Ask HN: Any active COBOL devs here? What are you working on?

Hacker News Discussion ↗

Where COBOL Is Actively Used

  • Heavy use in banking, insurance, government (tax, pensions, unemployment, health insurance, lotteries), education payroll, and healthcare “patient accounting.”
  • Common patterns:
    • Nightly/batch jobs (ACH, claims, billing, pensions, inventory, replenishment).
    • Online transaction processing via CICS/IMS.
    • Backends: DB2, IMS/DL/1, VSAM, sequential datasets, sometimes SQL databases.
  • Not strictly mainframe: PeopleSoft, vertical ERPs, and products like Global Shop use Micro Focus/AcuCOBOL on Windows, Linux, AIX, etc.

Modernization and Migration Efforts

  • Many are reverse‑engineering COBOL to:
    • Move to Java/.NET/TypeScript/low‑code/COTS systems.
    • Consolidate multiple mainframes or re‑host COBOL (e.g., Micro Focus on x86/private cloud).
  • Real difficulty is not COBOL syntax but:
    • 30–40 years of undocumented business logic.
    • Tight coupling and huge “uber‑monolith” systems.
  • Multiple stories of failed or massively over‑budget migrations (SAP/ERP replacements, AS/400 rewrites), leading some orgs to build new systems in‑house and use old devs mainly as domain historians.
  • Bridging tech: host gateways, custom Java/JS adapters, APIs, Kafka, DB‑driven integration.

Salaries, Careers, and Culture

  • Several reports that routine COBOL roles pay below typical software salaries; some consultancies hire cheap juniors while selling “COBOL scarcity.”
  • High rates exist mainly for niche experts on idiosyncratic, critical systems.
  • Many mainframe/COBOL devs are older, long‑tenured, extremely business‑process‑oriented, and often not active in online tech communities.
  • Cultural gaps noted:
    • Less exposure to modern security, tooling, and cloud paradigms.
    • Very strong system reliability, efficient keyboard‑driven workflows, and deep domain knowledge.

LLMs, Tooling, and Learning

  • Several commenters successfully used LLMs to generate or analyze COBOL, especially boilerplate batch code; others stress the surrounding ecosystem (mainframes, control languages, JCL) remains the hard part.
  • AI is being explored for code migration and test generation but seen as requiring heavy human oversight.
  • Learning/on‑ramping resources mentioned: IBM Z Xplore, Coursera mainframe courses, IBM Redbooks, mainframe emulators, and formal training programs.

Developer Experience and Perception

  • COBOL characterized as:
    • Verbose, boilerplate‑heavy, excellent for structured record processing and database‑centric logic.
    • Capable of surprisingly modern UIs (SCREEN SECTION, GUI controls in some compilers).
  • Opinions range from “boring, hated it” to “actually a decent 4GL‑like environment; not as bad as its reputation.”