Von Robert Jasiek, engl., 268 Seiten.
On the one hand, we consider one local endgame, an ensemble of the largest local endgames or several local endgames. On the other hand, we model the positional environment of the remaining local endgames as a value environment. During the early endgame, we model it by its largest move value called the temperature indicating the value of starting in the environment, although we sometimes also consider the second-largest value. During the late endgame, our model is an alternating sum of the move vales taken by the player minus those taken by opponent.
For the considered local endgames, we decide whether to play locally or in the environment. Sometimes both choices are right. We learn the first and last moments of playing in sente or reverse sente in every particular local endgame. We also learn exactly when to interrupt long successive local play and continue in the environment.
The decisions differ for the early and late endgame. They depend on the starting player, types of local endgames, their follow-ups, their calculated local values and the temperature. The early endgame is too complex to be solved completely while, in principle, we might solve the late endgame by tactical reading given much time. Instead of Black versus White, attacker versus defender, or sente player versus reverse sente player, it is often useful to distinguish the creator, for whom a follow-up or options are available, from the preventer, for whom they are unavailable. There are local gotes, ambiguous local endgames, local sentes, and local endgames with alternative gote and sente options. A non-iterative local endgame can have one or two simple follow-ups. We need some of these values: move value, follow-up move value, gain of a move, net profit of a sequence, its resulting count. The temperature can be low, medium or high.
Endgame 4 - Global Move Order states the correct decisions for all basic cases with these aspects for one local endgame with simple follow-ups in an environment. It also describes some decisions among several local endgames or, during the late endgame, local endgames with iterative follow-ups. During the early endgame, the latter do not have any explicit solution but we may apply the good approximations to them, too.
The book goes far beyond inventing just some new theory. Instead, the theory is already fully developed with different methods, such as comparing two well chosen test sequences, comparing counts, comparing net profits or applying a principle. Since Volume 5 proves their equivalence, everybody can choose his preferred method. Although such decision-making is very much faster than tactical reading, the theory offers further means of acceleration.
Examples illustrate every case. Finding the rarest examples took up to four days each. The book also studies how tiny changes greatly affect the outcome and decisions even in basic examples. Furthermore, the book explains various aspects of strategy. Besides, there is a brief history of endgame evaluation.
