cthia wrote:See above. That might be true if someone is there to oversee the process and budget power usage and tell the dumb capacitors not to attempt to maintain the last requested load. The capacitors will discharge very quickly if they are sending huge amounts of power thru the grids as last requested, which is wasted in dissipated heat and fried circuits.
ThinksMarkedly wrote:Why wouldn't the automation be able to adjust the power balance? Reduce the reaction rate to match the power consumption.
It
can adjust the power balance, but it does it in a generic fashion that will almost always be counterproductive to specific needs even in time of peace. The problem is not always the reaction rate. And the problem isn't always the power consumption. The problem is that, mostly, the power consumption in battle is the same (or increases), but the power itself needs to be
rerouted around damaged access points and junctions. The power distribution needs to be nuanced, finessed. It is a finely tuned orchestra.
It is much like water distribution to a city. It will never be designed for total computer control. The demand for water pressure (vs power in this scenario) may outweigh the need to shut down pumping stations (reactors) to decrease water pressure because of busted 16 inch mains that is shooting water up into the middle of a street. You don't want your software making that call at that point in time because firefighters just might happen to be battling a five alarm blaze in that same section of the city. So, the decision becomes, shut down the necessary pumping stations to reduce water pressure to prevent the geyser and loss of water in the middle of Fifth Street, or temporarily put up with the geyser and loss of water in favor of water pressure
until the raging battle against fire (or firepower in our case) is complete. The point is that an AI is not a substitute for a lifeform in decisions that are a matter of life or death. So you never design them to be such because you intend to always have warm bodies on tap (pardon the pun.)
ThinksMarkedly wrote:You don't need humans for that. In fact, you don't want humans to do that, the automation can react much more quickly to avoid overloads and "underloads."
The quickness of the decision may bite you in the ass. Even during black&white moments when the reactor needs ejecting.
So, uh uh, the reverse is true.
You don't want the software making that call in matters of life or death! In a raging battle with firepower in the middle of a space opera is life or death! Humans need to make the
big calls, and you can bet that is how it will be designed onboard ships. Software simply cannot decide that certain "risks" are more acceptable to the alternative. Humans may decide that in any particular scenario the ship is less important than the battle, that the crew is less important than the honor of the Queen, and that safety is the last thing to be concerned with. Death before dishonor! You can not program those decision banks into molycircs, and you'd be a fool to try. And, since compensator failure is "supposedly" a rare thing anyway. Why would you even try? But the true question becomes, since you need people aboard ships, why not use them and design with their decision making abilities and needs in mind?
You can only design black or white decision making capabilities into a reactor used aboard a warship, minus the gray areas that need to be decided by gray matter.
ThinksMarkedly wrote:Similarly, if reactors scram, the automation should start shutting down secondary systems to conserve resources.
Negative good buddy! That is another black or white decision that is really gray. You don't want the reactor to start shutting down any systems in the middle of a battle. Even secondary systems that may be in use by the crew who are trying to get the mains back on line to fire grasers!
The reactor will be programmed for
black & white operation only! These are warships, not homes worried about maintaining power to the HD TV in the middle of your favorite show. In fact, even in black and white situations that the programming can handle, it still defaults to a human. Instead of ejecting the core immediately, the computer will be programmed to give a countdown. Recall when Santos?, the more experienced gray matter aboard
Fearless, had to make the decision to bypass certain stages and eject the core.
ThinksMarkedly wrote:The human is not going to control each individual system any way, they're going to usually select from a very short menu of options, based on the operational needs: is the ship trying to flee or is it trying to fight back? The human is there to override the default choices, when the captain calls engineering with "Scotty, more power to the engines!" Or to make unsafe choices, like venting plasma because the heat exchangers can't cope with the load.
All true when the ship is at general quarters. The equation changes at the sound of the klaxon signaling red alert when every warm body hustles to his
duty station ready to become a well-oiled cog in the bigger decision making process, like venting plasma. Or choosing to NOT vent plasma because for whatever reason it is not recommended by the current situation.
ThinksMarkedly wrote:So, no, I don't see why a ship would tear itself apart if there are no humans aboard. It's perfectly able to select the course of action to preserve itself, provided that's what it is programmed to and provided that it's not damaged beyond repair anyway.
I'm certain the system can handle canned responses to canned scenarios, but you can't design for situations which are not canned, and since you rely on gray matter to make the decisions, you design for gray matter to make the decisions. If there is no gray matter around, the point is moot.
But!
Even the canned-response-mode has to be selected. Compensator failures are instantaneous, prohibiting prior selection.
The reactor's job is not to ponder why. The reactor's job is but to do,
or die.