Embracing the 3-Month Rule: A Guide to Unconventional Coding for Startups
In the startup ecosystem, one piece of wisdom often echoed is Paul Graham’s advice to “do things that don’t scale.” While this principle resonates, especially for early-stage ventures, the challenge lies in how to effectively implement it in the realm of coding.
Having dedicated the last eight months to developing my AI podcast platform, I’ve formulated a straightforward approach: every unscalable solution is granted a lifespan of three months. Within this timeframe, it must either demonstrate its worth and transition to a structured build or be phased out.
As engineers, we are conditioned to design our systems with scalability in mind from the very beginning. Concepts like design patterns, microservices, and distributed architectures appear enticing as they suggest our systems can comfortably accommodate millions of users. However, this perspective often aligns with large corporations more than nimble startups. In a startup environment, pursuing scalable solutions too early can become an expensive form of procrastination, focusing on hypothetical users and problems that may never materialize. This three-month framework encourages me to create straightforward, albeit imperfect, code that can be deployed quickly, allowing me to ascertain what users genuinely need.
My Current Infrastructure “Hacks” and Their Strategic Value
1. Consolidating Everything on a Single Virtual Machine
All components—database, web server, background jobs, and caching—run on a single $40/month virtual machine. This means absolute minimal redundancy and manual backups to my local machine.
Why is this a clever approach? In just two months, I’ve gained better insights into my actual resource requirements than any capacity planning document could provide. My so-called “AI-heavy” platform peaks at just 4GB of RAM. The Kubernetes infrastructure I nearly built? I would have been simply managing dormant containers.
When the server experiences outages (which has happened twice), I receive genuine data regarding the failure points. Surprisingly, it has never been what I initially anticipated.
2. Using Hardcoded Configuration Values
python
PRICE_TIER_1 = 9.99
PRICE_TIER_2 = 19.99
MAX_USERS = 100
AI_MODEL = "gpt-4"
Forget configuration files or environment variables. Instead, constants are directly embedded in the code. Modifications necessitate a redeployment.
What’s the advantage here? I can swiftly search my entire codebase for any configuration value. Each
