Those computers had to do differentials and other math calculations for small, sole purposes. Alot of the true technology was in the hardware. All the software did was read numbers from the hardware and send a signal to different hardware that then does most of the work. It was a complicated calculator
A modern application needs to keep track of positiin in the same manner for all items shown, relevant to each camera object, lit by another object at a position, the vector between them all for distance, and then use that to paint a 2d representation of the 3d emulated view which would be seen at a given position from that data, a bunch of flat texture files, and math. Then it does so 60 times a second or more for VR games. That's just to show you a frame alot more happens. Shaders, game logic, state management, job scheduling...
The rocket didn't need to figure out hundreds of curves it needed to take measurements on fuel usage, air drag, time, gyroscopes, etc. to figure out speed, direction, burn rate, and adjust these things in flight. It didn't need to do this all that quickly either. These are highschool math problems. Definitely a feat to do this all but something one very smart person can write the code for.
Modern applications do alot more than those rockets. Just for less useful purposes. So many devs touch it too. The language devs, the enfines devs, the library devs, and finally the game devs. Even a solo dev has hundreds of other developers code running with the most minimal engine and language.
I hate this comparison. Efficiency is a big problem, yes. We should care about it more and accept that we need to start making software slower, and more correct instead of using memory management solutions that obscure it all in a way that allows for every possible solution.
Its a complex problem but no its not comparable to the moon mission. You can't just go back to assembly, BASIC, and C directly managing memory and expect a 60 hour game to release every 3-5 years. Even with moderate solutions 3-5 years is a short timeframe when you get into the weeds of memory optimization.