Embracing the 3-Month Rule: A Pragmatic Approach to Non-Scalable Coding
In the world of startups, we often hear the mantra coined by Paul Graham: “Do things that don’t scale.” But how does this advice translate into actionable steps for developers? After eight months of developing my AI podcast platform, I’ve crafted a practical framework that revolves around the principle of giving every unscalable solution a lifespan of three months. If it demonstrates value, it gets the opportunity to evolve; if not, it is discontinued.
Rethinking Scalability in startup Culture
As engineers, our training often emphasizes the importance of crafting scalable solutions right from the start. We’re well-versed in concepts like design patterns, microservices, and distributed architectures—systems designed to accommodate millions of users. However, this focus on scalability can become a liability in a startup environment, where it may lead to investing resources into optimizing for hypothetical users.
Adopting my three-month rule has enabled me to focus on creating straightforward, albeit imperfect, code that can be deployed quickly. This hands-on approach allows me to understand audience needs in real time, enabling more informed decisions moving forward.
Current Low-Overhead Strategies I’m Exploring
1. Unified Server Architecture
I’m currently running my entire infrastructure—database, web server, background jobs, and caching—on a single $40/month virtual machine. While this setup lacks redundancy and relies on manual backups, it has provided invaluable insights into my actual resource requirements.
In just two months, I’ve realized that my platform’s peak usage is around 4GB of RAM. The elaborate multi-container Kubernetes configuration I almost implemented would have only served to manage unutilized resources. Plus, when issues arise (and they have), I receive critical feedback on what requires attention.
2. Simplified Configuration Management
My approach to configuration management is intentionally rudimentary: everything is hardcoded. Values like pricing tiers and maximum user counts are constants directly embedded in the code. Sure, this means that every change necessitates a redeployment, but it also allows for quick searches and a clear history of modifications through version control.
Building a separate configuration service would require a week of development. In contrast, I’ve only adjusted these values three times in the last three months—an investment of mere minutes compared to the extensive time it would have taken for a complex solution.
