EnJnDeSIgn/rand-and-rollers

rand-and-rollers feels like a very personal experimental lab for randomness, procedural writing, dice rollers, Fortran sketches, maze generation, tarot/story generators, and AI-assisted numerical play. The project has real energy: the README gives a sense of the author’s voice, and there are concrete runnable ideas like maze.py, maze0.py, die_roller.py, NRNG.f90, TarotV1.f90, and the story element generator. I also like that the repo has a long history, a security policy, an MIT license attempt, GitHub Actions files, and a few tests such as test_ss0.py; those are good signs that the maintainer is trying to make the work shareable rather than keeping it as private notes.

The biggest thing holding the project back is discoverability. Right now the root directory mixes Python, Fortran, prose, generated text, batch files, and many committed .exe files, so a new user cannot quickly tell which programs are canonical, which are experiments, and which are outputs. A small restructure would help a lot: put runnable source under src/ or named folders like dice/, maze/, story/, fortran/; move generated text and binaries into artifacts/ or GitHub Releases; and add a short “start here” section that recommends two or three stable demos. A requirements.txt or pyproject.toml would also make the Python pieces much easier to run, especially because some scripts depend on packages like numpy, pandas, scipy, and matplotlib.

There are some strong ideas worth preserving. The dice roller’s bit-string approach, the maze generator’s symbolic display, and the README’s notes about precision/carry-over in NRNG.f90 all show a distinctive angle on randomness rather than just wrapping random.randint. The next improvement would be to make those ideas testable and reproducible: add deterministic seed options, document expected inputs such as mock_tide_data.csv and simulated_glacier_data.csv, and turn representative scripts into functions with small tests. I would also clean the LICENSE file so it contains only the MIT license text, moving personal notes elsewhere; that matters because licensing clarity directly affects whether others feel comfortable using the code. Overall, this is an imaginative, unusual repository with a clear creative identity, and it would become much more approachable with tighter packaging, cleaner metadata, and a guided path through the best experiments.