v0.4.3
Wiki/Getting Started

The Research Manifest

Why StatCraft is built as an open exploration in memory-safe statistical computing.

1. The Premise#

Statistical software powers critical clinical, economic, and social decisions. Yet legacy statistical packages written decades ago in C/C++ remain susceptible to memory vulnerabilities, obscure crashes on malformed data, and steep commercial license fees that burden public universities.

StatCraft investigates a core hypothesis: Can a complete statistical suite be rewritten in 100% pure Safe Rust with zero `unsafe` blocks, zero panics, and native Model Context Protocol (MCP) integration without sacrificing speed or SPSS analytical fidelity?

2. Memory Safety vs. Statistical Correctness#

As highlighted by independent technical reviews (including remio.ai's analysis *"StatCraft Rust Rewrite Targets IBM SPSS With a Different Safety Promise"*):

Language memory safety does NOT automatically guarantee statistical correctness.

A program can avoid SIGSEGV while silently producing flawed degrees of freedom or misinterpreting missing value codes. This is why StatCraft enforces the Golden Oracle Invariant: every statistical procedure is formally audited and verified against Python scipy.stats, statsmodels, and the *IBM SPSS Statistics 24 Algorithms Guide* within an absolute error tolerance of ≤10−6\le 10^{-6}.

3. Invitation to Academic Validation#

StatCraft is an active research experiment (v0.4.x). We explicitly invite academic researchers, professors, and data scientists to challenge our numerical precision:

  • Run real-world .sav and .csv datasets through StatCraft;
  • Cross-check pp-values and test statistics against SPSS, R, and Stata;
  • Report any discrepancies via our anonymous submission engine without needing a GitHub account.