There is not one single thing a crew can do in the missile feed system. If something blows through, you certainly are not going to fix the problem with your own two hands in the next 20 minutes. One can barely change a single outlet in 10 minutes assuming you already have all of your tools at hand. Splintered tubes/rails, etc for moving 100 ton missiles? Dream on. Hours is reality. Days is more likely.
There is not one single thing a crew can do in an engineering room that a stand alone TTF+spare cannot do 10,000X better and WITHOUT errors. Heard of breakers? Heard of fuses? Heard of magnetic field fuses/breakers? Yea. That is why they exist without a human involved. DW's "cascading" power surge is complete BS. Can a short still be spectacular? Yes you betcha. Sparks flying everywhere when the bus bars melt when the breakers blow. Still is not going to blow up a fusion room.
Now you could argue that if a fusion plant was at max power setting, and its energy siphoned off to create electricity was instantly cut to zero and did not have a viable energy dump,
(There is not a single power system today that does not have a 100% capacity relief built in, why would that not be true in 2000 years?) Unless you wish to believe that all humans 2000 years from now are blatantly stupid of basic MURPHY and KISS principles could have interesting implications, but it certainly would not be called a cascading failure power surge! Voltage would spike, as the fusion plant pressure/temperature would climb as it is generating all that power without an outlet. Of course a very simple auto close multiple redundant circuit using halo effect due to overvoltage easily auto closes the Hydrogen feed line, or pressure, or temperature, or far more likely ALL OF THE ABOVE with multiple auto shut down on each sensor type! All of which would happen in a nanosecond to thousandths of a second. I do not know about you, but no human could even dream of stopping it in time that a multiple redundant auto system would not be faster, and far more reliable. No engineer would ever trust a human in the loop. No one trusts humans in the loop on our fission piles today, why the bloody hell would anyone trust a human in the loop for emergency shutdown in 2000 years? What you are going to have a human stand watch to physically turn a valve on the hydrogen feed line?
There is not one single thing a PDLC mount crew can do. Not a single thing. Even if the "communications" are cut. It is all computer or nothing. Incoming at 0.8c is just a tad faster than human reflexes...
There is not one single thing a graser mount crew can do when ships are separated by light seconds of distance. Now one could argue for a SINGLE human being able to figure out FfF and select a wedge shape for the computer to shoot at assuming communications are disrupted. Oh wait, that would require that both multiple hardline and wireless communications for target discrimination are offline. The human inquestion does not have a clue if that is a friendly ship or a foe they are firing at if they are not in communication with the bridge etc. This would require at least wireless coms to still work, in which case the mount itself still has its own coms as well. Yea, I don't buy that line either that a Graser mount would be useful with a human because by that time, the ship is a drifting hulk without a bridge, CIC, or a flag deck.