What is a compelling upper bound of reasonable (those that can happen in competition)…
- User
What is a compelling upper bound of reasonable (those that can happen in competition) number of chess games?
- Assistant
There’s no “small” number here—the answer is enormous. But one answer that many experts find “compelling” is to show that if you restrict yourself to only legal chess games (that is, those one might conceivably see in actual play, with the game‐ending rules in force) then a generous but rigorous upper bound is roughly
[ 80^{5900}\approx 10^{11220},. ]
Let me explain how one gets to a number like that.
The Basic Idea
-
Game termination by rule:
Although chess is “in principle” an infinite game if you ignore the rules, the rules force termination. In modern chess the 50–move rule (indeed, a modified version of it) and the rule against three–fold repetition guarantee that no legal game can go on forever. (In fact, careful analysis shows that no legal game can last more than about 5,900 moves.) -
A maximal game:
Assume that the very longest legal game has (to be generous) about 5,900 moves. (More precisely, one may prove that—for example using the 50–move rule—the maximum number of moves in a legal chess game is on the order of 5–6000 moves.) -
Average branching factor:
At each “ply” (that is, each half–move) there is a certain number of legal moves. (Many estimates for the average number in practical positions lie in the 30–40 range, but to be safe we may assume a generous upper bound of 80 legal moves at each ply.)
Putting it Together
Assume that each ply offers at most 80 choices and that there are at most 5,900 moves (i.e. plies) in a game. Then an extremely “loose” upper bound on the number of different games is
[ 80^{5900},. ]
Taking logarithms (base-10) we have
[ \log_{10}(80^{5900}) = 5900\log_{10}80 ,. ]
Since (\log_{10}80) is roughly 1.903, this gives
[ 5900\times 1.903\approx 11220,. ]
That is, the total number of possible games is bounded above by roughly
[ 10^{11220},. ]
How “Reasonable” Is This Upper Bound?
-
On the one hand, note that many actual tournament games last only 40–80 moves and often involve forced or restricted move choices. So the vast majority of games encountered in competition are a very, very tiny subset of the huge number (10^{11220}).
-
On the other hand, when we talk about the “number of chess games” in a combinatorial sense we mean all the different sequences of legal moves that could ever happen—regardless of how likely they are in practice.
In many texts this huge number is compared to the famous “Shannon Number” (often given as roughly (10^{120})) which is a lower bound on the game–tree complexity of chess. (Indeed, using rough arguments one is led to conclude that the number of possible games is “astronomically large” and far exceeds, for example, the number of atoms in the visible universe.)
Final Answer
Thus, a compelling upper bound on the number of possible chess games that can occur (i.e. those meeting the official rules of competition) is around
[ 10^{11220},. ]
This is, of course, an upper bound using very generous estimates. Any more realistic count—taking into account the typical game lengths and the fact that many moves are forced in critical positions—would be enormously smaller, but still a number of astronomical proportions.
-
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.