What is a compelling smallest (!) upper bound of reasonable (those that can happen in…
- User
What is a compelling smallest (!) upper bound of reasonable (those that can happen in competition) number of chess games?
- Assistant
A “mathematically possible” chess game tree is mind-blowingly huge (Shannon’s famous 10^120 figure).
What you asked for, though, is something far smaller:• only games that could conceivably appear in serious competition;
• i.e. moves a strong player (or engine) would regard as at least half-sensible.Below is a quick way to get a compelling but minimal upper bound for that smaller set.
────────────────────────────────────────
- How long is a competitive game?
────────────────────────────────────────
FIDE time-controls rarely see games exceed 80–90 plies (40–45 moves per side).
Extreme marathons happen, but are rare.
We will cap the length at 100 plies (50 moves each).
Anything longer is already “unreasonable” for this estimate.
──────────────────────────────────────── 2. How many reasonable choices per ply? ──────────────────────────────────────── Legal choices (≈30–35 on average) are heavily filtered by human/engine judgement:
Phase Typical moves considered ———————————————————————————————— Deep opening prep 1 Early middlegame 3 Mid-/late middlegame 3–4 Endgame (many forced) 1–2A good “generous yet realistic” single number is
B = 3 candidate moves per ply.
(Grandmaster think-aloud protocols show ≈2–4 real candidates most of the time.)──────────────────────────────────────── 3. Put it together ──────────────────────────────────────── With at most L = 100 plies and at most B = 3 sensible moves per ply,
N ≤ B^L = 3^100.Log-10 it:
log10(N) = 100 · log10(3) ≈ 100 · 0.4771 ≈ 47.7.So
N ≤ 3^100 ≈ 5 × 10^47.──────────────────────────────────────── 4. Trimming the bound even more ──────────────────────────────────────── • First 20 plies are often book (only one line followed),
leaving only 80 plies where branching really happens.
That alone knocks the exponent down by 20 × 0.477 ≈ 9.5 powers of ten.• Many late-game plies are forced (≈1 choice).
Replacing the last 20 plies with “1” rather than “3” shrinks another 9.5 powers.Do both adjustments:
Effective branching plies = 60 ⇒ log10(N) ≈ 60 · 0.4771 ≈ 28.6 ⇒ N ≈ 4 × 10^28.──────────────────────────────────────── 5. Round-off “compelling minimum upper bound” ──────────────────────────────────────── Because our assumptions were all deliberately generous,
anything beyond 10^30 distinct games is already well outside practical reach.Final answer (one-liner):
A safe, smallest-plausible upper bound for the number of competitive-quality chess games is on the order of 10^30–10^40, with 10^30 (one nonillion) being a convincing, conservatively small ceiling.That is trillions of trillions of trillions fewer than 10^120, yet still comfortably contains every game you could ever see in real over-the-board play.
- How long is a competitive game?
────────────────────────────────────────
FIDE time-controls rarely see games exceed 80–90 plies (40–45 moves per side).
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.