How BenchRecon briefs are built.
1. What we make
BenchRecon produces brief-form PDF exhibits — structured comparables reports drawn from public federal sentencing data. Each report is scoped to a specific offense-level / criminal-history-category cell and surfaces the cases the government's standard tool (JSIN) does not show: sentences the Commission collected but excludes from its published statistics.
The output is a litigation exhibit, not legal advice. It documents what the public record shows. How that record is used in plea or sentencing argument is counsel's work, not ours.
Every claim in every brief that can be tied to a specific source carries a data-source citation. The brief's footer reports the total count. If a number appears without a citation, it is labeled as estimated or approximate — never as sourced.
2. Where the data comes from
Active product data sources. Sentencing Snapshot and JSIN Exclusion Brief — the two products in general availability — draw exclusively from the U.S. Sentencing Commission's public individual-case datafiles. The Commission publishes annual datafiles covering every federal case sentenced under the Guidelines since FY 1987. The individual-case microdata includes offense level, criminal history category, final sentence, district, judge, and departure/variance notation. ussc.gov/research/datafiles. The Commission anonymizes case-level microdata by design — there is no per-case docket URL on ussc.gov; the verifiable source is the datafile itself.
Future product data sources. The waitlisted products (Officer Lookup, District Recon, Judge Snapshot, Compassionate Release, Forensic Foundation Pack) will draw on additional public datasets — CourtListener (free.law), RECAP (free.law/recap), FJC IDB (fjc.gov/research/idb), and CPDP (cpdb.co) for agency-level misconduct cross-reference. None of those sources is used in the active products today; they are listed here so waitlist customers can see the planned methodology.
3. The data-source citation convention
Every factual claim in a brief that derives from a specific record is paired with the USSC datafile vintage from which it was drawn. The Commission anonymizes case-level microdata by design, so verification is at the datafile level — not per-docket. External sources (statutes, opinions, secondary literature) are linked directly when cited. No paywalled citations are used.
The brief's footer reports a source-citation count. That number tells you how many claims in the brief are anchored to a checkable source. Our minimum threshold is one citation per substantive factual claim.
This convention exists because of Mata v. Avianca (S.D.N.Y. 2023) and the professional-responsibility consequences it put into permanent record — 1,227 documented hallucination incidents in 2026, with a record sanction of $109,700. BenchRecon ships zero generative content; every claim is an aggregated public record. If any data-source citation in our brief is broken or does not resolve to the USSC datafile, email us within 7 days and we will correct the brief or refund the purchase.
4. What we do not do
We do not generate argument text. We do not use large language models to synthesize legal analysis. The brief artifact is a structured data exhibit — numbers, distributions, and source citations, formatted for courtroom use. The JSIN Exclusion Brief includes a § 3006A factual-recital block: a fixed template with cell-specific values substituted in (not LLM-generated), labeled "edit before filing" in the brief itself. Counsel reviews and edits before any use.
We do not provide legal advice. Nothing in a BenchRecon brief constitutes legal advice, legal strategy, or legal representation. The brief is a research exhibit. Your judgment as counsel governs how — and whether — it enters the record.
We do not claim that any sentencing outcome in the data is binding, recommended, or appropriate for your client's case. The data is what the Commission collected. What it means for your client is a legal question, not a data question.
5. Refresh cadence
USSC annual datafiles are released each fall, typically covering the fiscal year ending September 30. We update our working dataset within 30 days of each annual release. The brief you purchase reflects the most recent available USSC datafile at the time of generation, which is identified in the brief header.
CourtListener / RECAP / FJC IDB refresh cadences will be documented when those sources enter active products.
6. UPL disclaimer
BenchRecon does not provide legal advice, legal representation, or legal services. The information presented in every brief is public-records aggregation drawn from the USSC public datafile. Nothing in any BenchRecon product constitutes an attorney-client relationship or a legal opinion. Your judgment as licensed counsel governs how this research is used, whether it is disclosed, and whether it is appropriate for your client's case. State UPL rules apply to any downstream use of this material.