|
computers don't "play" chess. A computer isn't playing chess any more than it's commanding a Civil War
army, racing a car, or swinging a light-saber. It's simulating a person playing chess -- and it's doing it by crunching ones
and zeros, not by using intuition or imagination. - Stephen A. Lopez
I don't think that we really have a clue as to the origin of intuition and imagination, and maybe never
will. I believe those qualities are among those that separate us from animals and aren't easily explained in scientific terms.
- Stephen A. Lopez
I think the single biggest difference between a strong chessplayer and an average one is not
based on calculative ability (the ability to see farther ahead, as the media stereotypically portrays it), but that the strong
player has the ability to choose better candidate moves right from the get-go which then leads to strong play. - Stephen A.
Lopez
Let's look at some quotes from books and see if we can synthesize a concept of strategy useful
for computers playing games, such as chess.
My approach is to ask many questions. These questions are designed to make the reader think about concepts
that have possibly many answers. We're going to look at some answers provided in existing literature, but these suggestions
are probably not the final word on the matter.
Strategy involves a lot of thinking, and if we are going to explore strategy we better get used to doing
a lot of thinking.
MBA Fundamentals Strategy (Ference, Thurman, 2009)
Strategy is a forward-looking concept. We desire a future state of affairs which
is better than what we have now. Aiming for this goal, we arrange our resources strategically so that our chances of arriving
at this "future" are improved.
p.4,6"When we engage in strategic thinking, we begin by conceiving of a future state of affairs... that
we find more desirable than the present state... thinking as a strategist... is... the way that we should approach all decisions
and actions... every choice that we make among alternative actions is, in effect, a strategic one."
Our strategy should be based on dynamic elements, because the "world" of the gameboard, the players, the relationships
between or among pieces/ tokens are all changing. We seek to manage a playing field undergoing transformation, and
basing our analysis or understanding on a single snapshot of the position, frozen in time, might not enable us to gain a complete
understanding.
p.28"the pursuit of a strategic vision - a desired future - takes place over a timeframe in which
the initial attributes and forecasts... on which our strategy is based are changing constantly... Thus, strategy is more properly
the study of dynamics rather than of statics."
Our strategy should involve a cycle of generating ideas, testing them, discarding some, and exploring the future that
is likely and promising. We seek flexibility to handle the unknown which lurks just beyond our ability to forecast. We also
need to know how to estimate the risk of NOT exploring a particular branch of our search tree.
p.33-34,35"Strategic thinking involves a continuing conscious, deliberate cycle, grounded in purpose and
aspiration, of visioning, analyzing, planning, executing, monitoring, and adapting where the adaptations based in experience
from the previous cycle become the bases for re-visioning and informing the next cycle... Another positive benefit of strategic
thinking and the strategic process as presented here is that it increases the flexibility of the organization in responding
to changing conditions."
We use our vision of the future to gather information and to steer our search efforts. Our search tree is possibly best
understood as an exploration tree. We explore paths that are interesting, promising, or have a chance of turning
favorable or unfavorable. We might use measurements taken from the present locations/ interactions of the pieces, but what
we are aiming at is the measurement of a forward-looking potential which can be used strategically
to gain a competitive advantage.
p.50"to be a strategic thinker... is to... think simultaneously
in time and space, to be visionary about the future and rigorously analytical about the present. The [strategist] is both
a right-brain and a left-brain thinker who is able to move fluidly back and forth between the desired future and the actual
present, constantly using the stream of information provided by ongoing events to monitor and adapt his/her strategic actions
in pursuit of his/her evolving vision."
A strategist should be first a diagnostician. We use a set of heuristics to reduce the complexity of the world to a manageable
level - this allows us to play with our model of reality and to explore positions which lie ahead.
Think of a mechanic taking a car on a test drive. When he returns he might have a better understanding of the
problem described by the customer and might have several ideas which are ready for exploring with his toolkit. What
"test drive" can we establish for better understanding the position in the game we are playing, which might lead to "ideas"
for exploring?
p.76,77"The most essential skill of the strategist is that he/she is first and foremost an excellent diagnostician...
It is the ability to ask the right questions in the right order and to listen carefully to the answers. It is based on the
integration of knowledge, experience, and insight. Our diagnostic approach, therefore, is composed of a set of heuristics
that provide us with the ability to reduce the seemingly overwhelming complexity of the organization and its environment to
a workable model through a series of carefully chosen and sequenced questions. Depending on the answers to these questions,
and to the further probing that the answer to each question suggests, we are able to form hypotheses, assess probabilities,
and make choices for action."
Strategy is based on diagnosis. Without diagnosis, there can be only the most basic of strategies.
p.88"Diagnosis,
not treatment, is fundamentally what drives the concept (or conceptual level) of strategy."
What guides our exploration of the future in the game we are playing? We might look to our values - these
are present at every choice we make. Perhaps we choose to examine positions in our search tree or choose to terminate
our exploration based on values which come from a "wisdom" derived from our experience.
p.119-120,122"We might think of our values as the ultimate criterion for choice and action, as the basis
by which we assess and judge the acceptability of options... We pursue this or that outcome because we believe it to be the
expression of something that truly matters - that has value for us. We evaluate possible courses of action on the bases of
whether they are consistent and compatible with our core values... If meaningful, our values have a prime seat at the table
in all of our decision-making and a fundamental question to be asked when considering any course of action would be: (If we
do this) Will we be living our values?"
We could develop measurements which enable us to determine whether we are doing (at any particular point in
time) what we should be doing. Are we executing our strategy, or are we just wasting time? How, specifically, would
we know?
p.126,127"we need to define our vision or strategic intent as we set out on our journey in terms that
will allow us to determine at the appropriate time whether or not we have done what we said we would do... organizations...
function more effectively and are better able to maintain focus when the timeframes chosen for visioning and acting are within
'eyesight' - in a range that is psychologically meaningful and that we can 'get our arms around.'... strategic intent should
be stated in terms that are accessible to the senses, that [is], can be seen... in some objective and reprintable way... we
should be able to specify going in - at the beginning - how we intend to assess achievement coming out - at the end [or at
the checkpoint]."
A car has a dashboard which tells us what is going on, and warns us if things are not going well. What performance
dashboard can we develop for our game which tells us the equivalent information? What dials and indicators would it display.
What alarms would sound? What indicator would show us how well we are executing our strategy?
p.191-194"The Balanced Scorecard: A Simple Operational 'Dashboard'... The balanced scorecard is so named
because it effectively balances our strategic intent with our actions while allowing us to measure... our efforts along the
way... It is critical that any measure used in the scorecard be objective, measurable, and repeatable... when executing a
strategy, keep it simple, keep it focused, and keep it clear... the fewer metrics you use, the better. This will make your
scorecard easier to manage, easier to communicate, and easier to track against."
What waypoints should we look for? What obstacles are likely to block our path? How can we use new information
we acquire to change our course to paths which are more likely to guide us to our goals?
p.228-232"What are the major stages/phases that we will have to accomplish along the way?... What obstacles
are likely to get in our way?... What are we going to do to overcome these obstacles?... these three questions constitute
the process of Folding Back... Folding Back is the Third KEY to Strategic Thinking."
What obstacles prevent us from moving directly to our goals? Identifying these obstacles allows us
to focus on ways to remove them.
p.229"we should expect, and we should anticipate, that we are likely to encounter any number of obstacles...
at every step of the way... we aggressively seek to identify and prepare for them. In fact, these obstacles... are the gaps
between our current state and our desired future state that our strategy must speak to and close."
Our strategy should be able to withstand multiple, unexpected shocks, as we collide head-on with an opponent
who likewise is crafting a strategy. What are our weaknesses and how will we protect them from exploitation?
p.231"our strategy... should be robust, that is, grounded in a strong consensus on mission, values, and
vision, and buttressed by rigorous, ongoing analysis... it can withstand major shocks... it should be resilient, that is,
it should have the flexibility and adaptability to be able to question current arrangements and to move quickly in response
to changing conditions"
Have we developed an action plan which can guide us along the path towards our objectives?
p.232"Our
strategic action plan provides the guidance and direction for implementing our chosen strategy. It specifies, step by step,
who has to do what when in order to move along the timeline - along the intended strategy... our strategic action plan can
and should focus on defining the actions and decisions that need to be undertaken to achieve the first milestone - to overcome
the initial obstacles."
How can we possibly approach the concept of a machine constructing a plan? Let's look at quotes from a classic text and
use the words of Darwin to suggest how a machine can evolve a plan over time by exploring the consequences of promising moves:
On the Origin of Species, Darwin
What if we constructed a planning model similar to the plan of nature in evolving a species - that of identifying
the few promising moves and letting them compete with each other for limited machine resources over time. Only the fittest
moves would survive, and the future positions that result might be more capable of competing with our opponent.
The survival of the fittest move, will then be the "choice" made by our machine:
p.73,74"A struggle for existence inevitably follows from the high rate at which all organic beings tend to increase...
as more individuals are produced than can possibly survive, there must in every case be a struggle for existence, either one
individual with another of the same species, or with the individuals of distinct species, or with the physical conditions
of life... Although some species may be now increasing, more or less rapidly, in numbers, all cannot do so, for the world
would not hold them."
It will not be possible to determine for certainty what the winning chances of each "promising" move are, ahead of time.
But we can simulate the evolution of the position over time and reduce, but not eliminate, our uncertainty.
p.79,81,82"Many
cases are on record showing how complex and unexpected are the checks and relations between organic beings, which have to
struggle together in the same country... I am tempted to give one more instance showing how plants and animals, remote in
the scale of nature, are bound together by a web of complex relations... In the case of every species, many different checks,
acting at different periods of life, and during different seasons or years, probably come into play; some one check or some
few being generally the most potent; but all will concur in determining the average number or even the existence of the species"
We might task our machine with the scrutinizing of the slightest variations in how our present position can develop over
time. We reject those positions in which our opponent is able to increase his positional advantage over us, and instead we
further explore those positions in which our struggle yields positions where we have the greater opportunity for advantage.
p.91"It
may metaphorically be said that natural selection is daily and hourly scrutinizing, throughout the world, the slightest variations;
rejecting those that are bad, preserving and adding up all that are good; silently and insensibly working, whenever and wherever
opportunity offers, at the improvement of each organic being in relation to its organic and inorganic conditions of life.
We see nothing of these slow changes in progress, until the hand of time has marked the lapse of ages, that we see only that
the forms of life are now different from what they formerly were."
The "offspring" mentioned by Darwin might instead be the future positions which result from the promising moves.
Our task then is to construct a heuristic which identifies the positions which are "promising", and a method which simulates
the competition with our opponent over time.
p.135"if variations useful to any organic being ever do occur,
assuredly individuals thus characterized will have the best chance of being preserved in the struggle for life; and from the
strong principle of inheritance, these will tend to produce offspring similarly characterized. This principle of preservation,
or the survival of the fittest, I have called Natural Selection."
Now let's look at another book:
Dynamics of Software Development (McCarthy, 1995)
McCarthy uses the concept of "scouts" in order to explore the future for his software organization. Scouts
can help us to "get good at change". Movement holds the potential for adaption - how can we become good at adaption?
p.36,38"From time immemorial, whenever significant groups of people have been on a dangerous journey, they
have sent scouts forward to check out the future... There is no end to the utility of scouts... The basic idea, one I'll come
at from several directions in this book, is not to slow the pace of change, or to create more stability; but to get good at
change, at managing technology in motion... Movement holds more potential for adaption and self-actualization than rigidity.
Scouts are used only by a people that are under way, in motion, migrating. If you are standing still, you don't need scouts."
What you don't know can hurt you. How can we organize our search efforts so that we don't get blind-sided
by the unknown? What paths do we need to explore, and which paths can we explore later - if we have more time?
p.99-101"It is essential not to profess to know, or even seem to know, or to accept that someone else knows,
that which is unknown. Almost without exception, the things that end up coming back to haunt you are the things you pretended
to understand early on but didn't... Bear down on the team until they realize that they haven't comprehensively assessed the
unknowns... Your job is to make uncertainty an unshakable fact, and then to coerce the reshaping of the organization to cope
with the uncertain situation."
How can we make our strategy visible? American football coaches often draw maps of plays with X's
and O's representing the players, and curved arcs indicating motion. A bracket might indicate that a player moves to contain
another player. If we made our strategy visible, what would it look like?
p.111"The product must be made visible. You have to see what you are doing."
We might want to create an adaptive search function that develops "ideas" and explores the consequences
of these ideas. Perhaps this is how humans play games. We might desire our machine to generate and validate ideas as part
of its strategy. How would we do this?
p.169"What you need for truly creative change, then, is an environment that transcends good health, an environment
that not only accepts a continuum of change, which is normal, but one that positively engenders, nurtures, and propels forward
wholly new dynamics. The transcendent organization values radical or revolutionary change and esteems utterly new modes of
thought. It's possible for a team to be healthy and not especially creative, but this state of affairs is not especially desirable.
What is desirable is team fecundity, the radiating of the new and the original from the normal and the healthy. This
kind of creativity requires a flexibility and a courage beyond the reach of most of us most of the time."
Frank Herbert, the author of the science fiction book Dune, establishes a connection between understanding
and dynamics. Our understanding of a dynamic process, such as game, must extend into concepts which relate to the ability
of the player to adapt and change in an uncertain future.
"A process cannot be understood by stopping it. Understanding must move with the flow of the process,
must join it and flow with it." - Frank Herbert (1920 - 1986), Dune (First Law of Mentat)
We must understand that the paths to our goals are going to be blocked by obstacles. What are the obstacles
we are facing?
"If you can find a path with no obstacles, it probably doesn’t lead anywhere." - Frank A. Clark
Julio Kaplan, Chess Life & Review
Julio Kaplan has this to say about strategy for chess. How can we develop a heuristic which allows us to measure these
concepts?
p.370"in practice the ability to calculate variations, to see a few moves ahead, seems by far the most important
attribute of the good chessplayer. Yet it is my contention that the two, abstract planning and concrete calculation, are really
closely tied together; that in one as in the other the most important thing is the ability to focus one's attention on the
few relevant moves, and dispose of the rest quickly and confidently. - Julio Kaplan, July 1978, Chess Life & Review
p.561"how much is your opponent really threatening? It is hard to give a dispassionate
answer to this question when his whole army is coming at you, but it is essential that you form an estimate of the strength
of his attack... try to get an objective measure of the strength of his attack. This will be very, very hard at first; but
the effort will make you find better moves" - Julio Kaplan, October 1978, Chess Life and Review
The pieces on a game board are a kind of complex adaptive system. What does this exactly mean?
Complex Adaptive Systems (Miller, Page, 2007)
We might form a model which is based on the game pieces and their interactions. This model might
allow us to unravel the complexity and help us to identify promising moves.
p.11"Regardless of the approach, the quest of any model is to ease thinking while still retaining some ability
to illuminate reality."
Our model should ideally allow us to understand the situation and serve as a predictor for forecasting events which
are going to happen or are likely to happen.
p.39"The success of a particular model is tied to its ability to
capture the behavior of the real world."
In natural systems, the static condition happens only at death. A natural system is dynamic and alive and our model should
capture this degree of "aliveness". We want to avoid the condition where we want to understand running water and resort instead
to catching it in a bucket.
p.83"Many of our existing analytic tools avoid an emphasis on dynamic processes and focus on equilibrium
states... In natural systems, however, equilibria are usually associated with the death of the system. The conditions that
favor equilibrium analysis are likely the exception rather than the rule in many complex adaptive social systems. If so, the
techniques that we traditionally use to analyze these systems may be like trying to 'understand running water by catching
it in a bucket.' "
Von Neuman and Morgenstern propose that dynamic analysis is the way to approach understanding, but that it must have
as its foundation, statics.
p.84"Social scientists have often recognized the importance of dynamic analysis but
have been very constrained by their tools. According to Von Neuman and Morgenstern (1944, 44) [JLJ - Theory of Games and Economic
Behavior], in their seminal work on game theory, 'We repeat most emphatically that our theory is thoroughly static. A dynamic
theory would unquestionably be more complete and therefore preferable. But there is ample evidence from other branches of
science that it is futile to try and build one as long as the static side is not thoroughly understood.' "
Our computer must ignore massive amounts of information in order to focus on the events that are promising and critical.
Perhaps the use of summarizing metrics, much like the daily announcement of the Stock Market Index, is what we are after.
Can we possibly develop a metric which indicates how promising a position is, in a game?
p.94,95"Given
the limits of information processing, agents must actively ignore most of the potential information they encounter... We suspect
that a full analysis of how agents selectively attend to information will provide some interesting scientific opportunities...
agents may develop ways, such as statistics, to summarize the flow of incoming information so that it is easily stored and
used... agents may... influence the actions of others... the reality is that most agents exist within a topology of connections
to other agents, and such connections may have an important influence on behavior."
We should aim for a simple model, even if the underlying behavior is complex.
p.101"Thus, even when the resulting behavior is complex, the underlying model should be simple."
Our model should have just enough of the right elements. Perhaps the construction of this kind of model is
more art than science.
p.246"Making sure that your model has just enough of the right elements and no more is the most fundamental
practice for any kind of modeling, and computational work is no exception... Scientific modelers must aim for... simplicity
and clarity. Modeling is like stone carving: the art is in removing what you do not need.
It is often easy to recognize
a simple, well-formulated model after the fact, as such models have strong intuitive appeal... Good models strip phenomena
down to their essentials, yet retain sufficient complication to produce the needed insights."
How should we judge the model we develop, which attempts to give our computer an understanding of what is
going on in the game?
p.254"Scientific judgments in this area [computational modeling] should focus not on the computer per se,
but on the quality and simplicity of the model, the cleverness of the experimental design, and the new insights gained by
the effort."
The Essence of Artificial Intelligence (Cawsey, 1998)
How exactly do we decide what to do when we search for the best move in a game? We might aim for an approach
that is similar to the way humans search for a move.
p.77"Where the search space is too big to search every node it may be possible to construct some scoring
function that can be used to provide an estimate as to which paths or nodes seem promising. Then the promising nodes are explored
before the less promising ones. Search methods that use such a scoring function are referred to as heuristic search
techniques.
The basic idea of heuristic search is that, rather than trying all possible search paths, you try
to focus on paths that seem to be getting you nearer your goal state. You generally can't be sure that you are really near
your goal state... But we might be able to have a good guess. Heuristics are used to help us make that guess."
Crucial is the concept of an evaluation function, which might allow our machine to determine the
moves that are promising and might contribute to the understanding of what's going on in the position:
p.77"To use heuristic search you need an evaluation function that scores a node in the search tree
according to how close to the target/goal state it seems to be. This will just be guess, but it should still be useful."
Let's now look at a book on economics. Economics and game theory are similar in that the future is unknown
and possibly unknowable. Perhaps we can aim at influencing the future, rather than worrying about not being able
to predict it.
The Origin of Wealth (Beinhocker)
Complexity arises out of interactions among networked people, agents, or even game pieces on a board.
p.141"Networks are an essential ingredient in any complex adaptive system. Without interactions between
agents, there can be no complexity."
Here we have the concept that economics is a kind of survival of the fittest. The competitive nature of
business drives the weak (and their weak strategies) out of business. Perhaps in game theory it is the survival of the strongest
strategies which drive the developmental evolution of the current position in a game.
p.187"Businesspeople, journalists, and academics all gravitate quite naturally to using images of ecosystems
and evolution when they speak about the economy. One of the strongest claims of Complexity Economics is that this language
is no mere metaphor - organizations, markets, and economies are not just like evolutionary systems; they truly, literally
are evolutionary systems."
Perhaps we can understand the complexity of a position through a process like evolution where we examine
the promising moves and the likely outcomes. The "unpromising" moves will not survive in our simulated evolution. The best
move then has the highest potential in the evolution of the current position.
p.187"evolution is not just about biology. Rather, evolution is a general-purpose and highly powerful recipe for
finding innovative solutions to complex problems. It is a learning algorithm that adapts to changing environments and accumulates
knowledge over time. It is the formula responsible for all the order, complexity, and diversity of the natural world... it
is the same formula that lies behind all the order, complexity, diversity, and, ultimately, wealth in the economic world...
Daniel Dennett, an evolutionary theorist... calls evolution a method for 'creating design without a designer.' "
Biological evolution gains efficiency by working in parallel, with multiple promising designs. The offspring
of these promising designs that survive to produce offspring are somehow a slightly better 'fit' with the environment.
A machine playing a game can simulate this biological process and 'discover' the moves (and the resulting positions)
that initially were only 'among the promising', but later, through tree searching, were discovered to offer a definite advantage,
when tested with the 'promising' moves from our opponent. In effect, we have a mechanism for 'evolving' an effective
plan. The only requirement is that we are able to produce good 'candidate' moves, from a given position.
p.216"In effect, evolution says, 'I will try lots of things and see what works and do more of what works
and less of what doesn't.' ... The algorithm learns what the fitness function 'wants,' ... the evolutionary process generates
novelty as it searches for fitter and fitter designs."
We might not be able to predict the future, but we can choose moves which create positions where we are
resilient and have potential for evolution towards our goals.
p.324"We may not be able to predict or direct economic evolution, but we can design our institutions and
societies to be better or worse evolvers."
Here is how we might define strategy, in the context of game theory:
p.324"Strategy can be defined as the determination of the basic long-term goals and objectives of an enterprise,
and the adoption of courses of action and the allocation of resources necessary for carrying out these goals... First, strategy
is inherently forward looking. To develop a strategy, one must make a determination about where one wants to be in the future.
Second, strategy is about creating a plan for getting to the desired future state and committing to a course of action defined
by that plan."
Our "plan" might be this: we develop a "portfolio of experiments", much like a venture capitalist. Some
of our experiments will fail. Others will succeed, and we can expand our experimental efforts once we have identified the
successful candidates.
p.334"Rather than thinking of strategy as a single plan built on predictions of the future, we should think
of strategy as a portfolio of experiments, a population of competing... plans that evolves over time."
What is required in order for us to create our "portfolio of experiments"?
p.337"Constructing a portfolio of experiments requires a collective understanding of the current situation
and shared aspirations... processes need to be established that enable the amplification of successful ... Plans and the elimination
of unsuccessful Plans."
Our "evaluation" which we use to develop these experimental plans might be constructed this way. Think of
Thomas Edison in his lab. He discovered many thousands of things that did not work, along with the thousands of things that
did work. He used a theoretical foundation, a box of junk, and a collection of ideas. Edison "knew" which approaches were
the promising ones, and he explored those first. If those didn't work, he tried the next most promising.
p.343,344"the aspiration must capture an important insight about the selection pressures that the outside
world is subjecting the organization to... one must strike an important balance in formulating an aspiration. It needs to
be specific enough to provide selection pressure, but not so specific as to require the ability to predict the future... a
good aspiration provides a powerful motivating force to keep the company in constant motion, trying new things, and supporting
the ethic of experimentation... To be adaptive, a company must be restless, never satisfied with its progress, constantly
searching and experimenting."
An evolutionary approach to strategy might be further described:
p.345,346"An evolutionary
approach to strategy emphasizes keeping the strategy tree bushy and the options open for as long as possible... one must know
which initiatives are promising and which are not... Every... Plan must have clear, thoughtful measures of success and a plan
for collecting data in as close to real time as possible... for an evolutionary approach to work, one needs real-time feedback
on where the nectar is [JLJ - an example was previously discussed involving Bees reporting the presence of nectar to other
bees and swarming to retrieve it] , and thus each strategy experiment must have a customary balanced scorecard [JLJ - a set
of performance metrics that is designed to provide visibility into value creation] and the measurement systems for carrying
it out."
We need to develop an adaptive mind-set. We focus on selecting "candidate moves", then we "test" them using
our model of how things interact on the game board. The moves with promise are the ones that survive our rigorous testing.
p.347,348"one's mind-set is perhaps the most important factor in creating an adaptive approach to strategy.
One way to picture an adaptive mind-set is to think more like a venture capitalist than a manager... venture capitalists use
their portfolios to learn their way into the future and thus generate high returns out of some of the riskiest and fastest
changing markets in the business world... Above all, an adaptive mind-set is willing to say, 'We learned something new; we
need to change course.' "
Perhaps what we need is a "Whack on the Side of the Head" (ouch)
A Whack on the Side of the Head (von Oech)
Our machine, in order to be successful, might attempt to duplicate the activity of a creative child at play.
p.14"Creative thinking requires an outlook that allows us to search for ideas and play with our knowledge
and experience."
The concept of a "Red Queen Race" governs the evolution of a species adapting in nature, and might even
explain how one side or the other must try as hard as possible simply to maintain "equal chances" in a hard-fought game.
p.61"One of my recent favorites [the subject being discussed is metaphors] was concocted by the American
biologist Leigh Van Valen, who was inspired by the Red Queen character from Lewis Carrol's Through the Looking Glass. She's
the one who runs hard but never gets anywhere because everything else in the landscape is also running. As she tells Alice,
'it takes all the running you can do to keep in place!" Van Valen used the Red Queen as a metaphor for his evolutionary principle
that regardless of how well a species adapts to its current environment, it must keep evolving to stay up with its competitors
and enemies who are also evolving. Thus the 'Red Queen' effect: do nothing and fall behind, or run hard to stay where you
are. This is also found in business, sports, new technology development, and arms races."
The identification of constraints helps us to determine whether we are making progress towards our goals.
p.114,115"Constraints can be a powerful stimulant to the creative process... [Limits] force us to think
beyond conventional solutions and find answers we might not otherwise have discovered... it can be argued the product of almost
every activity... can be made more creative if we'd take some time to playfully add a few constraints at the beginning of
the project."
If we tell the machine how to evaluate the positions, we might be surprised at the paths that the machine
is able to find which lead to those"better" positions:
p.180"The American General George S. Patton had similar ideas on how to stimulate people's creativity. He
said, 'If you tell people where to go, but not how to get there, you'll be amazed at the results.' He knew that posing a problem
in an ambiguous way would give more freedom to the imaginations of the people who were working on it."
We might need to look at the process of creating, since we desire our machine to make creative moves:
The Courage to Create (Rollo May)
We might desire that our machine be "engaged" in the game, with an intensity.
p.41,44"The first thing we notice in a creative act is that it is an encounter... there must be a specific
quality of engagement... This leads us to the second element in the creative act - namely, the intensity of the encounter."
Creativity requires limits - it is the struggle against these limits which for May produces the creative
act.
p.113"creativity itself requires limits, for the creative act arises out of the struggle of human beings
with and against that which limits them."
Human consciousness arises from the tension between possibilities and limitations. An organism cannot be
truly strategic unless it struggles with limits in the progress towards goals.
p.114-115"Consciousness is the awareness that emerges out of the dialectical tension between possibilities
and limitations... If there had been no limits, there would be no consciousness... the struggle with limits is actually the
source of creative productions... Creativity arises out of the tension between spontaneity and limitations, the latter...
forcing the spontaneity into the various forms which are essential to the work of art"
Perhaps what we ultimately desire for our machine is a kind of "imagination" which completes the picture,
substituting the "likely" for the unknown.
p.131"The human imagination leaps to form the whole, to complete the scene in order to make sense of it.
The instantaneous way this is done shows how we are driven to construct the remainder of the scene. To fill the gaps is essential
if the scene is to have meaning."
Now we look at a book by Jim Haudan. Can the principles used to engage people in executing a strategy
be applied, in a similar way, to machines?
The Art of Engagement by Jim Haudan
Ultimately, a machine playing a game must try to see the possibilities before they become obvious to the
opponent. Victory might depend on it.
p.10"We learned from Ted Levitt, former editor
of the Harvard Business Review, that 'the future belongs to people who see possibilities before they become obvious.'
"
Making sure that our machine is properly "executing the strategy" might be just as important as "seeing
ahead" into the future positions that have promise. We might want to periodically measure whether or not we are doing this.
p.12"We soon came to realize that success did
not actually depend on how masterfully a leader or manager could 'see forward' or perceive the detailed nuances of a new strategy.
It depended on successfully engaging the rest of the organization in... executing the strategy."
Stopping analysis in the lines that are not promising might be just as important as starting a new analysis
line. Our machine might not be able to focus on what is important if it is constantly exploring every possible move.
p.33"What is interesting is that the stop
activities must be just as deliberate and intense as the start actions."
We should aim, as a starting point, for a simple strategy to gain a competitive advantage. This will enable
us to use simple heuristics, and simple tests to verify that we are following the strategy.
p.36"This isn't about dumbing down the strategy,
but about making it sophisticated, elegant, and brilliantly simple so less experienced people can instantly grasp its meaning."
We might ask ourselves, what actions should we have our machine perform that are relevant? This might help
us to keep the machine engaged in the game.
p.40"Relevance is at the heart of engagement."
How might we sketch cause and effect relationships? If things are complicated, do we have first-order or
second-order simplifications that we can diagram?
p.44"leaders need to constantly bring motion
to the illustrations they create. The idea is to enable people to see cause and effect - how strategy works, where it's going,
the key links, and what's needed to make all the pieces work together."
We should aim for a system-wide level of understanding.
p.58"until people can think systemically, they'll
never be able to think big. They will not be engaged until they can see the whole system."
How can we make the strategy visible?
p.102-103"Visual iteration [JLJ - a kind of storyboarding]
allows people to see their ideas on paper in order to make sure that they're well thought through and convey what they intend
to convey... visual iteration enables people to think in systems."
We should be concerned with "what's going on" and with exploring how things on the gameboard change, or
are likely to change.
p.117"It's important to remember that sketching
the truth is valuable only if it... enables new energy and enthusiasm for finding a better way... The key is to use the sketch
to start a conversation that, in the end, is about how reality must change"
We would hope to achieve a strategic engagement for our machine.
p.176"If strategic engagement is a process, improving
it needs to be approached with the same discipline as any other process - with a solid assessment of the strengths and weaknesses
in the process and a clear understanding of how it is performing. This knowledge can open an enormous opportunity for better
aligning your resources to the areas with the greatest potential for success... What's lacking is a clear link to the behaviors
that indicate effective strategy execution. If you don't consider the strategic component of engagement, you may
be measuring engagement in an incomplete way."
We would want our machine to understand where things are now and where we would like them to go.
p.236"To execute a strategy, people need an honest
assessment of where they are and a clear picture of where they want to go... A strategy story, told properly, is an adventure
that invites others to join"
We should aim for a simple strategy that is effective. We should look for a more complex strategy only when
our simple strategy is not performing.
p.237"Albert Einstein... once said, 'If you can't
explain it simply, you don't understand it well enough.' "
How can we visualize our strategy?
p.237"Visualization creates simplicity. It forces
us to think more simply... Visualization acts as a mirror for our thinking, revealing just how complete our ideas are... or
aren't. If a strategy is not clear enough to visualize, it's not clear enough to deploy or to engage people."
What questions should we ask, and attempt to answer, when we attempt to evaluate how "promising" a position
is?
p.241"Success is the result of asking better
questions so that we reach deeper insights, which will allow us to more effectively solve the real problems."
How can we effectively engage the power of a computer to play this game?
p.242"The most sophisticated strategy is worthless
if humans can't embrace it or be engaged in it... It's up to us to constantly bridge the gaps between people and possibilities
by knowing - and practicing - what it takes to tap into the latent, unused potential that's just waiting to be awakened and
engaged."
Now, let's look at a book by David Apgar
Relevance by David Apgar
Relevance has a place in deciding what to do next. When overloaded with information, we should be able to
determine what is important and what is not.
p.2-3"This book is about telling in an age of information overload what's relevant to what you're trying
to do. The trouble with an overload of information isn't just that it's confusing. It's that the data have conflicting implications...
This book helps find what information is most relevant despite the overload."
Specificity and Relevance determine the usefulness of a piece of information.
p.6-7"A major idea in this book is that the usefulness of any piece of information depends on two complementary
values: its specificity - which you can think of as its content, or simply how surprising it is - and its relevance. Specificity
gives you a sense of the power of a piece of information because you can draw more conclusions from precise outcomes than
from vague ones... Relevance complements specificity. While specificity is intrinsic to information - it's the same no matter
what your goal is - relevance depends entirely on that goal, or at least on the assumptions behind it... specificity and relevance
drive the value of information"
We should test our strategy and modify it if it is not working.
p.16"the testable strategies, relevant metrics, and reviews of results proposed in the following pages are
more than a guide for coping with performance information overload. They embody an experimental approach to management and
problem solving. It's an approach that reflects a thorough fallibilism about the strategies we construct to meet our goals
and an optimism that we can always do better."
Learning and adapting depend on the ability to test the effectiveness of a strategy and to respond
differently if we are not making progress towards our goals.
p.18-19"A strategy is testable if it spells out goals and assumptions about how to achieve them that could
conceivably prove wrong... lack of a testable clarity in a company's strategy can impair its ability to learn and adapt."
Apgar believes that clear strategies are rare. We proceed by using our intuition, but we test it for effectiveness to
see that it is generating results. If not, we change what we are doing.
p.24-25"The rarity of strategic clarity helps explain why it's so hard to tell what matters in our burgeoning
performance reports... A strategy is testable if it spells out goals and assumptions about how to achieve them that could
conceivably prove wrong... clear strategies are few and far between... Strategies often become stagnant because their goals
are stagnant."
We might benefit from a short-term indicator which measures progress towards our long-term goals.
Our rule-of-thumb which measures progress should be part of a carefully crafted strategy if we are to make progress in
a competitive environment.
p.26"Some ideas take a long time to bear fruit... The problem with this defense of long-term goals is that
it affords no way to tell exactly which one really is your best new initiative... To see whether one of several possible initiatives
really is most promising, you need to devise some near-term indicator that it's working. Such an indicator lets you test and
adapt the strategy behind the initiative."
A clear strategy might be derived from experience.
p.29"clear strategies are rare... they're hard to specify. They need to set aggressive goals to avoid stagnation.
And they need to propose a way of achieving them that experience can help refine over time. Clear strategies are hard because
they really force you to find out, if not to know, what you're talking about."
We need to find metrics that matter and avoid collecting information that is not useful for testing
our strategy.
p.40-41"To learn anything from the heap of data our management information systems grind out, we need a
basis for telling which metrics matter. The examples of Chapter One suggest that strategic assumptions - basically, our biggest
bets or guesses - provide such a basis. The idea is to focus on the performance metrics that best test those assumptions and
to try out new assumptions in place of the ones that fail... Three things distinguish these organizations [JLJ - clients of
McKinsey consulting services that are deemed to be adaptable]: they make bold assumptions about the unknown as a matter of
course, they spend their time looking for errors rather than trying to prove themselves right, and when they find an error
in an assumption, they're quick to revise it and change course."
We make strategic assumptions and then derive performance metrics that test these assumptions. We seek to
measure progress towards our goals. There is nothing wrong with a strategic assumption, as long as it is working. When it
stops working we change our strategy.
p.71"Derive performance metrics from your assumptions... a concrete definition of relevance:
A performance indicator is relevant to a strategic assumption if the assumption's truth or falsity
greatly affects the results you expect.
If the results from an indicator are equally likely whether your assumptions are true or false,
then it is irrelevant."
Relevant performance metrics allow us to track whether or not we are following our strategy.
p.74-75"most organizations use a definition of relevance that's starkly different from the one I set out
in the previous section. Their performance metrics either ignore strategy altogether, track red herrings, or get lost in endless
demands for more data. As a result, they tend to become less adaptable or mired in complexity."
What indicator lets us test a strategic assumption? Is it specific enough? In what cases are we moving towards
our goals, but the indicator does not think so? What about the reverse?
p.99-103"To test a strategic assumption, you need to find an indicator that's relevant to it... If you want
to focus on data that can improve decisions, you must find a way to screen indicators for relevance... Choose a strategic assumption
to test... The first part of the relevance screen... asks whether an indicator's results could possibly disprove your assumption...
The second half of the screen focuses on possible results that would be favorable to an assumption... The main purpose of
this screen is to pick out the handful of indicators most relevant to each key assumption you want to test... Even if you
find an indicator that's highly relevant to a key assumption, it may be too vague to be useful. All other things equal, a
more specific performance measure will be more valuable than a less specific one."
If our strategic assumptions are correct, what would we see? How would we measure that? What would we count,
and how would we process this countable information to create an indicator?
p.111-112"Imagine what effects you would see if your assumptions were right... Look for ways to measure
those effects... Make sure what you measure would be unlikely if your assumptions were wrong."
The metrics that matter are tied to the strategic assumptions that matter. We should begin with the strategic
assumptions and then create a performance metric which lets us track whether or not we are pursuing our strategy.
p.181"The metrics that matter are the ones that are relevant to assumptions that matter... The best metric
for testing a strategic assumption is both relevant - in the sense of avoiding false positives and negatives - and specific.
Relevance makes sure your expectations are sensitive to the truth or falsity of the assumption. Specificity makes sure the
metric really captures those expectations. Together, relevance and specificity reflect the real value of a metric for someone
proceeding on the basis of the assumption."
Complexity by Melanie Mitchell
Our strategy exists in, and must operate within the bounds of a complex environment. Let's take a look at
the concept of complexity.
Melanie Mitchell offers this definition of a complex system:
p.13"Now I can propose a definition of the term complex system: a system in which large networks
of components with no central control and simple rules of operation give rise to complex collective behavior, sophisticated
information processing, and adaption via learning or evolution... Here is an alternative definition of complex system:
a system that exhibits nontrivial emergent and self-organizing behaviors."
Looking at snapshots in time of our system might not let us get the full picture of what's going on. We
might turn to dynamic systems theory to help us understand our system in motion, and the cause and effect relationships:
p.15-16"Dynamical systems theory (or dynamics) concerns the description and prediction of systems
that exhibit complex changing behavior at the macroscopic level, emerging from the collective actions of many interacting
components. The word dynamic means changing, and dynamical systems are systems that change over time in some way...
Dynamical systems theory describes in general terms the ways in which systems can change, what types of macroscopic behavior
are possible, and what kinds of predictions about that behavior can be made."
We ought to begin with an idea model - a simple concept which captures the essence of the interactions and
allows us to construct models of deeper complexity. Our understanding should begin with an understanding of the idea model.
p.38-39"idea models - models that are simple enough to study via mathematics or computers but that
nonetheless capture fundamental properties of natural complex systems. Idea models play a central role in this book, as they
do in the sciences of complex systems."
Our strategy might evolve over time, as we select candidate moves and examine paths which are promising.
p.79"According to this view [JLJ - a summary of the ideas of Darwin], the result of evolution by natural
selection is the appearance of 'design' but with no designer. The appearance of design comes from chance, natural selection,
and long periods of time. Entropy decreases (living systems become more organized, seemingly more designed) as a
result of the work done by natural selection."
We should look at the information present in the environment, and the possible computations we can perform
on this information, in order to find useful metrics for understanding what is going on and deciding what to do next.
p.170"At a very general level, one might say that computation is what a complex system does with
information in order to succeed or adapt in its environment."
The paths which are promising we should devote more resources to. We should take these resources from the
paths which are not deemed promising, based on our perception.
p.182"Many, if not all, complex systems in biology have a fine-grained architecture, in that they consist
of large numbers of relatively simple elements that work together in a highly parallel fashion.
Several possible advantages are conferred by this type of architecture, including robustness, efficiency,
and evolvability. One additional major advantage is that a fine-grained parallel system is able to carry out what Douglass
Hofstadter has called a 'parallel terraced scan.' This refers to a simultaneous exploration of many possibilities or pathways,
in which the resources given to each exploration at a given time depend on the perceived success of that exploration at that
time. The search is parallel in that many different possibilities are explored simultaneously, but is 'terraced' in that not
all possibilities are explored at the same speeds or to the same depth. Information is used as it is gained to continually
reassess what is important to explore.
Perhaps we should look to the ants for solutions to problems involving search and foraging.
p.182"Similarly, ant foraging uses parallel-terraced-scan strategy: many ants initially explore random directions
for food. If food is discovered in any of these directions, more of the system's resources (ants) are allocated, via the feedback
mechanisms described above, to explore those directions further. At all times, different paths are dynamically allocated exploration
resources in proportion to their relative promise (the amount and quality of the food that has been discovered at those locations).
However, due to the large number of ants and their intrinsic random elements, unpromising paths continue to be explored as
well, though with many fewer resources. After all, who knows - a better source of food might be discovered."
We should aim for a balance between a focused search and general-purpose foraging activity, when we search
for moves which are promising.
p.183-184"In all three example systems there is a continual interplay of unfocused, random explorations
and focused actions driven by the system's perceived needs... As in all adaptive systems, maintaining a correct balance between
these two modes of exploring is essential. Indeed, the optimal balance shifts over time. Early explorations, based on little
or no information, are largely random and unfocused. As information is obtained and acted on, exploration gradually becomes
more deterministic and focused in response to what has been perceived by the system. In short, the system both explores
to obtain information and exploits that information to successfully adapt. This balancing act between unfocused exploration
and focused exploitation has been hypothesized to be a general property of adaptive and intelligent systems."
What we are ultimately after is a model which allows us to form perceptions in our environment and allows
us to search for promising paths, based on our values.
p.209"A model, in the context of science, is a simplified representation of some 'real' phenomenon. Scientists
supposedly study nature, but in reality much of what they do is construct and study models of nature."
Our model can be used to predict the future, or at least to position ourselves for a future which is uncertain.
p.210-211"Models are ways for our minds to make sense of observed phenomena in terms of concepts that are
familiar to us, concepts that we can get our heads around... Models are also a means for predicting the future... computers
are often used to run detailed and complicated models that in turn make detailed predictions about the specific phenomena
being modeled... a major thrust of complex systems research has been the exploration of idea models: relatively simple
models meant to gain insights into a general concept without the necessity of making detailed predictions about any specific
system."
Our model does not have to be precise - it merely has to be useful.
p.222"All models are wrong, but some are useful. - George Box and Norman Draper"
Our strategy, in order to be effective, might have to focus to some degree on the relationships between
the entities present, rather than on the entities themselves.
p.233,252"Network thinking means focusing on relationships between entities rather than the entities themselves...
network thinking is providing novel ways to think about difficult problems... and, more generally, what kind of resilience
and vulnerabilities are intrinsic ... and how to exploit and protect such systems."
In order to understand complex systems, perhaps we should look to the suggestion of Norbert Wiener, that
we should examine feedback, control, information, communication, and purpose.
p.296"A prime mover of this group [JLJ - The Macy Foundation conferences] was the mathematician Norbert
Wiener, whose work on the control of anti-aircraft guns during World War II had convinced him that the science underlying
complex systems in both biology and engineering should focus not on the mass, energy, and force concepts of physics, but rather
on the concepts of feedback, control, information, communication, and purpose (or 'teleology')."
Fast Strategy by Doz and Kosonen
The metric we use to evaluate our chances must be sensitive in a strategic manner - that is, we must
be able to detect trends and opportunities as soon as they emerge.
xi"In our research, we saw that winning at the fast strategy game hinges on a few deciding differences.
One such difference is high strategic sensitivity: the early awareness of incipient trends and converging forces, the acute
perceptions of their importance, and the intense sense-making and reflective efforts they trigger."
Insight is crucial and must be part of our machine's perception.
p.10"Insight needs to replace foresight. The game you play appears only over time... In these emerging strategic
situations, fast pattern recognition... becomes key. The world around us keeps emerging, and our perception of it keeps reshaping
itself as we play... If you are fast in fitting pieces, the overall structure of the puzzle will evolve your way: you create
your own future and shape the markets and the competitive landscape to your advantage."
Our metric must have a strategic agility in order to succeed in a competitive environment.
p.17"Strategic agility is most needed in markets characterized by fast changes and growing systemic interdependencies...
More lasting value may come from identifying areas where complex, evolving, shifting interdependencies are likely to endure"
Once more, we ought to consider a metric with a heightened strategic sensitivity.
p.20"In sum, strategic foresight remains important. It remains important to anticipate the consequences
of key trends, to identify disruptions and discontinuities early, and either affect them to one's advantage, or have the lead
time to adjust to them effectively and in a timely fashion... Yet, where change is fast, complex, and systemic, and stable
sources of strategic advantage short-lived, strategic foresight needs to be strongly complemented by strategic insight: an
ability to perceive, analyse and make sense of complex strategic situations as they develop, and be ready to take advantage
of them. Heightened strategic sensitivity is required."
In extremely unclear situations, we ought to consider a simple metric, in fact any metric that works.
p.22-23"When uncertainty is extreme, and the complexity of the environment cannot be gauged, conventional
strategy making and decision models no longer apply. Even developing insight becomes impossible. Simple rules of variation
and selection may work better in these situations... But in the absence of enough foresight... a breakthrough can also be
achieved by just probing randomly at the enemy's positions along a broad frontline, and having the simple rule to concentrate
one's forces and attack where the enemy's reactions to exploratory probes are the weakest... rather than attempting to develop
a comprehensive understanding of a strategic situation simply devising action rules... may well be all what can be done, and
yield much better results than a late or wrong attempt at making sense of the environment."
Strategy is a theory of how to compete. The development of an effective strategy is perhaps the most important
thing we can do.
p.24"Think of strategy as a theory. A theory of how to compete, how to create and capture value, and developing
strategy requires theory-building skills... strategy is a theory of the future applied to an all-important sample of one...
and its likely validity, before it's implemented, can only be assessed based on the quality of the process that generates
it. In other words, the quality of the strategy making process is fundamental, yet it often receives little attention."
Choices should be based on plausible predictions of cause and effect relationships.
p.24"Discipline in logical thinking when developing a strategic commitment, and systemic attention to a
detailed 'How plausible is this really?' question, help improve the quality of choices."
Fluid resources make strategy possible and in the absence of clear paths, flexibility is perhaps the best
alternative.
p.29"Strategic sensitivity and collective commitments are of little value without resource fluidity, the
ability to redeploy resources quickly toward strategic opportunities as they develop. Fast decisions in complex environments
call for rapid resource deployment for their implementation."
Our plan will be to make sense of situations as they develop, and to immediately take advantage of situations
once we recognize that they might evolve in our favor. Strategy involves an exchange of intangible resources.
p.33-34"Strategic agility calls for three fundamental shifts in the emphasis of top management, in how they
steer the firm. First, it requires a shift from foresight-driven strategic planning to insight-based strategic sensitivity,
putting more emphasis on making sense of current situations as they develop than on anticipation of future strategic interactions...
Third, strategic agility needs a mindset and behavior shift from resource allocation and 'ownership', to resource sharing
and leverage, and from a focus on budgetary games and tournaments around capital allocation to a commitment to sharing and
exchange around intangible resources"
Insight and connection are related. Perhaps success comes only from an ability to play with a model of reality
that is useful.
p.55-56"insight results from connection, from a rich network of creative interactions... Connectedness,
experiments, and modeling do not pay off unless you are really smart about learning from external stimuli. Results come from
experiments, and from playfulness with one's own position."
Maintaining a portfolio of possible choices, at any point in time, is a very good strategy.
p.61"According to Chris Thomas, Chief Strategy Officer of Intel's Solutions Marketing Group:
You have to first pick up the most likely winners from the 360 degrees opportunity space and start supporting
these intellectually and financially. Then you gradually increase investments in the most likely winner."
Our metrics must be sensitive and diagnostic in nature or we risk missing opportunities.
p.69"Overambitious target setting and the wrong type of metrics can kill new insights and opportunities
too early."
Our resources should be fluid and flexible.
p.96"Without resource fluidity, strategic sensitivity and collective commitments toward new strategic opportunities
remain useless... resource fluidity is the ability to redeploy resources quickly toward strategic opportunities as they develop.
Only the ability to mobilize and reallocate resources toward new strategic opportunities with maximum fluidity makes strategic
agility real."
Summary
If you are designing a strategy for a game-playing computer, the chances are good that the above concepts have crossed
your mind at one point or another.
There are many ways you can answer these questions. Some approaches are simple and others are more complex.
The goal here was to develop a framework useful for developing a strategy. We are essentially, thinking about thinking.
Hopefully, this makes for a better strategy.
Perhaps the quotes from Stephen A. Lopez at the top of this paper are correct. The improvement of the strong
chess programs might be due to their ability to come up with better candidate moves.
Do you think that a computer can "play" chess?
|
