The Open Bankruptcy Project started when a Chapter 11 debtor in Kansas City taught himself Python to investigate his own attorney. It grew into a national-scale court data research platform - built entirely on publicly available federal court records, including PACER data accessed through infrastructure that SCALES helped make possible.
Using FJC public release data (37.9M rows, both the IDB main and supplementary files), we cross-referenced two fields that had never been systematically compared: prior filing history and case disposition.
Finding: 391,951 bankruptcy cases involving prior filers received a discharge with no evidence of 11 U.S.C. § 1328(f) eligibility verification. This represents 27.4% of all prior filers with known dispositions - a national compliance gap hiding in the government's own data.
The aggregate data showing the problem is free (FJC). The verification data that would confirm actual violations is behind the PACER paywall. That's the access-to-justice problem in one example: a federal rules committee is now acting on data that requires paid access to verify.
A non-lawyer used publicly available court data to identify a systemic compliance gap, and a federal rules committee accepted it for formal consideration. That pipeline - from raw PACER data to policy action - is what OBP exists to replicate at scale.
All tools are open-source Python, running against SQLite databases built from FJC and PACER data.
| Capability | What It Does |
|---|---|
| Attorney Scorecard | Multi-dimensional analysis of any bankruptcy attorney: dismissal rates, timing patterns, fee extraction, client outcomes vs. district norms |
| Blind Outlier Detection | Identifies statistical anomalies without specifying a target - the tool finds the outliers, not the user |
| ML Mill Probability | Random forest classifier trained on known mill indicators: case volume, dismissal timing, fee patterns, prior-filer rates |
| District Comparison | Benchmarks any attorney or firm against their district's baseline across 15+ metrics |
| 1328(f) Screener | Live web tool (1328f.com) - checks individual cases for discharge eligibility violations. Ranking nationally for "bankruptcy discharge screener" |
| Deep Docket Mining | Automated PACER docket analysis: extracts fees, timelines, dispositions, and attorney behavior patterns from raw HTML |
| National Audit Tool | 9-screen Madoff-inspired audit across all 94 districts: identifies firms with high volume + high dismissal + high fees simultaneously |
139 domains. 2,353 pages. Zero hosting cost (all GitHub Pages). Built in under three weeks. Every major bankruptcy statute, concept, and consumer question has a dedicated educational site with FAQ schema, cross-links, and plain-language explanations.
Key properties:
Google Analytics (3/26 - 4/2): 6,007 sessions from 5,992 users in the first 7 days of tracking. Peak day: 1,320 sessions.
OBP is a use case that demonstrates the full SCALES thesis: free court data, enriched with computational tools, in the hands of the public, producing policy outcomes.
| SCALES Mission | OBP Implementation |
|---|---|
| Make federal court records free | 1328(f) screener serves free case lookups; all tools open-source |
| Enrich records with NLP/AI | 347 Python tools extract structured data from raw PACER filings |
| Enable systematic research | 4.9M-case FJC database; reproducible methodology anyone can run |
| Bridge research and policy | 26-BK-3 accepted by federal rules committee |
| People-first court data framework | 139 consumer-facing educational sites translating court data into plain language |
On April 15, 2026, the Advisory Committee on Bankruptcy Rules meets in Charlotte, NC. Agenda Item 4C considers amendments to Rule 4004 - and the Consumer Subcommittee's memo cites Suggestion 26-BK-3 by name, noting that "some courts, in the absence of a motion, nevertheless grant discharges to debtors who are ineligible under 1328(f)."
The memo recommends the committee consider asking the Federal Judicial Center to study courts' discharge practices. OBP has already conducted that study using the FJC's own public release data - 37.9 million rows, cross-referencing prior filing history against case disposition across all 94 federal districts. The 391,951-case finding provides the empirical foundation that the committee's own research arm has not yet produced.
Two sitting federal judges (Connelly and Kahn) independently submitted the parent suggestion (25-BK-N) arguing Rule 4004 is inconsistent with the statute. OBP's data was cross-referenced to their suggestion and provides the national-scale evidence supporting their judicial concern.
The founder of OBP is a small business owner in Kansas City with no law degree, no CS degree, and no academic affiliation. He taught himself Python in 2024 while navigating a Chapter 11 bankruptcy as a pro se debtor. He built the entire platform - 347 tools, 4.9 million cases, 139 domains, a federal rules suggestion - in under 12 months, working nights and weekends while running his business and managing an active bankruptcy case.
You May Also Find Helpful