Reality First, Architecture Next
12-sep-2026 · Goa
I read an article where Sivers was building his dream house without predicting it. His plan was simple. He was living on his plot in a cabin and building only things that were absolutely essential. Instead of imagining the finished house first, he allowed everyday reality to tell him what the house needed to become.
This approach caught me in my tracks.
The process we usually jump into while building anything is architecture first, interpretation next, testing the features and functions, and then moving towards reality. We give way too much importance to architecture. And rightly so, because we want to imagine how the future looks today. That is gratifying. While we are engaged with the question, “How is the project going?”, we go back to the architecture and start working backwards.
Fuck No.
I had already shelved a project because I was building from architecture first. The entire project revolved around the end rather than the beginning. Midway, when the plumbing had to be fixed, tracing it back became a nightmare. I had to tear the whole code down just to understand what had happened.
Finally, I shelved the project because the cost of architecture first was way too much, especially considering that I am a solopreneur. I needed a first principle to hang on to and understand its second-order consequences. What is the cost or debt accrued at each stage, and can I reduce it before it becomes expensive?
The solution I used was simple. I inverted my thinking.
Reality first. Evidence next. Test the assumptions and false positives. Build small, bounded Work Units. Stress-test them. Look into the edge cases. Then build the architecture.
This is not an argument against architecture. Architecture matters. But it should be earned by evidence.
The first question becomes:What exactly is my evidence corpus? In practical words, what do I actually have in hand?
This drastically reduces the debt that comes with architecture first. Imagine building a beautiful home where the architect has sweetly hidden all the wiring and plumbing inside. Everything looks great until, after a few months or years, things go south. Then you need to dig the whole thing up just to find where the problem began.
When shit hits the ceiling, and it hits more often than not, I need to be able to trace the problem back to its origin, which is reality.
A reality-first system keeps the plumbing traceable. (and the fix needs to cheap and durable) This became the basis of the Engineering Operating System we developed for my new project, 07 Stealth. We bound the evidence first, interpreted it next, tested our reasoning and only then allowed it to influence the architecture.
We divided development into small Work Units and placed gates around each one. These gates protect what we are building from the smallest instance possible. Unsupported assumptions, unnecessary features and invented certainty cannot quietly slip into the product. ~Ravi Wiki~ became our smaller laboratory. It allowed us to test principles such as evidence preservation, authority, reproducibility, failure recovery and continuity at a much lower cost. We did not copy its architecture into the bigger project. We transferred only the principles that survived contact with reality.
When I adopted reality first, I had to stay with it for a longer time. It was like pitching a tent and sleeping on the plot before building anything permanent.
This is a little less elegant. However, it is a durable way.
Most things that last were built this way. Kingdoms were built this way. Civilisations thrived this way. They responded to the terrain, the weather, the threats and the needs of the people before their structures became permanent.
So I adopted the principle of Reality First and Architecture Next. This is the maxim I hold on to. Now, whenever I consider building a new feature, I ask myself: Can this work without having this feature?
If the answer is yes and the feature is merely decoration, can it. This has also helped me avoid the trap of building for appearances. Unfortunately, the modern world races towards appearances first. Architecture looks impressive. Reality usually does not. But reality is where all the useful information lives.
My plan is simple. Build small Work Units. Put gates around each unit. Preserve the evidence behind every important decision.
Fix needs to be cheap. Solutions need to be portable. Is this the best approach? I do not know. It works for me. I come from the angle that I need to be able to trace and fix something without breaking the whole house down.
So now I call this having your own Engineering Operating System. For builders and entrepreneurs, it is perhaps a useful way to think about how we create things, especially when we are building with limited time, limited money and increasingly powerful AI tools. It is an evolving state.
I learn this way. 🙂 PS: What prompted me to write this? I met Sachin, a super-interesting independent scientist working on quantum physics, particularly light and its effects on humans. His approach was close to what I had been doing, and our conversation made me want to mint this essay. We met at Ondo in Goa. BTW, Ondo means waves in spanish. 🙂
2026-9-12, Goa
You may like reading; Why I Aim to Be a Zero