Most bills that fail, fail on their language.

I spent over twenty-five years drafting legislation for the House of Representatives in the Massachusetts General Court. The defects I look for are the ones that survive a vote: the definition doing two jobs, the effective date that never reached the engrossed print, the “including” a court will read as exhaustive where the drafter meant it to illustrate.

Correctly drawn is not a slogan. It is the standard a bill must meet before it can pass to engrossment in the General Court — examined and certified as correctly drawn. For most of my career that certification was the job. This site is what came of it.

Services

Three things I do.

Each begins the same way: with the text, read cold, the way a judge will read it.

Drafting and review

Bills, amendments, home rule petitions. Before filing, or before a floor vote when the language has changed and nobody has read it cold.

Statutory analysis

What the text will be held to mean, and where it is ambiguous enough that either reading could win. With the canons named and the cases behind them.

Corpus work

Comparison sets for questions that span dozens of acts, so the drafting starts from what the General Court has already done rather than from a blank page.

Bias Spectrum Audit

Bias you can see is bias you can fix. Where a statute or bill sits between the worst version that would still look fair on paper and the most protective one — decision by decision, with the language that closes the gap. A commissioned engagement. Five published audits.

The method

Ambiguity is a condition to understand, not always a defect to fix.

Some ambiguity is deliberate; it is how bills pass. The job is to make sure the sponsor knows what the ambiguity creates, who bears the cost of it, and what a court will do with it. I built a system to do that work at scale, as a set of Claude skills that run in sequence: a lint for the phrases that generate litigation, a detector for the ambiguities themselves, an adversarial simulator for how each would be argued and resolved, and a scorer for the definitions that fix them.

  1. Step one

    Lint

    The cheapest check first. Relational phrases, bare time references, unclear cross-statutory references.

  2. Step two

    Detect

    Definitional conflicts across chapters, regulatory overlays, undefined terms. Every alert dispositioned with a reason.

  3. Step three

    Simulate

    The competing readings, the canons behind each, the predicted resolution — calibrated against what the SJC has actually done.

  4. Step four

    Score

    A proposed definition rated for coverage, precision, clarity and conflict against 1,800 existing ones.

Contact

Send me the language.

A bill, a section, an amendment, a petition — as filed or as it stands. I will tell you what a court would do with it, and what I would change. Email is the best way to reach me.