The following is a retrospective on the current job market for Software Development. The author raises an important question about the rise of new AI tools that can replace entry level (junior) developers and the problem of keeping great software library/API discovery, when there is so much new source-code being produced every day.
https://matt-rickard.ghost.io/the-implications-of-more-developers
There are 94M developers on GitHub.
What software will need to be built if there is an order of magnitude more developers?
Software development can be a highly leveraged profession – e.g., the average developer at Meta supports at least 100,000 users. And many undifferentiated but critical parts of the software stack (e.g., auth) will be outsourced to a SaaS provider. Yet, I believe that the trend will continue, and we’ll have even more developers in the future.
- Companies outside the tech sector hiring a significant number of developers (JP Morgan has 40,000+ developers).
- Better tooling (i.e., GitHub Copilot) lowers the bar for entry-level developers.
- Tangential technical roles for those who can’t program in a general-purpose language but know a specific niche (SQL for Data Analysts, configuration for Operations, vulnerability scanning for Security Analysts).
- More students studying CS in college, better recruitment pipelines (not just referral-based).
So what are the most important problems when there are 200M+ developers? An open question, but a few thoughts.
- How can we efficiently share code? Package managers must evolve to include better dependency management, security, and configuration.
- How do we discover new code? Discovery for APIs, libraries, and other software.
- Developer-first software – developer tools that increase developer productivity
- Minimal viable software – software that requires the narrowest set of skills to work with. More developers will simply be programming against a configuration (Kubernetes/YAML, dbt/YAML, etc.).