Embracing the 3-Month Rule: A Practical Approach to Non-Scalable Solutions in Coding
In the world of startups and programming, there’s a well-known piece of advice from Paul Graham that encourages developers to embrace the philosophy of “doing things that don’t scale.” However, implementing this principle in coding practices often raises more questions than answers. After spending eight months developing my AI podcast platform, I’ve created a straightforward framework that I call the “3-Month Rule”. This approach allows me to explore unscalable hacks for a limited time, after which they either transition into well-built solutions or are discarded.
As engineers, we are frequently conditioned to focus on creating scalable systems from the outset. We immerse ourselves in sophisticated architectures like microservices and distributed systems designed to handle vast numbers of users. While this is essential for large organizations, startups often find that pursuing scalability right away can lead to unnecessary complications. My 3-month rule challenges this conventional thinking by encouraging me to write straightforward code that allows for rapid iteration and real user feedback.
Current Infrastructure Hacks: Lessons from Iteration
1. Unified Virtual Machine Deployment
I’m operating all my services—database, web server, background jobs, and caching—on a single virtual machine, costing just $40 per month. Although this setup lacks redundancy and relies on manual backups to my own machine, it has provided crucial insights. In just two months, I’ve grasped my platform’s actual resource needs and discovered that it peaks at 4GB of RAM. The elaborate Kubernetes architecture I was contemplating would only have managed idle containers.
Whenever my system crashes, I gain valuable insights into real faults—often different from what I initially anticipated.
2. Directly Hardcoded Configurations
Instead of incorporating configuration files or environment variables, I’m using hardcoded constants throughout my codebase, such as:
python
PRICE_TIER_1 = 9.99
PRICE_TIER_2 = 19.99
MAX_USERS = 100
AI_MODEL = "gpt-4"
Yes, this means redeploying the application for any changes, but there’s hidden efficiency here. Quick searches allow me to locate any configuration value in seconds, and tracking price changes is as easy as checking my git history. With only three changes in three months, the time saved far outweighs the effort of developing a dedicated configuration service.
3. SQLite as a Production Database
I opted for SQLite
