Plain Text Accounting (PTA)

Original Article ↗ Hacker News Discussion ↗

Adoption & Use Cases

  • Many use plain text accounting (PTA) tools like Ledger, hledger, Beancount, and GnuCash for:
    • Personal finances, budgeting, and net-worth tracking.
    • Self-employment and small business bookkeeping, including multi-LLC setups.
    • Portfolios, pensions, RSUs, and multi-currency balances.
    • Group trip cost splitting and ad‑hoc “who owes what” scenarios.
  • Some run their entire business accounting in PTA; others confine it to side analyses while official books live in traditional software.

Data Import & Bank Integration

  • Biggest friction: getting bank and card data into a standard format.
    • Manual CSV/PDF downloads from multiple institutions with MFA are time‑consuming; missing a few months causes “catch‑up” pain.
    • OFX exists but is inconsistently implemented; open banking is mentioned in some regions, but global coverage is poor.
    • Aggregators (Plaid, Teller, etc.) can help but have country limits, approval hurdles, and pricing/usage constraints.

Workflows, Tooling & Editors

  • Common pattern: don’t type journals directly.
    • Use spreadsheets, GUI tools, banking app comments, notes apps, or custom UIs, then scripts to emit Ledger/Beancount/hledger.
    • CSV-to-journal transformers (hledger’s CSV rules, Paisa, banks2ledger, custom scripts) are heavily used.
  • Editor/IDE support: Emacs modes, VS Code extensions, hledger-ui/web, org-mode tables, Makefile/justfile wrappers, and even CoPilot for autocompletion.

Benefits & Motivations

  • Key attractions:
    • Plain text, version control, scripting, and custom reports.
    • Full transparency and distrust of opaque SaaS/bookkeepers.
    • Flexible “virtual envelopes” and projections that influence real behavior (e.g., reduced eating out, better cash-flow awareness).
  • Some find that data older than ~1 year has low practical value; “starting over” periodically is normalized.

Limitations, Pain Points & Skepticism

  • Time cost: even optimized setups may take 30–120 minutes per week/month.
  • Poor bank statement descriptions limit auto-categorization.
  • Attachments (receipts, statements) are awkward in pure text, leading to folder conventions or external tools.
  • Some accountants refuse or struggle to work from PTA outputs; this pushes some users back to QuickBooks/Xero despite disliking them.

Automation, AI & ML

  • Users experiment with:
    • Rules-based keyword matching for categories.
    • ML/LLM-based mapping of transaction descriptions to accounts, sometimes outperforming fragile regex pipelines.
  • Concerns raised about fragility and privacy when sending financial text to cloud LLMs.

Performance & Scalability

  • Large ledgers (≈1M transactions) are possible; performance varies by tool.
    • hledger can handle them with noticeable but manageable processing time; other tools may struggle on the same data.
    • Many prefer splitting data into per-account/per-year files to keep things responsive.