[self-promotion] Structural Audit Dataset Of 137 Official Government PDF Forms (19,512 Form Fields, XFA Layers, Encryption & Field-naming Metadata — CC BY 4.0 CSV)

Disclosure ([self-promotion] per Rule 1): I build JustFill, a browser PDF form filler and Excel-to-PDF batch mail-merge tool. While debugging why official government PDFs filled in Chrome or macOS Preview frequently open blank in Windows Adobe Acrobat—or block packet merging—we ran a programmatic structural inspection across 137 federal, state, and international government form entries and released the full table and raw CSV under CC BY 4.0 (no login or email gate).

Dataset Access (CC BY 4.0, Direct Links)

• Direct Raw CSV (137 rows × 26 columns): https://justfill.app/data/justfill-government-forms-fillability-2026-09-02.csv
• Interactive Filterable Table & Methodology: https://justfill.app/research/are-government-forms-fillable

Scope & Schema (26 columns)

Each row inspects a blank public government PDF form (IRS, USCIS, SSA, VA, State Dept, DOL, CMS, California Judicial Council, UK HMRC, etc.) using pypdf and structural dictionary checks: – Identity & Source: slug, formName, issuer, group, country, status (134 ok, 2 blocked, 1 no_pdf), pdfUrl, edition, pdfVersion, producer – Geometry & Widget Counts (628 pages, 19,512 terminal fields across 133 distinct PDFs): pages, sizeKb, fields, textFields, checkboxes, radios, choices, signatures, textLayer – Structural Locks & Accessibility / Naming Metadata: xfa, encrypted, modifyDenied, annotateDenied, namedFieldsShare (share of terminal widgets with semantic human-readable names rather than internal codes like f1_01[0]), verdict (xfa_hybrid, acroform, flat), note

Summary Findings in the CSV (134 analyzed entries)

1. **114 / 134 (85.1%) are xfa_hybrid PDFs:** They embed an Adobe LiveCycle /XFA XML packet alongside standard /AcroForm widgets (0 in this sample were pure dynamic XFA shells). Writing to /AcroForm without stripping /XFA (https://justfill.app/xfa-pdf-to-fillable-pdf) causes desktop Acrobat to prioritize the empty <xfa:datasets> stream.
2. **63 / 129 (48.8%) of interactive forms have namedFieldsShare < 0.10:** Terminal widgets use opaque internal identifiers such as topmostSubform[0].Page1[0].f1_1[0] instead of semantic labels (with human descriptions stored only in /TU tooltip strings or adjacent page text).
3. 44 / 134 (32.8%) set modifyDenied = true (16 also set annotateDenied = true): Empty-password owner encryption locks page assembly and stamp/annotation workflows despite opening without a user password.
4. **5 / 134 (3.7%) are flat PDFs (0 interactive fields):** Including DOL WH-347, State Dept DS-3053, DS-64, USPS Form 1583, and the UK HMRC Starter Checklist.

Happy to answer any questions about the extraction script, PDF /AcroForm vs. /XFA internals, or additional agencies worth adding in the next crawl.