Why Writing Correct Software Is Hard
Web page · https://pron.github.io/posts/correctness-and-complexity
First seen by Oriel on .
AI summary
This extended talk argues that complete software correctness is inherently costly because computability and complexity results place fundamental limits on understanding and verifying arbitrary programs.
- Halting is undecidable, and reductions extend this limit to program-state reachability and every non-trivial property of the partial function a program computes.
- Busy Beaver-style functions linking program size to maximum halting time or minimal termination-proof size are non-computable and cannot have known computable universal upper bounds.
- For six-state, two-symbol Turing machines, the maximum halting runtime is already known to exceed 7.4 × 10^36534 steps.
- Stefan O’Rear built a 1919-control-state Turing machine whose termination is independent of ZFC, comparable in size to a Python program of 100–200 lines.
- The article treats algorithms as physically carried-out processes and argues that decidability alone is insufficient, because finite verification may still require infeasible time and effort.
#computability theory #complexity theory #software verification #formal methods
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 365, latest observation ; +348 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 183, latest observation ; +182 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hacker News | 50465352 | https://pron.github.io/posts/correctness-and-complexity Read discussion |