Manifesto

What We Believe

These principles guide how xcodeflex.com approaches the study of programming languages. They are not claims about current findings — they are the commitments that shape the research.

  1. Outcomes unify; implementations differentiate.

    The goal of software is a defined outcome. The language is a means to that end. Comparing implementations requires holding the outcome constant.

  2. Every abstraction eventually meets the machine.

    Runtime behavior, memory layout, and system calls are not hidden from research. High-level language properties have low-level consequences.

  3. Context is part of correctness.

    An implementation that is correct in one environment may behave differently in another. Environment is always part of the finding.

  4. Every language has comparative advantages.

    No language is universally best. Suitability depends on the application, workload, team, constraints, and expected life of the software.

  5. Complexity can move, but it does not disappear.

    A language that simplifies one dimension often transfers complexity to another. Research tracks where complexity moves, not just where it is reduced.

  6. Native expression matters.

    An implementation that fights its language produces different results than one written in its natural idiom. Both are studied.

  7. Measurement requires interpretation.

    A number without context is not evidence. Every measurement is accompanied by its workload, environment, and declared limitations.

  8. Maintainability is a first-class outcome.

    The cost of changing software over time is as important as its initial performance. Changeability is measured, not assumed.

  9. AI changes the economics of comparison.

    AI assistance can accelerate implementation, translation, and test generation. It does not replace verification or reproducible evidence.

  10. The research corpus must remain alive.

    Language versions change. Runtimes evolve. Findings are dated and revisited. A static benchmark is a historical artifact, not a current finding.

Choose with context. Build with evidence. Learn across languages.

Code is plural. Outcomes are shared.