Btw, the reaction wheels are for attitude control of the missile.

My god I didn't even consider that. So in other words in the middle of all this bullshit they intend to... deploy their guidance system. While flipping from 10,000g lateral to zero g vertical they're going to spool up, spin out and control some goddamn FLYWHEELS to keep the thing from turning into confetti.

That might just be worse than the fucking airlock, which I've been obsessing over since last night. 'cuz if their gadget is 20 feet long, and their airlock is 40 feet long just for safety's sake, and they're flinging the fucker out at Mach 6 (at sea level, rather than 1mbar), they have 30 microseconds to get the first door completely closed and the second door completely open. And they need to do it in a non-turbulent, symmetrical fashion.

Just for comparison's sake, I've been learning a thing or two about EtherCAT, which is a machine control protocol that lives on Ethernet but below the TCP/IP stack... because TCP/IP is too slow. It's kind of like token ring, in that the frame goes from the controller to all the devices on the ring and they add and subtract instruction sets from the frame to make things as quick as possible. EtherCAT is quick, and it's becoming the dominant protocol for machine control because it's also cheap. EtherCAT? gets done what it needs to get done in 500 microseconds.

Assume the door is the size of a pizza box. Assume it's double-sided. I need one to close (both sides) in 15 microseconds and then I need the other to open (both sides) in 15 microseconds. Let's call it two feet - the actual creature needs to be bigger, but I like the pizza box. I need that pizza box to close one foot in 1/15e-6 seconds - in other words, my door needs to slam shut at 67,000 feet per second.

Now - you may shudder to think that this is a lot of performance to expect from, say, an explosive sledge. You're going to have to generate a lot of gas in a lot of hurry in order to make a mechanical object go from zero to 47,000mph. Not to worry, though - not only is that only 0.0007C, or a minuscule fraction of the speed of light, it's also nearly double the burn rate of C4, a good 60% faster than HMX and 50% faster than what wikipedia tells me is the fastest explosive we've ever developed so no worries, friend, if we tried to motivate this fukka with explosives the burn front wouldn't even be halfway through the charge before it was too late so we'll have to think of something else.

Maybe Bezos is good for 10k g's? Let's try

As demonstrated, a capacitor BARELY BENDS OVER in 10,000gs, no worries. How much you think their reaction wheels weigh?