Skip to content
View adamzwasserman's full-sized avatar

Block or report adamzwasserman

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
adamzwasserman/README.md

Adam Zachary Wasserman

Independent AI researcher and software architect. Founder of the Open Honest Foundation. Former CTO (IATA).

I build measurement instruments and standards for software and AI quality, and I run a pre-registered research program on the linguistic foundations of large language models. ORCID 0009-0002-8865-6583 · OSF osf.io/user/8t64r · openhonest.org · honestcode.software

What I build

  • Honest Framework: an architectural standard for software that is correct by construction: no Big State, pure-function dispatch tables in place of classes, no hidden state, named categories of defect eliminated by design. A linter is its operational definition. (Not a tool for explainable AI.)
  • Slop Audit: an open, reproducible instrument that scores an existing production codebase across eighteen dimensions mapped to SOC 2, NIST 800-53, OWASP ASVS, and ISO/IEC 25010, calibrated to detect the failure modes of AI-generated code at scale.
  • MÉTRON: a measurement-instrument family and no-code platform for controlled cross-linguistic AI research, governed by the Open Honest Foundation.

What I am researching

  • The Language-Only Hypothesis (OSF SJ48B): a pre-registered hypothesis that the emergent capabilities of large language models are properties of natural-language structure, not of neural-network architecture or scale, tested by controlled cross-linguistic transformer training.
  • The Linguistic Telescope (linguistictelescope.org): a project, a hypothesis, and a book treating an LLM as a telescope onto language, and language as a telescope onto the distributed human intelligence deposited into it.

Pinned work

  • fractal-language: the cross-linguistic training program behind the Language-Only Hypothesis.
  • babylm: Right Tool, Right Job, the BabyLM 2026 submission on French morphological efficiency at child scale.
  • wasserman-levin-2026: Polytope and Sphere Loss and Vector Invariance experiments.

Books

The Chaos Factory · Honest Code · The Robot in the Dark (in preparation).


Open Honest Foundation governs the Honest Framework, the Slop Audit, and MÉTRON. Everything here is by Adam Zachary Wasserman unless stated otherwise.

Pinned Loading

  1. fractal-language fractal-language Public

    Cross-linguistic training dynamics: evidence that scaling laws are language-contingent

    Python

  2. babylm babylm Public

    Born Speaking French: BabyLM 2026 submission — French morphological efficiency at child-scale

    Python

  3. wasserman-levin-2026 wasserman-levin-2026 Public

    Polytope Loss experiments: Can attention regularization synthesize morphological advantage?

    Python

  4. axioms axioms Public

    The 70% Rule: When Axiomatic Prompting Helps, and When It Hurts

    Python