Correct by Design vs Correct by Coincidence
Quick Answer: Code is "correct by coincidence" when it relies on duplicated, fragile details—like matching magic strings—that just happen to align perfectly to make the application work. Code is "correct by design" when you structurally enforce stability, such as centralizing shared values, ensuring the system cannot easily break from simple typos.
You can build an application a hundred different ways. From the outside, the users are happy, the behaviors are exactly what the product manager asked for, and your entire test suite is glowing green. But on the inside, the reality can be vastly different. The codebase might be a solid domain model accurately reflecting the real world—or it could be held together with tape, string, and pure luck.
This brings up a question I ask myself a lot: Is this codebase correct by coincidence, or is it correct by design?
What does "correct by coincidence" mean in programming?
Correct by coincidence means your software functions properly only because loosely connected, duplicated details happen to match up perfectly at runtime. If any single unlinked detail changes or a developer makes a tiny typo, the entire system breaks.
Imagine a city planner who builds a bridge where the support beams aren't bolted together, but they stay standing simply because the wind happens to be blowing with equal force from both sides. It works today, but it is incredibly fragile.
Let's look at a concrete software scenario. Imagine your team is building a microservice that needs to read a configuration file—let's call it config_init.json. This file is absolutely critical and needs to be read in ten different places across your application.
If you write the application so that in each of those ten places you manually type out the magic string "config_init.json", your code is correct by coincidence. It is purely a coincidence that each of those ten scattered locations happens to contain the exact same magic string required to successfully read the file. The moment someone renames the file but only updates nine of those ten strings, the application blows up.
How do you write code that is "correct by design"?
Writing code that is correct by design involves structuring your system so that fragile duplications are eliminated and invalid states are hard to represent. You establish a single source of truth—like a centralized constant—so correctness is enforced structurally rather than relying on human memory.
To fix the coincidental correctness in our configuration file example, the solution is remarkably simple. You extract that magic string, pull it into a single shared variable, and reference that variable in all ten locations.
// Correct by coincidence
const data = readFile("config_init.json");
// Correct by design
const CONFIG_FILE_PATH = "config_init.json";
const data = readFile(CONFIG_FILE_PATH);
By doing this, you've moved the responsibility of correctness away from human memory and handed it to the structure of the code itself. If the filename needs to change, you update it in exactly one place. If you typo the variable name, the compiler or linter yells at you immediately.
What are the main risks of coincidental correctness?
The primary risk is creating a brittle codebase where minor updates trigger unpredictable, cascading failures. This ultimately destroys developer confidence, slows down feature delivery, and turns routine refactoring into a high-risk operation.
When a codebase relies on coincidence, you will typically notice a few recurring symptoms:
- Hidden Coupling: Unrelated modules become secretly dependent on each other because they share the same hardcoded values.
- Refactoring Fear: Developers become terrified to change file names, database keys, or route paths because they don't know where else those strings are secretly hiding.
- Silent Failures: Because coincidences aren't structurally enforced, regressions often are not caught until runtime, sometimes making it all the way to production.
- Onboarding Friction: New engineers have no way of knowing the unspoken rules of the codebase, making it easy for them to accidentally break things when adding new features.
Frequently Asked Questions
How do I spot code that is correct by coincidence?
Look for magic strings, magic numbers, or duplicated logic that appears in multiple files without a shared reference. If you find yourself using global search-and-replace to safely rename a concept in your application, you are likely looking at coincidental correctness.
Is "correct by coincidence" the same as technical debt?
It is a specific type of technical debt. While technical debt can encompass anything from poor overall architecture to missing documentation, coincidental correctness specifically refers to fragile implementation details that rely on unlinked elements perfectly aligning to function.
Can unit tests prevent coincidental correctness?
Not always. If you write your tests using the same scattered magic strings as your implementation code, your tests will pass, but the underlying design flaw remains. Tests prove the behavior works, but structural design proves it will keep working when the system changes.
