries?  Complex knowledge emerges as the balancing along borders of systems of meaning.  Systems are both constrained and coupled as shown in this framework.  

(7.7)    If we balance well we have satisfied the conditions for social change; if poorly than we create the persistent resistance that we so often see.

(7.8)    Abductive reasoning is tightly constrained within certain dimensions and loosely constrained along others.  Contrast this with deductive reasoning that is uniformly constrained, and inductive reasoning which is generally loosely constrained.  Think of “best practice” and perfection as exclusivity seeking knowledge searches.  They focus on narrow criteria and as a result limit the possibility space they search.  On the other hand, “good enough” and novel need to be inclusive and thus they expand the search space of probabilities.