Can a Commander build itself?

Molecular manufacturing systems can move atoms into good positions, but they can't change the atom. The key difference is that there's a lot of nuclear material out there, but it is not the element or isotope that you want. Refinement means changing the atom, which is done with heavy machinery, not chemistry, and especially not with nano assembly.

Perhaps. But then you have a hundred tanks, while the guy making "perfect" machines has two or three. Quantity has a quality all its own.

The idea of "ultimate" or high quality tech is a definite possibility in the universe. The key factor is hitting that sweet spot on the "quality/quantity" curve. The nanite hits that sweet spot. It has everything needed to build a unit to the correct precision, and can then be used as part of the unit itself. This makes assembly simple, fast, and suited to anything with a lathe.

Higher quality units, while ideal, may very well sacrifice too much time or effort to be worthwhile. Atomic perfection is obviously going to be the slowest process of all. These factors would make a unit non competitive, or at least better suited to defensive efforts where time is the advantage.

For a Commander none of these factors are an issue, as their design limitations are size and mass. A bulk fab process would not be suited to creating the highest quality unit in the smallest/lightest package possible.

~~~~~~~~~~~~~~~~~~~

There is certainly a difference between making a unit or a structure. Units are more involved machines that demand many subsystems and complex instructions from the builder, slowing down the process. Structures tend to be bulk raw materials, big dumb machines where the only difficulty is getting materials on site. These differences can be best represented with "build power" and "Energy". More complex designs demand more build power, and more difficult features demand energy. Most of the games had this kind of system, creating general trends between units and structures. The fine details between energy/mass/build power didn't matter too much, and could certainly be simplified here. Zero-K removed the system entirely, as everything was built off of one number.

A perfected form of a unit wouldn't really cost a different amount of raw material, but there could be a vast increase in energy and difficulty required. In the end, all that you get is something akin to a "veterancy" system. Extra effort can be used to make the highest quality version of a unit possible, creating an "elite" caste of units that save money in the long term but are very difficult to build. This is a nifty way to create extra unit roles, but would be too cumbersome to apply to every single unit.

Yes, yes, my mistake. But really, point is that 10 nm already here. 22 nm for CPUs are only intermediate step onto the hard limit.

Well, if you say so. It's just matter of opinion. For me, placing resist and ions with atom precision is similar to atom-by-atom deposition (yes, I understand differences into process, but see below). You are just not making it by, naturally, taking an atom and placing it into correct place. That's "straightforward" way. But humanity is not known for going with straightforward ways of incredible complexity. Humanity is known for smart workarounds and cheating the nature so it makes all hard work itself. Nuclear centrifuges and photolithography is, imo, perfect example of that.

It's actually hard to make anything under constant pressure, demands, bombing threads and murders of your scientists. But yes, of course it's not trivial, but it's still easiest part of the process. Dirty-bomb could be done with nuclear waste from nuclear plants (see Chernobyl), no need for weapon-grade uranium. Basic "low-yield/high-fallout" nuclear explosion still require a good precision of explosions to hit two uranium rocks together. Or there will be no explosion at all. And that's actually part where molecular manufacturing good at. Plutonium bomb is top of modern nuclear tech, no surprise it's damn hard. Not sure about future though.

My mistake. By "heavy elements" I mentioned "transuranium elements", of course. They all are radioactive. But yes, reading the papers I may say that it's even more uncertainty about "stability". I always thought that "instability" is kind of quark instability, i.e. atom may exist only within very specific circumstances. Single uranium atom without any massive energy income (or neutron stream) is "stable". But you seem to have greater expertise in that field, so I'll agree with your point.

Well, of course they have disadvantages as any block-assembly tech. But @bobucles already answered that and I agree with him - if product is not perfect, but good, yet made 100 times faster and 1000 times cheaper - who cares?

Anyway, I really doubt that precise molecular composing (going back to atom-by-atom argue) is really viable for in-field production - even cheating processes require "clean room" which is hardly obtainable under constant fire.

"Growing" tech (as it's done by Aeon in SupCom) is actually more realistic approach, IMO - not much more precision than nanites assembly, but greater variety of possible resources use (as you actually streaming resources from within) and less overhead for nanites production.

Of course, but providing a solid explanation is also the vital thing, imo. You just need to first make a gameplay decision and then think of it's explanation.

Hey, aren't you supposed to oppose efficiency upgrades?

Why? It hasn't been critical in the vast majority of all games since ever...

Things need to be at least plausible for aloot of people to be able to immerse into a game. So some explanations (Even if they are bull if you take a closer look) are often nessesary.

You personaly might be one of those that dosent care, but that dosent meant there arent others that do and the game isent made for you personaly. If you take a look at other science fiction game i belive you would be suprised at the length some devs go to make the game more realistic (do note that "more realistic" is not in any way equal to "completely realistic" xD).

Like I mentioned, doing this for every unit is just dumb. Putting a star next to a unit's name is kind of silly as it bloats the effective unit count without giving new units. Rather, it offers another niche of units for the game; efficient on mass, expensive everywhere else. Such a thing might look like:

Code:

Same mass cost
20% more awesome
100% more energy
100% more build time

It certainly looks like a terrible choice, but maybe it's not. Players who can make the most of their mass can find themselves at a considerable advantage. Spearheading an invasion with higher quality units means easier transit, and they'll have more success breaking defenses or reaching objectives. They're also more juicy targets to destroy, as they are obviously more difficult to replace.

More importantly, it expands the options for making very similar units stand out from each other. A tank and gatling bot might do similar things, but one is priced for easy use while the other is the "elite" variant. A wind generator might be the "elite" variant of a solar panel. A pop up cannon might be the "elite" version of a typical cannon. Fighter => stealth fighter, Tank => Hover tank, the list is practically endless. It's easy to visually determine where the elite units are, and the strategic icons have an easy time marking these units (gold star it, man). It's just an option, anyway.

Aaaand... You've just invented SupCom teching. Congratulations. Only difference is that SupCom teching is real and badly balanced and your teching is imaginary and therefore perfectly balanced.

Since Evolution throw errors and survival of the fittest is not possible why not have evolution throw experimental units. anything made is perfect from a design point of view but untested, adding new abilities even just for a short term like agility(bot gets extra batteries to give the engines extra power for a short time before it is depleted and need recharging (this also adds extra weight to the bot and makes it slower in normal mode)).

They are having an argument (very vaguely) about using nano-lathes to build nuclear powered machines and comparing it to a real world process for building microchips. Personally I prefer the approach used in Total Annihilation, Supreme Commander, Supreme Commander: Forged Alliance and Supreme Commander 2; they build stuff with nano-lathes. No further explanation was required or provided.

I see, maybe i should have read a bit more of the discussion instead of jumping on your one post out of its context. Sorry.

Not to worry, the posts are pretty brain-numbing

Procedural generation.

Where's my Nobel Prize? (Of course, I assume that the Commander itself is made in such a way that you could procedurally generate it efficiently)

Assuming these are extremely high tech future robots, it probably wouldn't be so extreme just to recalculate the entire robotic "thought" processes that created the commander in the first place as a form of procedural generation.

If that's not believable, then what about simple lossless data compression? Modern voxel engines have managed to store voxels at less than one bit each - compressing the location of the voxel in 3 dimensions, the colour, and so on, at LESS than a simple 1 or 0.

If that's possible with current tech, I think it's not so far off to say futuristic robots that are at the literal limit of tech could pull a atom by atom recreation of themselves.

Spherical commander in vacuum.

Could you provide a link for that one?

Eh. The amount of data you may store into small atom grid is far greater than amount of data required to store position and composition of said atom grid. It's combinatorics.

Wow. That's so wrong it hurts. Supcom teching is "more expensive, bigger, stronger". It does not trade difficulty for resource efficiency, at least not on any explicit level.

The idea here is that two similar units cost the same money, except one is much more difficult to build for some bonus. This is done with vast increases in build time and energy cost, where the more common units are much less difficult to produce than the elite variants. It's just a different type of balancing point for units, compared to multiplying everything by 5 for a superior unit.

Spend a little effort before posting, okay? You might sound cooler in the long term.

Already got ya covered:

The thing about procedural generation is that it only provides the method, not the answer. Working up a full schematic can range from easy(armor, shape), to moderately difficult (circuits, nanites, elite tech), to insanely hard (removing every single flaw in the comm and all its units; refining a perfect no-waste package; thrive in every single simulated environment and beat every possible combat scenario). Super robots may be able to solve everything up to "moderately difficult" with no hitch. Beyond that, the simplest answer is to just clone yourself.

Going for the super insane challenge is already covered in the first post. Getting everything procedurally refined and perfected is a job for a super computer(in an age of supercomputers, no less), not a Commander. The Comm can certainly know how to build that kind of computer, which no doubt is a very expensive and perhaps even a campaign-defining device. But why try to reinvent the von neuman wheel, when the answer already exists in physical form?

Okay, this is fairly elementary, but apparently it needs saying:
It is impossible to store the information contained on a storage device, without actually storing the information.

These death machines already use the best data storage possible. How could they store the schematics of that device, information and all, in some way that's besterer?

You actually can't differentiate idea from it's implementation.

Build time and energy is "money" too. Slight increase of mass within glowing economy is "same money" as well, so it's "more expensive, bigger, better" in macro scale. With imaginably better balance.

Go for it, clone yourself

Your brain (2-3 kg of biomass) "schematics" are stored into DNA, 3Tb HDD schematics could be stored on floppy disk. Quantum computer is capable of operating Petabytes of data, but it's schematics are stored onto some ordinary HDD disk or flash stick. So, what's the deal?

Crap, even infinite amount of data (pi constant, Fibonacci sequence, whatever) could be stored using finite memory (generation algorithm)

You argument stands true only if there is no quantum memory (which is not true even in modern times) and you're trying to store exact atom-by-atom image of memory whole device without any compression techniques or any common sense.

If you do use common sense, than you'll describe only one memory block that allows you to store 1 bit and algorithm for placing this blocks. Or you may just compress your data.

You are insisting that ACU is Perfect Tech (tm), they why it should not use perfect (optimal) algorithms as well?

I'm sorry, but after several pages of endless waffle, your post made me laugh. :lol: Procedural generation, voxel compression heh.

On a more serious note, and keeping away from the stigma of reality, a commander possesses the knowledge and capability to build facilities capable of (eventually) constructing units that in at least some ways (usually in firepower for example) are superior to the commander. As these tools are very versatile it is not too unreasonable to think that they can be upgraded to the point where they are capable of creating a commander.

The real question is, is a commander capable of improving upon itself? (Making up lore here) When taking into account that technology has been refined for thousands/millions (who's counting?) of years to the point where all armies use the same unit designs and yet there are still variations in commanders eg Alpha/Theta/Progenitor it is reasonable to assume that Commanders can indeed improve upon themselves.

It's hardware vs. software. A human brain takes over 20 years to program. A 3Tb HDD starts off empty. Quantum computers perform very specific tasks. In every single situation, the software is stored as a real world change in chemistry or physical structure. Without it, the information would not exist.

A procedural algorithm is NOT the information, no matter how much magic you think there is. Rather, it details HOW the information may be discovered. As already mentioned the process can be easy or hard. But in all cases if the information for a process is lost or destroyed(as often happens in battle), then the algorithm has to be run again. For structural repair this is easy, as any physical shape is easy for killbots. For advanced circuitry and tech, it is tricky to figure out where the yellow wire is supposed to go. For the programming and experimental tech(which is itself an atomic layout), anything less than complete error recovery has to be rediscovered anew. A blind "just do what you did before" would not work because it is not possible to store more information than what a unit contains.

Basically, your argument is centered around the idea of self repair. Of course a Comm can fix itself. It might even be able to reprogram itself, in some capacity. But a Comm can not create something from nothing. If an arm is blown off, the replacement will never be the same as the original. If a core CPU is shorted out, exabytes of data could be corrupted. These problems, while disastrous, are not within a Commander's purview to rediscover as new originals. It is at war, and it makes do with what it has, no matter if the fix is inferior to what it had before.

It sounds a bit like Supcom2 research, and in a way it is. But instead of building things up, massive mainframes serve to redesign and replace things that were torn down. As cool as it may be to suffer critical malfunctions, it really wouldn't work too well for this kind of game. After all, the typical critical malfunction is to go nuclear.

When it comes to making a new Commander, it is not possible to give more than what you have. A perfect clone accounts for everything that a Comm has.

I don't actually understand what you are talking about... Yeap, to it required 20 years to grow human brain. Ok, so why not do the same with ACU? Quite an explanation why it is hard task. 3Tb HDD starts off empty, yeap (not precisely - it starts with garbage, but anyway) - but you may fill it with information of this HDD schematics and all production process in every detail. And you will have a lot of space left to fill it with computer games too =). So it's perfectly good idea to store schematics of some device on same device, instead of reverse-engineering it each time you need to produce new one. And it's even better idea, actually, if checksums, replication techniques, redundant information and other protective stuff taken into consideration.

WUT? Man, you just turned all known computer science upside-down. Every program has algorithm, every algorithm could be programmed (it's not proven, but it's widely-accepted assumption). And program is just a bytesting - a sequence of numbers. A one huge number. Algorithm is a number (there is even special term - "algorithm number" and a lot of things based on that, like "extra algorithm" or "universal algorithm"). And number is a data/information.

So you may describe pi with finite number. It's not pi itself, but you may rebuild it up to any required precision using this number. And note, please, that you can't just "store pi" completely - it's infinite. You may even build your pi-dependent algorithms, so they won't use some pre-calculated "pi constant", but will operate on pi-generator itself.

It's called "finite representation". Computers could work only with things that have "finite representation" of any kind..

So, you still don't believe in compression? It's not religion, it's proven - it DOES work, so you may just check out. Really, create black image 10000x10000 pixels. Save it as bmp - it will take a lot of space. Now zip it. Cool, isn't it? Now take any your favorite movie and recode it into RAW format. Oh crap, now it takes gigabytes of your space! How did that happen?? And then go read about MPEG.

Hello? He have an alien here... What? No! I'm not kidding - he IS an alien, he is blabbing some nonsense I could not understand. No! I'm NOT drunk! I'm tel- ah crap...
Really, man, you need some storage on your ACU to store unit schematics and command program. If this storage is damaged beyond repair - you just lost your army, your algorithms. Actually, that's a system error, your reactor go wild as there no more control over it and "kaboom!". To prevent this, you probably have a lot of storage devices embedded into your ACU unit replicating each other. Why not just store ACU unit schematics on same disks? You CAN do this, believe me. By one million means, and none of them is "magic".

http://en.wikipedia.org/wiki/Cyclic_redundancy_check
http://en.wikipedia.org/wiki/Replication_(computing)

If your arm is blown up you just build new one same as old one. Well it's not same arm - it's not composed from same atoms, but who cares. You know that YOUR arm consists from tiny biological cells dying and subdividing every second? So every second your arm is not same as it was second ago. But do you care about that if it sill works like second ago?

Again - go, clone yourself, that's easy. You do know that we don't know how exactly human body operates from macro to micro level, aren't you? And to even "repair" human we need a hell lot of research. And that research results in "human body schematics" (which are far from complete). And we can't actually rebuild human, we may just try to trick it's embedded self-replicating mechanism to create more-or-less same thing.

To "clone" a commander, you need to "read" a commander. Leaving aside question of such process possibility without destruction of original (hint: it's not possible), we still got a problem of constructing new commander. You can't just replicate it atom-by-atom - it's would be very long process and it's atoms already placed could decay in process. You need to do it fast and in right order. And for that you need a buffer of data read from commander, so you may analyze it and build more efficiently. But "buffer of data read" is the commander schematics.

Okay. Uhm. Yeah. You don't get it. Fine. It happens. I'm done trying to explain things that you clearly can not understand. Go read up on information entropy. Go read up on procedural generation. Learn what it means to clone a computer, and what the limits of compression really are. Heck, go have a little fun with Russell's paradox for a bit.

Come back when you've understood that, while such a purely self replicating, perfect memory self improving machine isn't technically impossible, a complete schematic for a Commander will necessarily contain its own complete schematic in one way or another. While storing a self referential schematic sounds like a lot of fun(hello infinite fractals!), at some point it suffices to say "This is what it is, figure the rest out some other time". That's when the infinite recursive data storage stops. Quite simply, there is no way for a Comm to store itself.

A Comm self schematic is not a complete schematic, and it can not be. The Comm's total systems, tweaks, and accumulated knowledge will always be greater than what its own memory can possibly store. This is the fundamental problem, because at some point a piece has to be generated ex nihilo. It can be easy, moderately difficult or very hard, as already explained before. But in all cases, the schematic depends on pulling some type of information directly from the source Commander. It could be a physical scan of the most complex components. It could be a software package for all the AI and units. It could be a complete copy of all the nanites. It doesn't really matter what it is, only that without it, the new copy is going to end up missing something from original, creating something that will either be forever inferior, or demanding that it both rediscovers and rebuilds the missing elements in its design.

I mean, it's simple robo genetics!

Considering how there are different types of commander it is likely that new commanders are built to become a base model of a commander.

Lacking the tactical experience of the parent and any field modifications that go along with it, new commanders would effectively have to grow into the position of commander by learning from the parent, so more like starting life as a sub-commander.

That way, commanders don't clone themselves in a way that is apparently impossible, and explain why every commander isn't exactly the same.

I would imagine Commanders also have personality disorders.

How do we know the commander wasn't made using procedural generation algorithms in the first place?

I mean, it's not unlikely that instead of whatever designed it treating it on a per-atom basis, that instead it created procedural algorithms for "leg design", "torso design", etc. Store a 3D size, armour thickness, etc, and let procedural algorithms fill in the little details and handle animation and weight distribution.

Even if it wasn't, we have the perfect procedural algorithm to make it already stored. If a computer was able to create the blueprints for a commander, then surely this commander (designed to be the absolute top of technology, able to have the computing power to control massive armies of thousands of units) could just re-create the exact thought processes that created itself?

Get the seed (the idea for a perfect machine of warfare), then run it through something to emulate the machine/s that created it through this vastly powerful commander. It's certainly not impossible - and even if would take absolute years of computing power to accomplish, maybe that's why Commanders don't constantly re-create themselves?

http://www.atomontage.com/?id=tech_overv

"It is critical for a modern volumetric game engine to provide a compression method that shrinks a typical voxel to less than one bit. Atomontage Engine features an elegant compression scheme that makes it possbile to store a large, multi-gigavoxel scene into a few hundred megabytes. The engine also provides efficient means to alter volumetric content in real-time and make it ready for physics simulation and rendering instantly. "