I spent the last few months researching the economics of software engineering. Specifically, what happens to costs when a product grows past what one team can maintain.
The same pattern kept showing up. Teams spend more time coordinating than building. Changes in one module break unrelated parts of the system. Features take twice as long as they should, not because the code is hard to write, but because the code is coupled to things it shouldn't be coupled to.
I call it the Sync Tax. It's a multiplier on every engineering hour. A multiplier of 2.0 means everything costs twice as much as it should. A multiplier of 4.0 means you're burning most of your budget on coordination and firefighting, not on output.
I built a small API around it. You plug in a few numbers about your codebase and team structure, and it returns the multiplier plus a dollar figure. No credit card needed for the free tier.
There's also an MCP server if you want to hook it into Claude, Cursor, or any agent.
https://complexity-cost-calculator.beamercloud.com/
I'm curious what numbers people get when they run their own teams through it.
Top comments (2)
Putting a number on complexity is valuable precisely because complexity is the cost everyone feels and nobody measures - it shows up as slower features, more bugs, longer onboarding, and "don't touch that file," but it stays invisible on any dashboard, so it never gets prioritized against shiny new work. A concrete metric gives teams a way to argue for refactoring in business terms instead of vibes, which is the only way that work ever gets funded. The hard part, and where I'd love to know your approach, is making the metric trusted: cyclomatic/cognitive complexity, coupling, churn-vs-complexity hotspots all capture real signal but each is gameable and none alone equals "cost," so the credibility comes from how you combine them and whether the score actually predicts pain.
This resonates because it's the same instinct behind why I obsess over cost in Moonshift, the thing I build - a multi-agent pipeline that takes a prompt to a deployed SaaS, where I treat per-build cost as a first-class measured number (route to the cheapest capable model, cache aggressively, so a full build lands ~$3 flat) instead of an invisible bill. Measure the thing you want to control. First run free, no card. Genuinely useful API. What's the underlying metric - cognitive complexity, or a churn-x-complexity hotspot model? And have you validated that a high score actually correlates with where teams report the most pain? That validation is what turns a number into a decision tool.
Some comments may only be visible to logged-in visitors. Sign in to view all comments.