p.30 From the functional perspective, the concept of predisposition parallels that of adaption: in both
cases, the key task is not so much to forecast the future, but to prepare for an unknown future. In this
context, adaptation implies passive adjustment of a system to a turbulent environment. On the other hand, the creation
of predispositions presumes an active interaction with the environment (in effect, shaping the environment). This
subtle point informs the structural difference between the two concepts. Under adaptation, the structure of a system
is based primarily on versatile essentials; that is, flexible elements and methods of their transformation,
which allows that system to switch (adjust) from one set of circumstances to another. In forming predisposition, the
essentials are treated holistically (in totality), which calls for the introduction of relational parameters that describe
the state of the system.
p.48 In actuality chess is a mixture of deterministic and indeterministic methods of play
with the latter predominating... The key component of any indeterministic procedure is the evaluation of a position.
Since it is impossible to devise a deterministic chain linking the intermediate state with the outcome of the game, the most
complex component of any indeterministic method is assessing these intermediate stages. It is precisely the function of predispositions
to assess the impact of an intermediate state upon the future course of development.
p.49 To sum up, whatever method of play is selected, the key ingredient is the valuation of the position.
As Alexander Lysenko (1990) mentioned, William Steinitz's great feat was to conceptualize the key role of positional evaluation...
The formation of a position as a "fundamental and universal principle may be encapsulated into the following dictum: a superior
plan is invariably based on positional valuation. To evaluate, to judge, to assess is certainly not synonymous with exact
knowledge. But an estimate, a judgment, or an evaluation, though not precise, is, according to the principles laid down by
Steinitz, still an effective guide for the master" (p. 190).
p.50 We still have not devised an algorithm-based formation of the wide spectrum of parameters and their
valuations included in the weight function, that is, a formal link between the various methods of positional valuation and
actual game strategy.
p.55 There is no known deductive theory to construct positional parameters or their valuations - a task
performed by the intuition of chess players.
The set of positional parameters is very diverse; it includes parameters pertaining to pawn structure
(doubled pawns, backward pawn, isolated pawn, pawn chain, etc.); to the relationship between a piece and some part of the
board (rook on an open file, bishop on an open diagonal, etc.); to piece interaction (two bishops, bishop and knight, variety
of pieces on the board, etc) and to many other types of structures.
One of the most intriguing and difficult issues is to establish relative total value of positional
parameters to material (the latter is considered decisive).
p.59 The crux of the proposed algorithm [by Botvinnik] is to define the most effective trajectories
for the pieces partaking in the attainment of the set goal... The passage below sheds some light on Botvinnik's philosophy
of chess as manifest in the Pioneer chess program [from Botvinnik, Analytical and Critical Works, 1987]:
Over the last 25 years I worked on the problem of an "artificial chess master." I have come to the
conclusion that the key component of such an algorithm is positional analysis incorporated into the weight function. This
function allows the master to select high priority moves to be included in the search, to determine when to terminate the
search, and to evaluate positions at the end of the search - in other words to select a move.
I believe grandmaster approach is based on three factors:
1. The balance of material available to each player...
2. Expectation of material gains...
3. Purely positional factor (p): control over certain squares of the board and traversability of the trajectories
leading to an attack of the opponents pieces (p.255)
p.60 Perhaps one reason for Botvinnik's failure to develop such an algorithm was his neglect for
the overall positional valuation, which embodies positional parameters pertaining to the position as a whole rather than to
the individual pieces... Perhaps, the main reason for Botvinnik's lack of success was his infatuation with one single
method rather than a strategy based on the synthesis of diverse methods of play. If so, it is all the more important
not to discard Botvinnik's results in developing an algorithm based primarily on combinatorial style, but to
integrate it with other methods.
p.61 To conclude my brief excursion into Botvinnik's approach, I believe that in general, the author's
mindset exerts a strong influence upon the methods he proposes. Of course, we need to qualify the statement, for
the long chain linking the author's mental attitudes and his scholarly output is mediated by numerous intermediate steps.
However, one key intermediate stage is the development of heuristic principles, which is particularly crucial
in developing innovative solutions.
p.66 Key to strategic positional play is some global action, such as an attack against
the opponent's king or building up a position ripe for such an attack. This action governs a plan, which incorporates
intermediate tasks and respective goals. these goals define certain positional features, which are composed
to a varying degree... In effect, the formulation of intermediate tasks defines tactical aims with goals represented
by dominant positional parameters... the strategic positional play merges strategy and tactics.
p.67 As Bronstein (1981) notes, unlike the combinational style, the comprehensive positional style
allows "the creation of long-range plans which do not have to take all the possible variations into account" (p.
176).
p.69 It seems that the most promising avenue of development of computer chess engenders the synthesis of
the various styles discussed above [JLJ - combinational style, defensive techniques, positional style]. Perhaps, the initial
version of such a synthetic program would fuse reactive method (particularly in the openings), optimization techniques (especially,
in the endgame), and positional style based on strategic planning.[24] We can then complement this amalgam with the combinational
techniques, which can be viewed as a particular case of the positional style. If no combination can be defined and
no intermediate positional goals are apparent, the program may employ the incremental step-by-step positional method.
p.69 My principle thesis regarding positional valuations is the following: while the set of material parameters
and their valuations (particularly semiconditional ones) can be unified and objectivized, the set of positional
parameters and their valuations is basically individualized and subjective.
p.71 The choice of positional parameters and the creation of new ones reflects one's personal style of play...
There exists no complete (particularly no deductive) program to devise positional parameters
