tegwick
469d00d679
CB-RES-0008 with a runnable baseline (games/ground/examples/difficulty-baseline.rs), and the measurement produced a finding before any solver exists. A GREEDY BOT WINS 200 OF 200 GAMES AT FIVE AND SIX SEATS. Median margin +3, 11.8-12.0 points available against a threshold of 9. The curve is 66% / 82.5% / 95% / 100% / 100% across 2/3/4/5/6 seats. Row-level, as GROUND-WP-0004 T02 requires: available points 6 / 9 / 12 against thresholds 5 / 7 / 9, so the ratio RISES with seat count (1.20, 1.29, 1.33) while the table also gains actions per round to clear it with. Three multipliers pointing the same direction. It also explains the maintainer's report without needing a solver at all: "I felt it was too easy but then we lost" is two true statements about different seat counts. Cost measured and it rules out the obvious approach. Branching is small (mean 4.7-9.1) but legal_commands costs 112-161 us per call because it filters candidates through full validate. Exhaustive search is out at every seat count; 10^4-10^5 nodes is 1.4-14 seconds, which is the budget the ADR must design inside. Prior art names the trap: determinized search (PIMC) suffers strategy fusion (Frank, Basin & Matsubara 1998) -- the search picks different actions in states a real player cannot distinguish, so the witness may require knowing what was on top of the deck. Such a line still replays green, so the checkability benchmark does not catch it. Honesty and checkability are different properties; stated explicitly so T03 cannot conflate them. The survey states its own most likely killer up front (§6): a view-only search cannot fold events, so making the information boundary structural rather than a promise may not be affordable. Better found here than in T05. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> |
||
|---|---|---|
| .forgejo/workflows | ||
| benchmarks | ||
| crates | ||
| decisions | ||
| editions/ground-darvo-r0 | ||
| evidence | ||
| games/ground | ||
| history | ||
| research | ||
| scenarios | ||
| specs | ||
| tools | ||
| workplans | ||
| .custodian-brief.md | ||
| .gitignore | ||
| Cargo.lock | ||
| Cargo.toml | ||
| clippy.toml | ||
| facts.toml | ||
| gates.toml | ||
| INTENT.md | ||
| LICENSE | ||
| Makefile | ||
| README.md | ||
| rust-toolchain.toml | ||
| WORK-RECORDS.md | ||
clay-borg
A rebuild from scratch simulation and games engine framework set up to assimilate and optimize techniques and implementations useful for games, simulations, robotics.
Licensed under the Target Revenue Source License (TRSL V1C1) — see LICENSE; canonical text lives in the org's target-revenue repository.
Running the gates
make all # every gate, from a clean shell
make -C /path/to/clay-borg all # …or from any other directory
There is no environment setup step. No cd, no export PATH, no
activation script. make locates the repo from its own path and cargo
from the standard rustup locations; the Python tools do the same via
tools/repo.py. The only prerequisites are a rustup toolchain and Python
3.11+.
This is deliberate and enforced: make env-test runs every tool from /
with a PATH containing no cargo, and make all includes it. CB-RES-0003
measured 84 agent turns and $15.33 spent prefixing commands with cd and
export PATH before that friction was fixed at the root (CB-WP-0004 T01).
Other useful targets: make cost (spend per task), make cost-budget
(spend since the last commit), make cost-mix (mechanical vs judgment
turns), make loop-lint (executable InnerLoop rules), make self-tests
(every tool's positive control).
GROUND
The first product vertical is a virtual tabletop implementation of GROUND — A Game of Bonds and Rivalry: DARVO Edition. The boardgame itself (rules, editions, content) is at home in the sister repository ground-game — that repo is authoritative for what GROUND is; clay-borg implements the engine that runs it.