The Three-Month Rule: A Pragmatic Approach to Non-Scalable Development

In the realm of startup innovation, many entrepreneurs often find themselves grappling with the notion of scalability. Paul Graham famously advocates for “doing things that don’t scale,” yet the intricacies of applying this philosophy in software development often remain unexplored.

After eight months of developing my AI podcast platform, I’ve crafted a straightforward framework: any unscalable approach I implement is given a three-month trial period. During this time, the hack must either justify its existence or be phased out.

Rethinking the Engineer’s Mindset

As engineers, we are typically conditioned to create scalable solutions right from the outset. We delve into design patterns, microservices, and architecting systems capable of managing millions of users, which is really more aligned with the big corporation mentality. However, in a startup environment, striving for scalability from day one can lead to wasted resources and time, focusing on users that are yet to come and problems that may never arise. My three-month rule encourages me to write simple, straightforward, and at times ‘bad’ code, allowing me to truly understand my users’ needs.

Current Infrastructure Hacks That Are Surprisingly Effective

1. Consolidating Everything on a Single VM

Currently, I run my database, web server, background jobs, and Redis on a single Virtual Machine costing $40 per month. Although there’s zero redundancy and I perform manual backups to my local machine, this setup has proven to be invaluable. It has provided insights into my actual resource demands much quicker than any capacity planning document could. I learned that my AI-oriented platform peaks at 4GB of RAM; the complex Kubernetes architecture I almost implemented would have been managing idle containers. Each time the system crashes, which has happened a couple of times, I gain real insights into what genuinely fails—typically not the components I anticipated.

2. Utilizing Hardcoded Configurations

For the time being, I’m operating with hardcoded configurations spread throughout my codebase, such as:

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

This may seem unprofessional, but it has a hidden advantage: immediate accessibility. I can quickly search all configurations across my codebase, track price changes via git history, and implement code reviews (even if it’s just me