Science! (and what it can do for -you-)

[Daemonknight]

I would disagree that smaller always means harder in the computer world. If you are attempting to scale something down without changing functionality, I would agree, but this is something different. Going from DNI/VDNI systems to full automations is atleast doubling the required functions on a computer. Not to mention that you can't just stick a 'logical' system into a mech and make it work.

You've been in combat CS? I don't remember if you ever told me your MOS. For anyone that has; how many things are going on in your head once you hear that first bullet zip past? And I don't mean the fear, thats a given, I mean just things that actually pertain to your training.


A computer can be programmed to know those things, but can it be programmed to actually understand them? Does a computer know the different between 50% armor on it's center torso and 15% on it's right torso? It knows there is a difference, and that it needs to protect its right side. It would analyze it's own weapons, analyze those of it's enemy, and make a calculated decision to attempt to minimize it's own risk, and also maximize it's chances of taking it's target down. However, I garuntee that a human will come to that decision quicker, because so much of what the human mind processes can be 'tuned out' due to training. You can train a person to recognize things and react without actually 'thinking' about them. The term 'incoming' comes to mind. I have a friend who was a Marine for 15 years, and to this day, if someone yells 'incoming', he will immediatly drop to the floor without hesitation. He drops, covers his head with his arms, and burys his face in the ground, on pure instinct. A computer can't do that, it HAS to run the computations to come to a decision, by checking every possible factor. Granted it can check each individual datum faster than a human, it still has to go through the list.

And its an exponential problem: every second requires millions more calculations, as the situation in a combat zone is so fast-paced. There is no way to scale that type of system down to an implant. Its actually the inverse, because as a machine gets larger, it has more variables, which requires even more data checks, and logic paths. Making a Battle Armor suit an automation is likly about as difficult as a tank, or a battlemech. They all operate on essentially the same principles, although BA and Mechs have the whole gyro and jumping issues, those are minor compared to the tactical aspects.

However, what about an aerospace fighter? You have easily doubled if not more than doubled the ammount of processes that computer needs to run, simply by adding a third axis of movement. Then add in the dynamics of flight, and its likly another doubled process. Warships? Their sheer size, the number of systems required to allow them to function makes them the most difficult of all.


I would actually say that if you have EI or MD, it doesn't really provide any assistance to creating the types of systems I'm interested in. Likewise, the systems I'm discussing have little to do with EI or MD, because their purpose is to replace the human directing the action, not merely make him better.

[Cannonshop]

I would disagree that smaller always means harder in the computer world. If you are attempting to scale something down without changing functionality, I would agree, but this is something different. Going from DNI/VDNI systems to full automations is atleast doubling the required functions on a computer. Not to mention that you can't just stick a 'logical' system into a mech and make it work.

You've been in combat CS? I don't remember if you ever told me your MOS. For anyone that has; how many things are going on in your head once you hear that first bullet zip past? And I don't mean the fear, thats a given, I mean just things that actually pertain to your training.


A computer can be programmed to know those things, but can it be programmed to actually understand them? Does a computer know the different between 50% armor on it's center torso and 15% on it's right torso? It knows there is a difference, and that it needs to protect its right side. It would analyze it's own weapons, analyze those of it's enemy, and make a calculated decision to attempt to minimize it's own risk, and also maximize it's chances of taking it's target down. However, I garuntee that a human will come to that decision quicker, because so much of what the human mind processes can be 'tuned out' due to training. You can train a person to recognize things and react without actually 'thinking' about them. The term 'incoming' comes to mind. I have a friend who was a Marine for 15 years, and to this day, if someone yells 'incoming', he will immediatly drop to the floor without hesitation. He drops, covers his head with his arms, and burys his face in the ground, on pure instinct. A computer can't do that, it HAS to run the computations to come to a decision, by checking every possible factor. Granted it can check each individual datum faster than a human, it still has to go through the list.

And its an exponential problem: every second requires millions more calculations, as the situation in a combat zone is so fast-paced. There is no way to scale that type of system down to an implant. Its actually the inverse, because as a machine gets larger, it has more variables, which requires even more data checks, and logic paths. Making a Battle Armor suit an automation is likly about as difficult as a tank, or a battlemech. They all operate on essentially the same principles, although BA and Mechs have the whole gyro and jumping issues, those are minor compared to the tactical aspects.

However, what about an aerospace fighter? You have easily doubled if not more than doubled the ammount of processes that computer needs to run, simply by adding a third axis of movement. Then add in the dynamics of flight, and its likly another doubled process. Warships? Their sheer size, the number of systems required to allow them to function makes them the most difficult of all.


I would actually say that if you have EI or MD, it doesn't really provide any assistance to creating the types of systems I'm interested in. Likewise, the systems I'm discussing have little to do with EI or MD, because their purpose is to replace the human directing the action, not merely make him better.

Luckily for me, no combat time when I was in-at least I was spared that, thanks.  It's interesting that at the end of your comment, you essentially make the same point I was trying to make with my post: that "Cyborg-wear" isn't necessarily something that shares a tech-tree with AI or automation of weapons systems.

Though it might be interesting indeed to have a tech-tree that leads to AI augmentation of soldiers ala HALO or Red Vs. Blue.

[Cannonshop]

Tech Tree for Autonomous Munitions/robot soldiers:
Stage Zero: Hi-Scout Drone Carrier, C3(inner Sphere Only, including C3i Variants), TarComp (Clan OR Inner Sphere), Artemis IV, Guardian ECM (for the ECCM applications),Beagle Probe.
Stage Zero because these are current-service systems that can work as a basis for your development project.

Stage 1
Btt7i/augmented pilot training 'expert systems'-(TRO 2750, Page 60, "Swift" entry) they already exist in the game-universe (check TRO 2750) but would require dedicated effort to develop.  Game Mechanic: Reduces training time/resource cost to move units from R to V.  this is your 'stage one'.  Why? because you have to teach machines how to "Think like a fighter", and it also permits your programmers and dev. staff to view/review/examine real-time training and/or combat footage to help develop the needed reflexes for a future autonomous system.

Battle Armor C3: this one is found in (iirc) Strategic Ops, though it might be in TacOps (the damn book's laid out so funny I can't find much in either one...except by accident).  This teaches your officers how to "Think Cloud" with improved intercommunications between relatively small elements across a broader frontage.

Kanga Computer: the jump computer in the Kanga hovertank was complex, so complex nobody remembers how to build it-except your faction, who've doped out all the little microsensors needed to give a hovercraft a 'sense of balance' in free-fall/vertical envelopment operations. 

those're "Stage One" technologies-stuff the Star League had scads of at its height, open to all factions who have access to a good pre-war technical library or good reverse-engineering people.

Stage Twos:

Blackwasp "Brain"-the first units the SLDF automated were fighters-there's a reason for that-a fighter can 'coast' in space while making up its mind.  (Relatively common in the 28th century for guarding special installations, the Mk-19 Blackwasp was an automated, unmanned aerospace drone fighter with a very limited intellect-the same tech forms the in-canon basis of the "Caspar II" SDS systems used by WoB-possibly because the more advanced Caspar I's were too independent.  We have Canon game-rules for the II's, thanks to the Jihad sourcebooks, but even Herb Beas is afraid to tackle REAL Caspars as they were described in the Star League book back in the 1980s.)

Fredasa Automation (Clantech Only)-reduces the number of crew needed to run the ship-even in combat.  This is stage two because it fairly represents a bridge between "Expert systems" like the BTT7i and "SDS" quality automated ships-you still need a crew, but not as much.

The various "Caspar" systems detailed in the 3085 TRO would be Stage Twos-they don't work well without a human in the loop (-4 to initiative rolls)

Stage Three:

Caspar SDS/SYBIL Computer systems.  These are true Autonomous Warcraft, includes the Mk-39 Voidseeker and similar advanced systems.  These systems can operate on a tactical level with only an objective, and make limited value judgements (See fluff: Luxor class Warship-drew off Caspar and SDS units because they evaluated the class as the most threatening of Kerensky's navy.) 

Stage Four:
Beyond Caspars, true Human-level judgement and/or intellect.  If you're a Fanfic reader from the CBT boards, then names like "Sibyl Luddington", "Tabiranth", "Nike" and "Chobi" (or Man'yoshou) illustrate this level of development along this tech-tree.

You need to be at Stage Four to get things like autonomous Battlesuits that operate in "cloud" intelligence models with any reliability (esp. in an ECM rich environment).  These suits are likely to be Quads because Canon rules don't allow for autonomous bipedal robots at ANY level of development-including 'mechs, so it is logical that any extension beyond FanPro/Catalyst rules would follow similar trends (much as the development of the Inner Sphere version of the Clan Heavy Machine gun has the exact same defects as the Clan HMG-one can take some comfort in assuming, therefore, that some physical rules and engineering choices are necessity, rather than mere education.  Apparently it's not possible to make a Heavy machine gun that does Machine Gun damage over longer ranges as the trade up for the additional mass... Maybe in the BT universe, if you increase the proportion of propellant to projectile too far, you get a bomb, instead of a bullet...or something.)

For "Cloud Computing" AI "armies", the tech development would follow these steps, (based on canon sources and extrapolating beyond what Herb&Co. are comfortable dealing with...)
Stage Zero:
Hi-Scout Drone Carrier
C3i

Stage 1: Kanga 'brains/reflexes' computer, BA C3, BTT7i "Expert" pilot training aid.

Stage 2: Blackwasp equivalent 'brain' computer technology.

Stage 3: SIBYL computer (I forget the canon entry for this one-but it was the Star League's first true "AI").

Stage 4: Integration of what you've learned-the machine is probably a networked system-quad BA (unmanned) with reduced-size "Kanga" computers for their reflexes and balance, reduced-capacity Blackwasp equivalent brains, and BA scale C3, with a "master" computer proportionally installed in a vehicle (such as, say, a modified APC or urban combat track, or modded Hi-Scout carrier).  Each individual unit has about the brain-power of a Bumble-bee, but through the link, it increases in ability with numbers-a 'gestalt' intelligence, the more you have, the more it can do on its own-the likely 'cap' in this case, would be how many networks can work together-based on C3 tech, you can't interlink an entire battalion of piloted machines, the most you can manage is a single company-likely a SIBYL level computer can handle MORE, but perhaps not as well.  Optimum performance would be at Company level, with battalion level decisions coming through as simplified instruction sets ("Take that hill", "Retreat to Rally Point and re-group", "Exterminate all carbon based organisms in your sector", "Do Not engage vehicles using the following IFF codes" or "prepare L-shaped Ambush at site XYZ coordinates, set pre-programmed fires and suppress emissions until condition W occurs or condition X, Y, or Z invalidate this order.")


Another direction in this tech-tree goes toward human interaction (but not Cyborg humans...)

Stage zero: BAP, ECM, TC, C3

Stage One: BTT7i (Betty) Expert systems, BA C3.

Stage Two: Fredasa style automation of mundane functions (Logistics/Maintenance)

Stage Three: SYBIL style high-end computers in network.

Stage Four: You still have humans in the loop, on the ground, but they have AI assistance with additional sensors on vehicles, sort of a "Super C3" network where crews can put the tank/'mech on "autopilot" or BA troops can 'rest' while the suit keeps going.  On this tree, a Stage Four force can keep going under conditions where non-augmented forces can not, and enjoys a much more omniscient view of the battlefield with recon data updating displays based on multiple points of input-best uses would be in coordination with "Force Multipliers' such as artillery, air-strikes, or OrBom under high-tempo operations-basically a stage four force in this case would have better targeting, gunnery, initiative numbers, but would be horrendously expensive on a per-unit-cost basis-the trade off being that this higher level of integration improves the force's ability to inflict damage and out-manuever enemies in the field.  Game Mechanic would be paying for "Elite" forces right out of the factory.  The down-side being that you're paying for Elite forces, right out of the factory.  A single infantryman's load in this case would be equivalent in cost to a squad of Assault BA troopers, and if he's separated from the other arms, he's only got an infantryman's weapons on hand to defend himself-and he loses most of his bonuses if he lacks air-cover, or artillery support, or vehicle support, and if the force "Loses the Hub" then everyone is suddenly green, because their gear doesn't work, or "Regular" because it's only working across a company-scale range and the force multipliers don't work RIGHT anymore-but the RP cost for the force itself remains the same as if everything DID work like it was supposed to.

[Jeyar]

I just thought of something: how about a tech tree or 2 that the canon universe DIDN'T ever go down.

Now, one problem is that there is the risk of whatever it is being broken. But the bonus is that in theory those IN universe shouldn't know just what everything leads to before hand.

Like "information tech" may result in being able to use blackboxes as low level HPG at its endpoint, but we don't know that. We might be under the impression that it aids spies in figuring things out.

[Cannonshop]

I just thought of something: how about a tech tree or 2 that the canon universe DIDN'T ever go down.

Now, one problem is that there is the risk of whatever it is being broken. But the bonus is that in theory those IN universe shouldn't know just what everything leads to before hand.

Like "information tech" may result in being able to use blackboxes as low level HPG at its endpoint, but we don't know that. We might be under the impression that it aids spies in figuring things out.

That actually sounds to me like a neat idea-and some of hte more advanced models in canon actually DO work that way.

Here's an idea on an unrelated note...


Skyhooks/space elevators as a possible tech-tree (not considered in Canon).  this would allow "Portage" for ships that don't land (Like the Behemoth Dropship) to move high-bulk cargoes between planets.  If you can make shipping, say, GRAIN cheap enough, it can be a better revenue generator than shipping germanium ore.

Here's how I would envision it working in "Game Effects" rules:

Requirements: Tech roll/discovery success

Cost: 5RP per "tower", time to build each tower 4 turns (doesn't reflect so much the builiding time, as getting your supporting infra laid out-ships to use the towers, convincing people they're safe, staffing and training, etc.)

Benefit: Resource Multiplication.  Let's say we multiply the RP output by the number of such towers you've built in your nation...of course, there needs to be a 'cap' on this, but that can be worked down to a reasonable level-say, a maximum of x4 for ease of maintenance.  that means if you have four built, that gives you 1 RP (Link it to the output of the specific hexes! four hexes have towers, those four hexes each generate 1 RP instead of .25, the more you build, the more the multiplier is.)

Restrictions: there has to be a cap on the multipliers, so we cap it at a nice, round number...say, x10, past that, you're just building in redundancy.

This could, concievably, make smaller factions resource-competitive against larger factions-but at the cost of increased vulnerability of that resource base to attack-if you have, say, eight hexes with eight control worlds, and each one has a tower, sure, you're going to make sixty four RP (assuming no other infra is built)-but if the neighbours raid you on multiple worlds, you won't make that next turn even if you beat them back (provided the dice gods don't like you that turn!)