Embracing the 3-Month Rule: A Framework for Building Non-Scalable Solutions

In the tech world, we often hear the mantra from Paul Graham: “Do things that don’t scale.” However, the challenge lies in translating this advice into actionable strategies, especially in software development. Throughout my journey of developing an AI podcast platform over the past eight months, I’ve formulated a straightforward yet effective framework: any hack that isn’t inherently scalable is given a lifespan of three months. After this period, we either enhance it based on proven value or phase it out entirely.

As engineers, our instinct is to deliver scalable solutions from the outset. We focus on design patterns, microservices, and distributed systems, all tailored for high user volumes. Unfortunately, this is often the mindset of larger organizations, while startups face unique constraints. In the early stages of a startup, crafting scalable code can become a costly form of delaying crucial progress. Instead of concentrating on future users, my three-month rule encourages me to write straightforward, albeit “imperfect,” code that actively delivers results and uncovers real user needs.

Current Hacks and Their Hidden Wisdom

1. Unified Infrastructure on a Single VM

By operating my database, web server, background processes, and Redis all on a single $40/month virtual machine, I’ve opted for simplicity over redundancy. While this approach may seem risky, it has provided clarity about my actual resource requirements within just two months. Initial assumptions about needing complex setups like Kubernetes were quickly dispelled as I identified that my platform’s peak usage only demands 4GB of RAM. Moreover, every time the system crashes, I gain valuable insights into what issues truly occur, which are often unexpected.

2. Hardcoded Configurations

Every configuration is hardcoded directly into the application:

python
PRICE_TIER_1 = 9.99
PRICE_TIER_2 = 19.99
MAX_USERS = 100
AI_MODEL = "gpt-4"

While some may view this as a limitation, it actually streamlines my workflow. Searching through the codebase for configuration values takes mere seconds, and each adjustment is meticulously documented in version control. Instead of developing a configuration service that would consume a week of engineering time, I’ve adjusted values three times in three months—resulting in far less overhead.

3. Utilizing SQLite in Production

Surprisingly, I’m employing SQLite to support a multi-user application, with my entire database encompassing