back to the topic of the this thread....
Whether you like it or not from a math point of view,
it's related to some pratical experience in the game of chess.
https://en.wikipedia.org/wiki/First-mov ... e_in_chess
For a mathematical 'proof', as i suggested earlier, the easiest
way imho is not with computer calculation, whether it's
with brute force, or using tricks as graph theory (*).
What i would suggest in math, although not an expert, is
that there is a class of games, where can be logically derived
from the rules (and possibly with some way of (backward/complete)
induction, that the first move is an advantage, and with best play
cannot lose. Think of tictactoe maybe, and then expand upward
in complexity. In other words, we need to show that in such
a game, by principle, the firstmover can avoid a loss (thus there
is (probably) no zugzwang, or other disadvantage for the first move
player in such games, although there still can be a win for the
second player if the first player makes one -or more- mistakes
(in practical, human chess, it often means more than one (little) mistake.
Conjecture 1): there exist a class of games (as described above)
with such an 'advantage' for the first move, that in principle
the first mover with perfect play cannot lose.
Yes this conjecture nr 1 is a rather broad one, i must confess but
right now i don't know of a better word in the discipline of math/game theory
NB don't lecture me about what conjecture can mean in urban slang
or whatever; i'm now talking math (almost upon request) so look it up
https://en.wikipedia.org/wiki/Conjectur ... onjectures
conjecture2) for lots of such as on above conj1) games,
starting from the most simple ones, it's possible to derive
with logical reasoning that the first mover indeed cannot
lose, and thus such a game indeed falls under the class (as in conj1);
this looks like circular reasoning (tautology) but it's not.
For the rest i leave it to the math professors
(not the programmers, although such things become interesting
in the broad field of logical AI, not simple chess algorithms)
(*) in general within game theory, the field of graph/network
theory seems to be coming up as an important/useful discipline
https://www.or2018.be/slides/Ljubic.pdf
So also in chess this field is coming up
https://www.or2018.be/slides/Ljubic.pdf
https://blogs.cornell.edu/info2040/2017 ... ibly-more/
In chronological order:
https://arxiv.org/abs/math/9905198
https://www.jstor.org/stable/10.4169/co ... .278?seq=1
http://snap.stanford.edu/class/cs224w-2 ... -final.pdf
https://syncedreview.com/2020/11/06/dee ... landscape/
Two tools
https://www.sciencedirect.com/science/a ... 1018301687
Maybe interesting for (AI) programmers still interested in chess but
becoming bored with simple search algorithms:
PS apparently in the zipproth/cerebellum some sort of network
is used (in the backsolving) but it's still 'brute force' ofcourse.
Although (DC) it's not about draw games (like apparently in your CAP)
but -like normally in chess opening theory- about trying to seek
a fundamental advantage from the start with an optimal opening strategy
(NB forgot to mention earlier as 7th project, Chess Assistant, the only
program to my knowledge where you can combine stuff as statistics
and score (eg. from CAP) but then the minimax (as in Aquarium) didn't
work in the versions i (years ago) i looked at (something to fix maybe !??!!
Or update Aquarium to also being able to combine score with statistics
and (as i suggested earlier, 'sharpness').