# Browser puzzle data

These files are generated from the same public game datasets as the native app by `node tools/build_web_catalog.mjs`. They contain no player profiles, progress, or private account data. Row and answer-pool order is preserved for deterministic daily selection.

Spanish accepted guesses derive from [words/an-array-of-spanish-words 2.0.0](https://github.com/words/an-array-of-spanish-words/tree/612349c44d0a9127fb317f7d56d078eda7ae6e65), copyright 2016 Zeke Sikelianos. Its MIT notice is retained in `LICENSE.spanish.txt`; exact source provenance is recorded in `spanish-provenance.json`. Safe Spanish answer and SubWordy solution lists are the game's editorial allowlists. The broad accepted-guess dictionaries are not used as hidden-answer pools.

Namely spellings and aggregate counts originate in the U.S. Social Security Administration national baby-name data. English Wordy answer tiers are generated using wordfreq 3.1.1 (Apache-2.0) frequency classifications. Comparison data retain the native game's curated animal roster, FuelEconomy.gov vehicle model families, World Bank country indicators, U.S. states, and chemical elements. Native source documentation is in `data/names/README.md`, `data/WORD_ANSWER_TIERS.md`, and `data/comparison/README.md` in the project repository.

The JavaScript picker implements the Godot 4.6 String hash and PCG32 number-generation algorithms for native puzzle parity. Reference implementations: [Godot String hash](https://github.com/godotengine/godot/blob/4.6/core/string/ustring.cpp), [Godot RandomPCG](https://github.com/godotengine/godot/blob/4.6/core/math/random_pcg.cpp), and [PCG](https://www.pcg-random.org/), by M. E. O'Neill.