"Please take this as my 2-week notice." 

Congratulations!  This means we are the manager and the team member did not quit immediately in a huff, need us to fire them, or force HR and corporate security to inform us the results of a background check uncovering an ongoing criminal investigation (I had the last experience with HR calling me about an otherwise perfect employee). Nevertheless, we are left with an impending skills gap and we have at most 2 weeks to fill it.  The time spent finding someone equally qualified reduces the time available for the departing team member to develop the new recruit. Exceed 2 weeks and we are starting from square one.

So what is a manager to do?  Hiring willy nilly is insanely frightening - like a rebound relationship. Prior existing redundancies are expensive luxuries that disappear in tough business environments. Cross developing other staff reduces efficiency - plus runs the risk of the telephone game if the departing member transfers skills to a temporary worker who in turn leaves or is otherwise distracted before transferring them to a new hire. 

The answer is that institutions are built by the books. The manuals. The policies and procedures. The US government is not Congress, the Supreme Court, and the President. (Thank Heavens for that!) It is the Constitution that establishes the policies and procedures that guides, directs, constrains, and controls government actions. A well oiled institution has clearly defined books, policies, and procedures.  A departing employee is a departing friend. But the policies and procedures remain with the institution to guide and direct the new hire in the same manner as the departing member.  The institutional intelligence resides in the books.

But institutions are built not by the books but by the books that people built. People who understood what needed to be done. People who understood how to explain it. People who understood people so as to  explain, to bridge the gap between new hires and skilled team members. A book is inert. An ordinary book on English can have far different meanings to a French speaker vs. a Chinese speaker.  A French speaker for example needs to drop conjugations and gender.  A Chinese speaker needs to add conjugations and gender.  An ordinary book is one to one. An institutional intelligence book needs to be many to many. We have all been to school. Great place. Some students like to highlight their texts. Others like to video record the lessons. Yet others write summaries that emphasize anomalies.  Studying together means understanding and adapting methods to fit our individual students.  Finding the identical new hire match to a departing member may take years. A suitable hire with whom we can adapt, much less.

But at last, institutions are not built by the books that people built but by the books that people build.  Institutional intelligence is continuously adaptive. Always, and especially in today's world with lower barriers to entry and disruptive innovation, a survivable institution finds new opportunities and adjusts to take advantage of them.   Here's to new hires continuing the expansion to greater heights.

This answer is easier said than done.  Developing institutional intelligence to the policies and procedures that remain to guide and grow has a glaring gap.  How do we build that book exactly?  What is it about institutional intelligence, that macrocosm of human intelligence?  How do we understand it, the mechanisms of it, and the institution of it?  What is the art of designing intelligence?

Translating the same concepts of institutional intelligence to individual psychological person terms borrows from the work of Piaget (1954) for early cognitive development.  Piaget's seminal theory asks, "What is intelligence?" and "Where does knowledge come from?" (Siegler and Alibali, 2005).  The theory speaks heavily of qualitative stages of intellectual ability throughout an individual lifetime.  It still forms the basic framework for public schools systems around the world.  But the relevant mechanism of how a person continually adapts and shifts from stage to stage mirrors the above mechanism of adaptive institutional intelligence. The three keys are assimilation, accommodation, and equilibration.

Assimilation anchors environments to the self. It occurs when a person sees a stimulus, say a purple color, and adaptively translates it in terms of what they know, say only either red or blue. 

Accommodation anchors the self to the environment. It occurs when a person grows to adaptively learn that a new color existing between red and blue - purple - is important to keep distinct. 

Equilibration guides assimilation and accommodation by determining when one or the other is appropriate.  It occurs when a person realizes that they are not sufficiently anchored to the environment, say when they declare purple is a red and everyone else disapproves.  The slight embarrassment that ensues causes disequilibrium, causing said person to re-evaluate their anchors, and adjust accordingly to recognize purple as a distinct standalone color.  It also occurs when a person realizes they are sufficiently anchored to the environment, say when they declare purple is red and everyone approves.  Said person understands red and blue as the only important colors.

Equilibration guides the way for triangulating on intelligence.  In artificial intelligence terms, there is a central challenge called the "stability-plasticity dilemma."  Stability means knowledge is constant in stone.  Plasticity means new knowledge can be learned.  They are polar opposites forcing an entity to choose more stability and less plasticity or vice versa.  In machine learning and data science terms, there is a basic "curve fitting stiffness" problem.  A regression graphical summary of the data can be very stiff to resist noise - and possibly show a meaningless average line - or be very loose to capture every value in the curve  - and possibly show a meaningless squiggle.

Stability is to stiffness is to assimilation. This is a classic artificial intelligence script with no reasoning that leads to an, "I am right, you are wrong, and this is the way to do it," attitude that helplessly watches falling market share.

Plasticity is to looseness is to accommodation. This is a classic eidetic memory machine learner, like a k-nearest neighbors algorithm with no reasoning that leads to a, "Whatever you say must be right," attitude that mindlessly imitates and does nothing else.

The trap is in framing the problem only on this axis.  The trap is in even searching for an optimal sweet spot.

The answer is to get off the axis.  The answer focuses on equilibration, the "...keystone to developmental change within the Piaget system." (Siegler & Alibali, 2005).  Equilibration has no real analog in artificial intelligence, machine learning, or data science.  It implies constant communication to balance initiative and emulation.  It implies a clear, concise, complete, and continuous discussion on goals and feedback.  It implies trust and teamwork.  There is a vaguely defined analog in management called leadership.    

It means we do not spend excessive time reacting to find the right new hire with the right accommodation-assimilation mix to write the book.  It means we do not spend excessive resources on a data science and machine learning artificial intelligence team attempting to add more complexity to low level learning, computerized Chess and Go notwithstanding.  Rather it means to proactively lead by clearly defining the goal and continually coordinating with the team to create the impact, scale, and intelligence to get it done.  It means that regarding the books that people build, these people are the leaders.  These are the leaders who understand what needs to be done and understand the people so as to explain it.  

An inert book is a rules set.  Following one by rote is being an automaton or a machine.  Following them faster and more efficiently is being a more effective robot.  The ultimate goal of artificial intelligence and machine learning is to build robots.  Want to see what society looks like with thousands of robots operating perfectly, giving people orders and making people safer?  Go to South Africa.  They have friendly traffic robots at every street corner.  These can automate and replace people in doing menial, repetitive, dangerous, or boring tasks.  

An institutional intelligence book by contrast is a line of communication between team members such that whoever writes is the manager.  Whoever reads becomes future managers.  This is all for the sole purpose of generating the scale and impact to accomplish a particular objective.  The institutional intelligence book by definition cannot automate or replace the team.  It can only augment it.  

If we are to find and develop institutional intelligence - that macrocosm of human intelligence - we need to narrow our attention to the people who build the books.  We focus on the leadership process rather than the particular team skills.