Can Logic Alone Solve the Game of Chess?

Play & Learn Chess
142 followers
This to you might look like a extremely easy chess puzzle. You have a rook and king, black has king
behind three of its own pawns. I assume almost everyone watching this video can solve this
puzzle instantly. The move is obviously just rook to e8 checkmate. But if you pause and ask yourself
how did that thought come to you? You won't find a single solid answer. You didn't calculate
anything. You didn't logically deduce the moves. The move came in your mind instantly because of
Now, let's take an extremely complex position as an example. Unlike the previous position,
here your mind doesn't know what to focus on. I assume the first thing that comes to most people's
minds is to play h7 check. But that's not the correct move. Maybe you actually did find the
right move. But how much do you trust that instinct of yours? Does it have any basis?
In science, logic usually starts with axioms, which are statements we assume or accept as true,
without proof. From there, we build a system using rules of inference. An example of a formal logic
is if a is true and a equals b, then b must be true. But there's a catch. Scientific logic is
almost always incomplete and provisional. We can build internally consistent systems,
but they rest on assumptions that might later be overturned. Newton's laws worked until Einstein
showed they were just approximations under certain conditions. Logic itself didn't fail,
but its scope of application did. So science never forms a truly complete logic of the universe.
It forms a provisional logical model, always open to revision by evidence. Logic formation requires
a lot of assumption, and that is the reason why logic has yet to solve things like curing cancer,
understanding the human brain, solving world hunger, and also the game of chess.
Chess is a deterministic, perfect information, finite game. There is no fog of war, no randomness,
no information delay, and no loops. Every game begins from the same position, and each player
can see the entire board. That makes chess theoretically solvable, even if we haven't
solved it yet. But then a question arises, why haven't we solved chess yet? And the answer to
that question is simple. Even though chess is finite, it is astronomically complex. The number
of possible legal positions is estimated to be around 10 duo decillion, and the number of distinct
games could stretch beyond 100,000 trededillion. This means that even a purely logical system would
need to evaluate an unfathomable number of possibilities to determine the best move in
every position. And without shortcuts like intuition, memory, or pattern recognition,
logic is forced to examine every branch, which quickly becomes computationally impossible.
So, while the path to perfect play exists, finding it is like trying to find a single grain of sand
which could be hidden anywhere in the universe. Computers and engines have made remarkable
progress in navigating this complexity, using brute force calculation, heuristics, and more
recently, machine learning. Engines like Stockfish evaluate millions of positions per second,
relying on human-coded evaluation functions and efficient pruning methods to focus only on the
most promising lines. Meanwhile, systems like AlphaZero learned purely through self-play.
And in fact, computers have solved chess, at least partially, when there are seven pieces or less on
the board. That was pure brute force computation, and it is called the chess endgame table base.
The complete seven-piece endgame table base requires over 140 terabytes of storage,
representing every possible configuration and perfect move sequence. Going to eight pieces
would require about 10 petabytes of storage, which is approximately 10 million gigabytes,
due to the exponential growth in possible positions, making it far beyond practical
limits for current hardware and storage systems. So now what? For us humans, is there no way to
understand the game in a purely logical way? Perhaps logic, as powerful as it is,
has limits. Not because it's flawed, but because reality is simply too rich, too complex,
and too nuanced to fully capture. But no one ever said that we couldn't give it a try.
Humans, through sheer effort and relentless curiosity, have achieved unimaginable things.
We split the atom, mapped the human genome, landed on the moon, and built machines that learn.
We achieved those things not because it was easy, but because we chose to try,
even when the odds seemed impossible. Which is why I am going to give chess a try,
to understand it through pure logic and reasoning, without any preconceived notions about the game.
The only information I will use is the rules of chess. Chess logic is complex,
but if we were to shrink that complexity by some degree, if we were to apply logic,
find shortcuts, and create fast logical paths, then we might actually have a shot. Perhaps not
in solving the game, but just in improving and playing it differently than normal.
So, starting from the next video, I'll be starting a series called Numits Chess,
where I put all of this to the test. I'll build logical systems, step by step, and apply them
in real games to see how far pure reasoning can take us. This is more like an experiment,
and who knows, we might end up discovering a new way to learn chess altogether.
So if you're interested in seeing this and want to be part of it, make sure to like and subscribe.
It's the best way to let me know you're interested and want me to keep going with this experiment.
