BBright BBright

When the Bluebook Is Silent: Building a Best-in-Class Engine for Civil Forfeiture Case Names

January 2026 · BBright Editorial Note

This post is part of BBright’s Editorial Notes series: practical explanations of how the engine resolves hard Bluebook questions, especially where the rules under-specify real-world captions.

← Back to Editorial Notes

Civil forfeiture cases sit in one of the strangest corners of legal citation.

They are ubiquitous in federal practice, especially in criminal and constitutional litigation, yet the Bluebook offers only partial guidance on how to name them. The result is a quiet but persistent problem: even prestigious law reviews often disagree about how these cases should be cited — not because editors are careless, but because the rules simply don’t cover the edge cases.

Over the past several weeks, we undertook a deep dive into one of the most difficult subsets of forfeiture cases: in rem actions where the defendant is real property. What began as a modest cleanup effort ended with a complete re-thinking of how a citation engine should behave when the Bluebook runs out of road.

This post explains what we found, how we approached the problem, and the rule system we ultimately implemented.

The Problem: Real Property Defendants Don’t Behave Like Parties

Most case-name rules assume that the parties are people, corporations, or governments. Real property doesn’t fit neatly into any of those categories.

A single forfeiture caption might include:

All of that may be technically accurate — but none of it answers the editorial question a citation is meant to answer:

What is the shortest name that lets a human reader recognize this case?

The Bluebook tells us to prefer street addresses when available, but it does not tell us:

That silence is where inconsistency creeps in.

Our Methodology: Editorial Reality Over Abstract Rules

Rather than inventing rules in the abstract, we anchored our work in real-world usage.

1. We collected a corpus of hard cases

We assembled dozens of forfeiture cases from multiple circuits, deliberately focusing on:

2. We checked how serious law journals actually cite them

When the Bluebook was silent, we looked to practice — especially:

The striking pattern was not randomness, but convergence. Journals often disagreed in the margins, but they almost always agreed on the shape of a sensible short caption.

3. We encoded rules incrementally, backed by gold tests

Every insight was locked into a growing gold test suite. If a new rule fixed one case but broke another, it didn’t survive. The result is an engine that is conservative, explainable, and regression-resistant.

Why These Cases Fall Between Bluebook Rules

Real-property forfeiture cases expose gaps in the Bluebook because they combine doctrines that are usually separate:

The Bluebook gives us tools for each domain, but not for how they interact.

That means a good engine can’t just “apply rules” — it has to reason editorially, the way a careful human editor would.

The Rule Schema We Settled On

After working through the corpus, we arrived at the following hierarchy. These rules are now encoded directly into the engine.

1. Prefer a street address when it is the primary identifier

If a caption presents a street address as the main way the property is identified, we use it.

Addresses short-circuit all other logic when they are primary.

2. Treat “also known as” addresses as secondary

If an address appears only as an alias (“also known as”), we do not let it override a complete legal description.

This matches both editorial practice and the semantic signal in the caption itself.

3. If no address exists, prefer the smallest complete land identifier

We fall back to:

4. If no land identifier exists, use parcel count + smallest geography

When the only distinguishing feature is geography, we retain the parcel description and append the first geographic unit mentioned.

This avoids collapsing the case name into meaningless geography alone.

5. Jurisdiction-complete systems suppress extra geography

Some land-description systems are already complete.

In Washington, D.C., a square–lot identifier uniquely identifies property citywide. Adding “District of Columbia” does no editorial work.

This rule is jurisdiction-specific and deliberately narrow.

6. Ownership and claimant language is never part of the case name

We always drop:

This is true even when no better identifier exists. As a last resort, we output:

The Result: Transparent, Explainable, and Editor-Faithful

The goal was never to produce “the only correct answer.”

It was to produce answers that are:

If someone disagrees with an output, they can now see why the engine did what it did — and that disagreement can be about judgment, not mystery.

That’s the standard we think a modern legal citation engine should meet.

Questions or disagreements?

We’re happy to compare approaches and explain why the engine makes a particular editorial choice. Email support@thebluebooker.com.