Software can be finished

Software that is feature-complete, stable, secure, and has no runtime dependencies — is not only possible, but worth striving for in certain contexts.

🔍 What “Finished” Means

According to Ross, finished software:

• Is feature complete — it does what it needs to do, nothing more.

• Works reliably and securely.

• Has no runtime dependencies (except maybe an interpreter).

• Doesn’t require future updates unless the environment changes.

He acknowledges this isn’t always practical or desirable — especially in commercial software — but argues it’s a useful mental model.

🕹️ Real-World Examples

1. Nintendo Gameboy: Its OS and games haven’t changed in 35 years — and they still work.

2. Job Sheet Manager: A Pascal app Ross built for his parents’ business that ran for 17 years without updates.

3. Embedded Systems: Appliances like microwaves and thermostats run software that never gets updated.

4. Small Web Apps: Ross’s own projects like a Scrabble clone and chatbot library are intentionally “finished.”

🛠️ Lessons for Developers

• Keep scope small and fixed: Simplicity leads to durability.

• Reduce dependencies: Fewer moving parts = fewer failure points.

• Produce static output: Avoid reliance on external services.

• Invest in quality upfront: Linting, testing, and clear requirements reduce bugs.

• Design for longevity: Think about how your software might survive without you.

💡 Final Thought

Ross isn’t saying all software should be finished — just that thinking about what “finished” looks like can lead to better design, fewer bugs, and more resilient systems.