B1 Droid AI and Starfighters vs. Capital ships

[Seydlitz_k]

I'm generally a lurker around these parts, but reading the posts in the Clone Wars series threads and watching the series itself, I've been tossing a couple of theories around in my head that I would like to present.

The first regards the B1 battle droids and their antics and stupid personalities. While reading the ideas of why they are suddenly so silly (Poor maintenance, not enough memory wipes, etc) there was Coyote's post about how it might have been partially intentional as well, in a bid to create veterans with a sort of 'sixth sense'. This immediately reminded me of a friend of mine who was 'evolving' an AI for his final year project in uni.

Basically, he created a simple game, and made a bunch of AI 'entities' play it against each other. Each entity would play a couple of games against the other entities. After a certain number of games, he would pick out the AI entities that won the most, and create a new generation of AI entities based on the winners of the previous round of games. He did this for a couple of generations, until the project was due, and by that point you could clearly see the AI had evolved. It wasn't perfect, but from the completely random AI of the first generation, the newest generation of AI was actually using simple strategies in their games. The AI had evolved through a sort of natural selection. Most of you have probably heard about AI development experiments like these, as a lot of people have been doing similar things these days.

What if this was partially the intention of the CIS? While originally, the B1 droids could be maintained and only needed to run basic security and tactical programs because they just needed to provide protection for freighters or dignitaries, this limited programming would not have been sufficient to adapt to the various scenarios of a galactic war. Obviously, with quadrillions of droids being pumped out, it would be impossibly expensive to maintain them all as well as write a program or an AI capable of adapting to every scenario. Instead, they left their droids on their own, in an attempt to evolve a capable AI. The droids which survived the encounters would be monitored and have their memories copied an then printed onto the new production round of droids. So, given enough generations the inept droids would be weeded out, in favour of the ones like the B1 in the subway system who knew not to fire on a Jedi when his lightsabre was out.

Naturally, the major issue would be that most droids would never see front line combat or be destroyed (Especially those on capital ships, who can escape destruction due to poor combat AI if their ship retreats from a fight it's losing), so the majority of droids in service would stay stupid, especially if it is too costly to retrofit them with the AI from the more intelligent droids. This would explain why the droids on Malevolence and Providence are so stupid.

It would also explain why droids would get promoted. Those who show more initiative and tactical ability would be marked out, especially as candidates for having their memory copied onto newer models (Or even, droid commandos?).

Perhaps this is also why Asaj Ventress and General Grievous have such inept droids, as they are such "good" commanders, and suffer less casualties. They wouldn't get the newer, more intelligent droids due to the logistical nightmare of re-equipping whole fleets, or because the intelligent droids would be sent to the more dangerous theaters with the highest casualty rates in an effort to balance things out with better tactical AI (Or even to speed up the process of AI natural selection).

In summary, if this was tl;dnr: The CIS let their droids develop personalities partially to try and create a more sophisticated AI through natural selection (As well as because quadrillions of memory wipes and maintenance would be much to expensive). These smarter survivor droids would have their AI copied on to the newer model droids so that each new droid produced would be smarter than the previous ones. The main CIS players (Grievous, Ventress) have such stupid droids, because there droids suffer less casualties due to better leadership (I'm assuming their tactical and strategic abilities are some of the best in the CIS, otherwise they probably wouldn't be such important characters) and so aren't replaced as often. That, or the newer more intelligent droids are sent to theaters where the CIS is losing ground to try and turn the tide, and perhaps speed up the AI development.

I'm putting the starfighter vs. capital ship part in a seperate post, to improve readability.

[Bounty]

It's an intriguing idea - the only problem I see with it is that the SW galaxy already had thousands of years of AI technology behind it, so it seems wasteful to "evolve" an entirely new AI program. Then again, I don't know how widespread the use of droid soldiers was before the Trade Federation's antics, so while high-quality AI exists there may not have been any readily available for large-scale armies.

[Vehrec]

The major problem with 'natural selection' AIs is they tend to become sub-optimal due to 'local peaks' of the fitness landscape. Imagine that you can only step up and never down-well if we put you at the bottom of mount Everest and blindfold you, then let you walk to the top, odds are you'll wind up stuck on an ice block, unable to get down and unable to reach the actual peak. Needless to say, this is also a problem for biological systems, and you can only get away from such micro-peaks by jumping across large areas of the landscape or by changing the landscape itself.

[Seydlitz_k]

Starfighter vs. Capital ships

Watching the 2nd episode in the Malevolence story arc, if we assume that missiles are less powerful than lasers, than their is a problem with what the Y-wings acheive.

I don't know if this has ever come up in the EU (I'm guessing it's probably been considered around these parts?), but I'm pretty sure so far it's always been assumed that Star Wars missiles are just warheads with rockets attached. Considering what our current day missiles and our anti-tank rounds can do, might it not be fair to guess that Star Wars missiles are more advanced?

From the game X-Wing Alliance we know there is such a thing as an Ion Pulse warhead, a missile which with one shot disables a fighter.

Why is it not possible that standard anti-capital ship missiles have similar technologies? Why can they not be similar to modern day HEAT rounds, and operate in various stages?

For example, the missiles the Y-Wing's launched could by a 3 stage warhead. The first stage is some sort of Ion pulse, which temporarily weakens or disables local shields. Once this is done, a third stage that works on similar principles as a HEAT round could create a hole in the hull armour which would focus the blast from the explosives carried by the missiles into the interior of the capital ship, causing significant local interior damage.

Of course a single missile would not be enough to significantly weaken the shields of a capital ship, which is why to attack the bridge of the Malevolence Anakin needed the whole squadron. If each Y-wing launched a pair of missiles simultaneously, that would mean 20+ missiles unloading ion charges onto the shields at the same time, and even if 19 of those warheads get stopped by the shields (Which would place extra load on them), chances are one would likely get through and cause significant interior damage.

Perhaps the ion cannon charge up sucked enough of the power from the local shields enough so that a smaller amount of missiles could overload the shields and penetrate the hull to cause enough internal damage to overload the weapon system.

This would also go to explaining why warheads are actually useful at all, and also why starfighters are still quite popular. A dedicated missile attack by enough starfighters could cause local damage to a capital ship, and while the starfighters alone wouldn't be able to take it out it would greatly help any friendly capital ships.

[VT-16]

I always thought the missiles and torpedoes were much stronger than regular shots, which is why they kept them in reserve in ANH.

[Seydlitz_k]

The major problem with 'natural selection' AIs is they tend to become sub-optimal due to 'local peaks' of the fitness landscape. Imagine that you can only step up and never down-well if we put you at the bottom of mount Everest and blindfold you, then let you walk to the top, odds are you'll wind up stuck on an ice block, unable to get down and unable to reach the actual peak. Needless to say, this is also a problem for biological systems, and you can only get away from such micro-peaks by jumping across large areas of the landscape or by changing the landscape itself.

Could that not be solved if droid formations are periodically rotated through different theaters of war? Although that might not be very cost effective, or very practical.

It's an intriguing idea - the only problem I see with it is that the SW galaxy already had thousands of years of AI technology behind it, so it seems wasteful to "evolve" an entirely new AI program. Then again, I don't know how widespread the use of droid soldiers was before the Trade Federation's antics, so while high-quality AI exists there may not have been any readily available for large-scale armies.

Yeah, it would defintley be wasteful to evolve a new AI program if a better one is readily available, but maybe it's a lot cheaper evolving your own on the field as well. You don't need to hire any new programmers, or buy and modify existing software. The B1's probably weren't designed to fight in a galactic war, and retrofitting them might be too much of a nuisance.

Perhaps a more advanced battlefield AI already exists, but the CIS need a license for each droid? Intergalactic DRM issues!?

[Master_Baerne]

Starfighter vs. Capital ships

<snip>

I'm pretty sure this was brought up the New Essential Guide to Weapons and Technology, in the concussion missile section. They were apparently fired in pairs, staggered by a fraction of a second, in order to achieve exactly this result: A momentary shield breach through which the second missile would fly.

[Samuel]

If missles are so good, why don't capital ships use them exclusively?

As for the droids, a better system would just get a system so that they all vary, let them loose and copy the programming of the surviving droids onto the next batch.

Of course, your system seems like the one they are doing- the advanced battle droids can be fooled by peek a boo

[TK-984]

^ I would hazard a guess that Cap ship heavy turbo lasers are better than missiles, but ultimately if you want to effect precision damage to localized areas, you want to go in with fighters equipped with missiles.

[evillejedi]

I still like the idea that the B-1 idiocy is trauma from being exposed to Grevious

Warheads could be effected less by the deflection action of the 'deflector' shields than turbolasers and might impart more kinetic energy in an attempt to overload the local shielding equipment ability to absorb shockwaves in the shield. At long range the warheads could be targeted and destroyed (some of the capital scale missiles are larger than a star fighter by most indications so they would be able to be tracked and intercepted with some regularity and would take up significant space in warships) so their utility seems to be best when used against large targets that can't effectively destroy them(orbital and surface), for ultra long range stand off capability against moving targets (the missiles would get to a very high fraction of light speed very quick and would be able to track unlike TLs) and for get-the-fuck-away short range salvos. I've personally always wanted to see an SSD open up with its launchers on some poor ship that gets too close. (or just a visual of the torpedo sphere vs SSD battle over ?corellia?)

[Master_Baerne]

Missiles don't destroy the entire ship. The function of turbolasers is to pump as much energy into a target as possible, while missiles seem to be a bit more...fine-tuned.

[Seydlitz_k]

Starfighter vs. Capital ships

<snip>

I'm pretty sure this was brought up the New Essential Guide to Weapons and Technology, in the concussion missile section. They were apparently fired in pairs, staggered by a fraction of a second, in order to achieve exactly this result: A momentary shield breach through which the second missile would fly.

Ah yeah! I thought I heard that somewhere, wasn't sure about it though.

But why do they have to be fired in pairs? That seems a bit wasteful if you consider most Star Wars Fighters can only carry around 6 missiles. Shouldn't each missile be capable of doing something similar by itself? With the Star Wars technology level, it shouldn't be hard to do.

Checking out wookieepedia it seems most missiles have some sort of armour-penetrating ability, but no way to overcome shields. I highly doubt the missiles guidance system or propellant would take up so much space that the warheads themselves could not be extremely specialized.

evillejedi, wouldn't particle shields be designed to stop such physical, high KE attacks? I'm assuming most capital ships would be equipped with them to stop such eventualities. You are right though, most combat warships in the SW galaxy probably have extremely advanced countermeasure systems which would make missile attacks from normal ranges fairly useless. That could also explain why fighters raids against capital ships using missiles are so effective; The ship probably just dosn't have enough time to react with countermeasures. Missiles might never be used against other fighters for the same reasons, that SW countermeasures are just too good. (By the way! If you are the EvilleJedi of Warlords fame, I love your mod man! Great work. I still always cackle with glee when I see the SSD open up with missiles and turbolasers! )

I'm also guessing that starships are constructed so that if any missiles do reach and penetrate the hull, the resulting blast would be mostly contained by the rooms/cells of the area around it. So Anakin's Y-wing squadron missile attack in Destroy the Malevolence! would have devastated the bridge room, but not have done much to anything else in that area.

[Coalition]

Could that not be solved if droid formations are periodically rotated through different theaters of war? Although that might not be very cost effective, or very practical.

It could also be that if you rotate them, you might also have to refit the droids for the new envronment. I.e. transferring a droid from a forest region to an arctic region would require different materials, lubricants, etc. You'd have to rebuild it. Of course, if you just swap around the program, that gets past that problem.

Perhaps if you leave them in a forested region for a long time, they could become better jungle fighters, rather than the generic trooper.

It would be interesting if the ships are also local nodes for improving local droids. Maybe every day the ship polls the planetary droids, and the survivors have their programming adjusted. When it receives a new batch of droids (or updates from a higher node) it distributes the upgrades to its troops, and monitors their effectiveness.

[Darth Ruinus]

[That could also explain why fighters raids against capital ships using missiles are so effective;

Wait hold on, I haven't been paying attention to Star Wars in a while (been busy with my upcoming midterm) but I thought it was that fighter attacks weren't effective against capital ships unless the fighters were being backed by a capital ship (or the capital ship being assisted by fighters, whichever way you want to look at it)

EDIT: Grammar

[lord Martiya]

As far I know, the fighters are an unstoppable force of destruction only in the flying simulator games and in the Rogue Squadron novels. And I remember that in the Shadow of the Empire novel the Rogue Squadron plus Luke considered a lucky shot his only possibility to damage (not destroy, damage) a lone ISD.

[Samuel]

If fighters are able to do precision strikes to damage capital ships (which would make a bit of sense), they probably can only hit and damage sensor systems or a weapons turret. They rest of a capitol ship is probably hardened. Except, of course, one of the auxiliary bridges.

I'm not sure having a central droid control and programming center is something the CIS would do- they learnt the last time it is a bad idea.

[Swindle1984]

I always thought the missiles and torpedoes were much stronger than regular shots, which is why they kept them in reserve in ANH.

That and the blaster cannons on the Rebel fighters fire in a straight Line-Of-Sight fashion, whereas the proton torpedoes were capable of executing the 90 degree turn and then following the exhaust port to the Death Star's main reactor. Which is what they needed.

[Seydlitz_k]

[That could also explain why fighters raids against capital ships using missiles are so effective;

Wait hold on, I haven't been paying attention to Star Wars in a while (been busy with my upcoming midterm) but I thought it was that fighter attacks weren't effective against capital ships unless the fighters were being backed by a capital ship (or the capital ship being assisted by fighters, whichever way you want to look at it)

EDIT: Grammar

Well, I don't mean effective in destroying the ship. They should most likely be able to damage or destroy components of it, like sensor suites, turrets, or small areas of the ship near the surface (ie. unprotected bridges.), as Samuel mentioned.

I'm not saying that the depictions of fighters in the X-Wing games or in the Rogue Squadron novels is correct, but that the heavy reliance on fighters in the movie and in the CW cartoon means that they are probably quite useful, and that in groups they can be a severe nuisance to something like a Star Destroyer. By closing to point blank, they can fire missiles at targets that might otherwise be protected from longer range fire by countermeasures or shields (Assuming missiles have some sort of local shield 'piercing' capabilities).

I always thought torpedoes were used in ANH as a last resort ("the exhaust port is ray-shielded, so you will have to use your torpedoes"). Although unless the shaft is perfectly straight, lasers would probably be pretty useless.

[Samuel]

My theory is that due to the high proliferation of personal craft they simply reconverted alot of the production into fighters. The need for fighters is due to the fact that people would use their own for raids and other mischief and you'd need a cheap counter.

After all, in the movies, the rebels are fighter heavy probably due to the fact they can't go around making cap ships as easily. The Empire doesn't use so many fighters considering- they have 144 fightercraft in a SD.

As for the prequels... they are ramping up to wartime production with peacetime fleets used for policing. They probably went with what they know.

As for bombers, they probably exist to take a ship apart once the shields are down. That way the caps can concentrate on other targets.

[Lord Relvenous]

IIRC, the raids in the Rogue Squadron Bacta War were only successful against capital ships when the figthers has some back-up, expected or not. It's said righ tin the book that the Star Destroyer captains are smart enough to just roll their ships, presenting new shields to the fighters, and exhausting their supplies. So while the fighters are useful because they can drain shields with proton torpedoes, they can't do it alone, even in Micheal Stackpol's books, where the Starfighter wank is strong.

Also, as to why cap ships don't exclusively use concussion missiles and torpedoes: Don't Victory-class Destroyers have something along the lines of 80 missile tubes? That seems pretty extensive to me. Also, the price of torpedoes and missiles seems to be very high. They are high cost muninitions, while turbolasers are one time cost weapons. When implemented on a scale as large as cap ship weapons, this is probably a major consideration.

[Lord Revan]

it also should be noted that SD are capable of shooting down missiles with alot better chance of succes then modern ships, so that might account for the rarety in which we see missile weapons used by cap ships (at least partly) as you can shoot down a turbolaser bolt.

[Samuel]

Turbo lasers are NOT one cost weapons. The power they use is generated from hypermatter. Given the sheer unobtanium of that substance, it means they cost alot to fire.

[Bilbo]

I always thought torpedoes were used in ANH as a last resort ("the exhaust port is ray-shielded, so you will have to use your torpedoes"). Although unless the shaft is perfectly straight, lasers would probably be pretty useless.

I thought the implication was that the entire length of the port was so well shielded (you dont want your exhaust port's tube leaking radiation into your Death Star) that any shot that made it into the port would go straight down into the reactor like pouring water through a bendy straw.

[Solauren]

It's been stated in source material that warheads are generally much, much more expensive then Turbolasers both to obtain and maintain, and that's why they are used on starfighters. Starfighters are a better delivery system, and are capable of much more precise attacks then a fired warhead.

[Darth Ruinus]

Turbo lasers are NOT one cost weapons. The power they use is generated from hypermatter. Given the sheer unobtanium of that substance, it means they cost alot to fire.

What? Aren't hypermatter reactors used in almost everything, from warships to starfighters, so it can't be expensive to use or find. Its unobtanium to us, but not to them.