PACER charges $0.10/page for public court records. EFF wants it gone.

What happened

The Electronic Frontier Foundation published a fresh call this week for Congress to abolish fees on PACER — the federal judiciary's Public Access to Court Electronic Records system. The argument is straightforward: court records are public documents, the public has already funded the courts through taxes, and charging $0.10 per page to read a docket entry is a paywall on democracy.

The specifics are damning. PACER charges $0.10 per page retrieved, capped at $3.00 per document, with a $30/quarter waiver for low-volume users. Sounds modest until you try to actually research a case. A single complex docket can run thousands of pages. Investigative journalists, academic researchers, and pro-se litigants routinely face bills in the hundreds or thousands of dollars to do work the First Amendment arguably guarantees them the right to do.

Meanwhile, PACER is a cash cow. The judiciary collects roughly $145 million per year in fees — far more than the cost of running the system. Surplus funds have been diverted to courtroom audio systems, flat-panel displays, and other line items that have nothing to do with electronic records access. A federal court ruled in 2020 (National Veterans Legal Services Program v. United States) that the judiciary had been illegally overcharging for years, but the fee structure itself survived the lawsuit.

Why it matters

This is not a niche civics complaint. It's an infrastructure problem with direct implications for anyone building tools on top of legal data.

The workaround ecosystem is already substantial. The Free Law Project's RECAP browser extension intercepts PACER document fetches and uploads them to CourtListener, a free public archive. Harvard's Caselaw Access Project digitized 6.7 million state and federal cases. Justia, Court Listener, and PlainSite scrape and re-publish what they can. Every one of these projects exists because the official system is gated — they're a multi-million-dollar volunteer effort to rebuild a public utility the government already built and is sitting on.

The AI angle is now unavoidable. Legal-tech startups training models on case law are doing it on whatever they can scrape, supplemented by paid commercial databases (Westlaw, LexisNexis) whose own corpora are built on top of public records. Harvey, EvenUp, Casetext (now part of Thomson Reuters), and dozens of others are racing to build domain models without a clean, complete, freely-licensed training corpus for federal litigation. A fee-free PACER with bulk API access would change that overnight — and probably commoditize a chunk of the legal-AI moat that depends on data access asymmetry.

The judiciary's stated reason for keeping fees is funding. The CBO scored the elimination of PACER fees in the Open Courts Act at roughly $2 billion over ten years — real money, but a rounding error in the federal budget and trivial compared to what's spent on, say, courtroom carpet. The deeper reason fees persist is institutional: the federal courts run PACER as a self-funded fiefdom and resist congressional oversight of how the surplus is spent. The Open Courts Act has passed the House twice (2020, 2022) and died in the Senate both times, opposed by a judiciary that, per the EFF's filing, lobbied against it directly.

Compare this to the SEC's EDGAR system. EDGAR is free, has a documented bulk API, ships filings as structured XBRL, and underpins an entire ecosystem of fintech and research tools. The financial-disclosure equivalent of a PACER document costs $0. Nobody seriously argues this has harmed the SEC's operations. The federal courts could have built EDGAR-for-litigation 15 years ago. They built a paywall instead.

What this means for your stack

If you ship anything that touches legal data — contract analysis, compliance tooling, case-law search, e-discovery, regulatory monitoring — you're already paying the PACER tax indirectly through your data vendor's costs. Watch the Open Courts Act in this Congress; if it passes, your data licensing math changes. Bulk download access would let you reduce dependence on Westlaw/LexisNexis for federal court material, which is a meaningful line item for legal-tech startups.

If you're building in civic-tech or journalism tooling, RECAP and CourtListener already expose REST APIs. CourtListener's bulk data dumps are free for research and include opinions, oral argument audio, and a growing slice of PACER docket entries contributed via RECAP users. Wire your scraper through RECAP rather than hitting PACER directly — you'll pay the same fees but the documents become public the moment you fetch them, multiplying the social return on your spend.

For AI/ML practitioners: federal court documents are one of the largest untapped high-quality English corpora outside of books and code. Adversarial briefs, judicial opinions, expert depositions — adversarial, structured, dense. The CourtListener corpus is already on Hugging Face in pieces. If PACER opens, expect a wave of legal-domain foundation models within 12 months, and expect the bar for legal-AI products to rise accordingly.

Looking ahead

The EFF's push lands in a Congress that has, twice, been a single Senate vote away from killing PACER fees. The political coalition is genuinely bipartisan — Sasse and Wyden co-sponsored the 2022 version — because access-to-records is one of the rare issues where libertarians, civil-liberties groups, journalists, and AI companies all want the same outcome. The holdout is the judiciary itself. The realistic 18-month forecast: another Open Courts Act attempt, more judiciary resistance, and continued growth of the parallel free archive built by RECAP and Free Law Project. Eventually the volunteer mirror will be more usable than the official system, at which point the fee structure collapses under its own irrelevance. The only question is whether Congress beats the volunteers to the finish line.