I Designed a Physics Engine Before Asking Who Needed It (otw.)

I've already imagined someone typing import alle. The embarrassing part? I still can't confidently tell you why they'd want to.

ALLE (Analytical Lagrangian Learning Engine) is a physics engine I've been conceptualizing for robotics research. I've drawn its architecture in LaTeX, thought through the mathematics, and spent countless hours imagining how simulation, dynamics, and optimization could work together.

I absolutely love this stuff. I could lose an entire evening thinking about one problem in contact dynamics and consider it a fantastic evening.

But why would a researcher abandon MuJoCo or Drake for something I've built?

I recently watched Strategyzer's "The Value Proposition Canvas Explained," and it made that question much harder to ignore.

The video explores customer jobs, pains, and gains. It got me thinking about two researchers who might both complain about their simulation software. One could be struggling with unstable contact dynamics, while another might be spending days connecting tools and debugging scripts. They could describe the same frustration and need completely different solutions.

And here I am, dreaming up an entire physics engine.

I want to start sitting with researchers, looking at their actual workflows, understanding the workarounds they've learned to live with, and figuring out what genuinely costs them time. I'm beginning to wonder whether the most valuable thing I could discover about ALLE is something that makes me rethink the architecture I've grown so attached to.

Because I still want to build an extraordinary physics engine. That ambition hasn't changed.

But I want the problems researchers live with every day to have a much bigger say in what it becomes.

So here's something I'd genuinely love to ask other engineers and founders: Have you ever been so fascinated by what you were building that you almost forgot to question whether it deserved to exist? And what finally made you question it?