Embracing the 3-Month Rule: A Pragmatic Approach to Non-Scalable Coding
In the tech world, few pieces of advice resonate as deeply as Paul Graham’s call to “do things that don’t scale.” However, while the sentiment is widely acknowledged, the methods of applying this philosophy, especially in coding, are less frequently discussed. Over the past eight months, while developing my AI podcast platform, I have crafted a personal framework that I like to call the “3-Month Rule.” This approach allows each unscalable solution a three-month trial period; after that, it either proves its value and receives the attention it deserves, or it is discarded.
The Dilemma of startup Engineering
As engineers, we often fall into the trap of creating solutions designed for scalability right from the outset. With concepts like design patterns, microservices, and distributed systems, it’s easy to get caught up in creating intricate architectures fit for massive user bases. However, this mindset can lead to wasted resources and excessive complexity, particularly in a startup environment where the focus should be on establishing a foundational understanding of user needs.
My 3-month approach compels me to produce straightforward, albeit “imperfect,” code that can be deployed swiftly. By doing so, I gain invaluable insights into what users genuinely require, rather than what I assume they need.
My Current Infrastructure Experiments: Simple Yet Effective
1. Consolidated Operations on a Single VM
I currently run everything—from the database to the web server—on a single $40/month virtual machine. While this setup lacks redundancy and relies on manual backups, it has provided me with real data about my resource needs more quickly than any capacity planning document could. I’ve discovered that my platform requires only about 4GB of RAM at peak times, revealing that a complex Kubernetes configuration would have been more hassle than benefit.
2. Hardcoded Configurations
My configuration settings are hardcoded directly into my codebase. For example:
python
PRICE_TIER_1 = 9.99
PRICE_TIER_2 = 19.99
MAX_USERS = 100
AI_MODEL = "gpt-4"
This approach might seem primitive, but it allows quick access to values and ensures that changes are easily tracked through version control. By minimizing the complexity of managing configurations, I save a significant amount of engineering time while still effectively implementing necessary updates.
