"fast" FTL travel and Von Neuman machines

[SomeGuy123]

     Observation : the visible universe seen by telescopes does not appear to be majorly edited by intelligent life.

Hypothesis : "fast" FTL travel is impossible.  The reason is simple.  In science fiction, it is common for there to be no ultimate speed limit.  More and more advanced alien races can get around quicker and quicker.  N00bs like humanity might take a few days to reach another star (like Starfleet), but there are methods that are faster and faster, to the point of reaching other galaxies in hours.  

What would be the consequence if this actually is possible?  (never mind existing knowledge of physics, let's just say there is an exploit not yet found that lets you basically teleport anywhere in the universe instantly)

Well, all it takes is one species with the ability to (1) teleport around near instantly (2) consume arbitrary solid matter to make self replicating robots.

Basically, once they have the technology for one and two, ZOOMP, whole universe is converted.  We as a species never come to exist because this happened long ago.  Ergo, since humans do exist, it's probably not possible to do (1).  (we basically already can do (2) as we ourselves are such replicating robots.  We just need some tools we already know how to make to let us turn arbitrary chunks of planets and asteroids into things we can eat and live in)

 

 

 

[K^2]

We already have technology for self-replicating matter-consuming robots. It's called life. It's absurd to think that robots are going to spread more rapidly, or consume more matter than living things, because a self-replicating robot is just another living thing. And it will follow all the same rules. We use technology to augment our ability to survive and replicate, but our entire planet isn't made up of people. It never will be. This is a silly concept.

[problemecium]

Perhaps the universe is self-stabilizing with respect to apocalyptic gray goo scenarios.

So a civilization invents hyper-ultra-warp and accidentally also makes a self-replicating robot. Within a few days their entire interstellar empire is besieged by self-replicating robots and they go extinct.

Ten million years later, another civilization arises in a galaxy a million light-years away. They proceed to go through their Kardeshev stages, miraculously avoiding self-destruction and eventually inventing their own hyper-warp. They explore the universe and find a distant galaxy populated entirely by self-replicating robots.

Aware of the danger posed by these robots, they send in their fleet of Death Stars and Halo rings and blow them all up.

Fortunately for us, the universe is a very, very, very big place. Even our own galaxy is big enough to house every popular sci-fi civilization concurrently, and there are billions of galaxies in the known universe, presumably accompanied by billions more undiscovered. Even if a civilization invents very fast travel, it will still take some amount of time for anything they do to propagate through all that space. The universe as they know it may fall into ruin very fast, but never infinitely fast, and therefore over sufficiently large distances it will still take a long time for the damage to spread. Combined with the vast timescales available for civilizations to rebuild themselves and for new ones to appear, it seems believable to me that the universe could simply tough out events like this and go back to its usual rhythm.
It's easy to say things are impossible when they have scary implications, but the simplest answer often is wrong and gets in the way of progress

[SomeGuy123]

2 hours ago, K^2 said:

We already have technology for self-replicating matter-consuming robots. It's called life. It's absurd to think that robots are going to spread more rapidly, or consume more matter than living things, because a self-replicating robot is just another living thing. And it will follow all the same rules. We use technology to augment our ability to survive and replicate, but our entire planet isn't made up of people. It never will be. This is a silly concept.

Try reading Nanosystems, K^2.  You're talking outside your area of expertise.  To summarize, the world's leading expert on the subject points out that vastly faster optimizations on the replication speed of life is possible.(he works it out based on fundamental theory and points out the absurd inefficiency and floppiness of existing ribosomes)

  The reason life isn't this fast is because (1) it isn't intelligently designed, and the kind of replicating machinery that would be orders of magnitudes quicker would be radically different from anything evolution has reached (2) current life depends on liquid chemistry for legacy reasons (3)  designs that are faster than existing life by several orders of magnitude by using stiffness and tightly designed feeding chains would be very vulnerable to design error.  A really well made, high performance machine will fail completely if any part is wrong.  So evolution can't random walk it's way there, as it is sort of a skyscraper on the evolutionary landscape surrounded by a vast moat of dead zones.  

Ironically, real self replicating nanomachines might easily be so fast that waste heat becomes the limiting factor.  You'd be burning megawatts per cubic inch if you can reject the heat fast enough.  

 

Edited February 14, 2016 by SomeGuy123

[K^2]

11 minutes ago, SomeGuy123 said:

So evolution can't random walk it's way there, as it is sort of a skyscraper on the evolutionary landscape surrounded by a vast moat of dead zones. 

I've heard this argument before. From creationists. In fact, that's pretty much their entire argument.

I'm prepared to listen to experts on these matters when we have them. When they figure out how to create a self-replicating machine from scratch, or at least, greatly modify existing ones to do anything remotely close to the claims.

So far, every attempt has simply revealed a new limitation that real self-replicating systems have to deal with.

[SomeGuy123]

19 minutes ago, K^2 said:

I've heard this argument before. From creationists. In fact, that's pretty much their entire argument.

I'm prepared to listen to experts on these matters when we have them. When they figure out how to create a self-replicating machine from scratch, or at least, greatly modify existing ones to do anything remotely close to the claims.

So far, every attempt has simply revealed a new limitation that real self-replicating systems have to deal with.

We have self replicating machines today, and they have the properties described.

They don't need to be nanoscale, and qualified experts never thought they would be.  (some day, pieces of the machines would be nanoscale but the overall system would not be)  We currently do in fact have industrial equipment sophisticated enough to perform every task needed to produce the equipment itself, and partial automation - enough to prove that self replication is possible.  There just isn't the money or economic need for full 100% self replication at the present time, but there are massive factories that make every component used in those factories, and robots that can duplicate any task a human hand is capable of doing, so long as sophisticated judgement is not required...(you would reject any part that doesn't meet close enough tolerances so that you can use the same motion on every part going through that particular station)

So yeah, we have such experts right now.  

As for creationists...the reason their beliefs are wrong is because scientists can show how evolved mechanisms did get their, piecewise.  Each predecessor, simpler version of something evolution made was functional in a simpler form.

In the evolutionary world, it turns out that plain RNA, without anything else - just RNA in a test tube - is self replicating.   Certain magic RNA sequences are able to catalyze copying themselves because they conform into a certain functional shape.   Reason the creationists are wrong is because the special environment that nature operates in, and the chemistry it uses allows for gradual evolution.  Intelligently designed equipment that humans make does not allow for this.

Take a rep rap 3d printer.  How well's that printer going to work if half it's parts are missing?  Describe to me a version of the printer that is simpler that still works.  All a self replicating machine is something like a 3d printer, a set of assembly robots, a CNC mill, some chemical reactors, plasma furnaces...it's a lot of machines that the sum total of all their collective efforts is self replication.  We already have all of these machines today, we just don't have the money and design resources to make them closed loop in this way.(and they ARE closed loop, actually, just no single facility has all of the equipment under the same rooof)  A nanomachinery version is just the same equipment, shrunken...

So no.  You aren't qualified to make any other comments on this subject.  You're wrong about the lack of experts - plenty of industrial and process engineers today are more than qualified.  You're wrong about nature.  You're wrong about operation speeds.  You're wrong about the design fragility of intelligently designed systems.  You're basically just blowing hot air.  Aren't you like a grad student in physics?

Edited February 14, 2016 by SomeGuy123

[K^2]

You are still falling into the same fallacy. Describe to me how a human survives to reproduce without a digestive system? You are grabbing a single element, and focusing on it.

If factory-style reproduction was evolutionary advantageous, it would evolve. Yes, from simpler self-replicating systems, going back to strands of wild RNA in a puddle of organics. Same way we went from single self-replicating gene to a self replicating cell, with ton of machinery without which it couldn't work. To multi-cellular organisms that couldn't work without a single type of cell. To complex organisms that couldn't function without a specific organ. And then yes, into factories, that couldn't function without a specific gear.

Evolution is just an optimization algorithm solving the problem of optimizing energy consumption. And what we have is about as efficient as it practically gets, without either falling apart due to errors, overheating, or running out of resources.

[SomeGuy123]

6 minutes ago, K^2 said:

You are still falling into the same fallacy. Describe to me how a human survives to reproduce without a digestive system? You are grabbing a single element, and focusing on it.

If factory-style reproduction was evolutionary advantageous, it would evolve. Yes, from simpler self-replicating systems, going back to strands of wild RNA in a puddle of organics. Same way we went from single self-replicating gene to a self replicating cell, with ton of machinery without which it couldn't work. To multi-cellular organisms that couldn't work without a single type of cell. To complex organisms that couldn't function without a specific organ. And then yes, into factories, that couldn't function without a specific gear.

Evolution is just an optimization algorithm solving the problem of optimizing energy consumption. And what we have is about as efficient as it practically gets, without either falling apart due to errors, overheating, or running out of resources.

What's your background?   You're wasting both our times posting about things you don't understand.

Factory style reproduction can't evolve because of how evolution works as a process.  I'm not going to try to explain it to you - go take a course in it.  I gave you a big hint when I mentioned dead zones, you should have already remembered if you had the background.  You don't have a clue how evolution actually works because you either didn't take evolutionary bio or you didn't have a good course in it.

Evolution is a mathematical algorithm and it can't just do anything, especially when it's constrained by the needs of physical replication like life on earth is.  It has limitations.  This in no way supports the arguments creationist nutjobs, but it does mean that life can't just be and do anything that is possible within the design space.

Edited February 14, 2016 by SomeGuy123

[magnemoe]

17 minutes ago, SomeGuy123 said:

Try reading Nanosystems, K^2.  You're talking outside your area of expertise.  To summarize, the world's leading expert on the subject points out that vastly faster optimizations on the replication speed of life is possible.(he works it out based on fundamental theory and points out the absurd inefficiency and floppiness of existing ribosomes)

  The reason life isn't this fast is because (1) it isn't intelligently designed, and the kind of replicating machinery that would be orders of magnitudes quicker would be radically different from anything evolution has reached (2) current life depends on liquid chemistry for legacy reasons (3)  designs that are faster than existing life by several orders of magnitude by using stiffness and tightly designed feeding chains would be very vulnerable to design error.  A really well made, high performance machine will fail completely if any part is wrong.  So evolution can't random walk it's way there, as it is sort of a skyscraper on the evolutionary landscape surrounded by a vast moat of dead zones.  

Ironically, real self replicating nanomachines might easily be so fast that waste heat becomes the limiting factor.  You'd be burning megawatts per cubic inch if you can reject the heat fast enough.  

 

True, life is not optimal, its however resistant and adaptive. An nano scale von-neuman machine would be bound by solar output too. It would also get problems getting power and resources at the same time.
Liquid chemistry has the benefit of low energy cost. Trying to build something of diamond using the solar power falling on you and you will not be fast.
Main benefit with nanotech as an life form is that would work in space. 

[SomeGuy123]

Just now, magnemoe said:

True, life is not optimal, its however resistant and adaptive.

Actually it's neither.  One constraint of evolution in a confined environment like Earth is actually a nasty form of version lock in.  TLDR, everything alive on Earth actually is only 1 branch of many possibilities, and it can never leave that branch.  The reason is that the codon space is full and the capability to migrate an existing genome to a new codon base space is too complex a task for any mechanism to evolve to accomplish it.  So everything on earth is actually stuck on the same basic set of amino acids and has been stuck for several billion years...  A lot has been accomplished but it's nothing like K^2 thinks is possible.  

[K^2]

1 hour ago, SomeGuy123 said:

What's your background?   You're wasting both our times posting about things you don't understand.

Factory style reproduction can't evolve because of how evolution works as a process.  I'm not going to try to explain it to you - go take a course in it.  I gave you a big hint when I mentioned dead zones, you should have already remembered if you had the background.  You don't have a clue how evolution actually works because you either didn't take evolutionary bio or you didn't have a good course in it.

Evolution is a mathematical algorithm and it can't just do anything, especially when it's constrained by the needs of physical replication like life on earth is.  It has limitations.  This in no way supports the arguments creationist nutjobs, but it does mean that life can't just be and do anything that is possible within the design space.

I am not at all an expert in the field. I have decent background in evolution algorithms, as well as a number of other algorithms that optimize similar functionals. I have taken courses in cellular biology and genetics. I have done laboratory work on gene sequencing and splicing. And there were other misc bio classes here and there. Now, as limited as that scope is, you come off as a person who've taken a single course on the subject and considers himself an expert without any practical experience.

I am well aware of how local optima work and difficulties they present in achieving true optimal solutions. It's not just a buzz word to me, like "dead zones" seem to be to you. What I also happen to understand is that there can be no local optima preventing factory-style reproduction from evolving, because that's how every single living cell reproduces. Unlike complex organisms, cells do not consist of self-reproducing components. They rely on an entire process for all of their components to be assembled, starting with genetic material transcription. Why don't you go and look up how ATP is produced in animal cells, and they come back and tell me that a factory cannot evolve.

[SomeGuy123]

7 minutes ago, K^2 said:

I am not at all an expert in the field. I have decent background in evolution algorithms, as well as a number of other algorithms that optimize similar functionals. I have taken courses in cellular biology and genetics. I have done laboratory work on gene sequencing and splicing. And there were other misc bio classes here and there. Now, as limited as that scope is, you come off as a person who've taken a single course on the subject and considers himself an expert without any practical experience.

I am well aware of how local optima work and difficulties they present in achieving true optimal solutions. It's not just a buzz word to me, like "dead zones" seem to be to you. What I also happen to understand is that there can be no local optima preventing factory-style reproduction from evolving, because that's how every single living cell reproduces. Unlike complex organisms, cells do not consist of self-reproducing components. They rely on an entire process for all of their components to be assembled, starting with genetic material transcription. Why don't you go and look up how ATP is produced in animal cells, and they come back and tell me that a factory cannot evolve.

 What you're missing - because you didn't read nanosystems - is that Drexler is talking about machinery that is tight.  As you must know, per your stated background, living systems rely heavily on diffusion and random processes.

A true factory has a direct supply chain and active transport of every reactant and product.  It would run like a gatling gun fed by a belt instead of relying on concentration gradients to eventually bring together the reacting species.

That's the main way you get a performance boost of orders of magnitudes.  That's what I'm talking about.  A cytoplasm full of components sharing a common pool - or even a sub pool like organelles - is not as quick as this.

Also, the second property is stiffness.  This is a function of temperature and the materials used.  A hypothetical nano machinery system would use parts that are stiff - meaning they flex only between useful conformations.  Out of every state of a biological molecule - something you'd know about if you've done any protein folding or NMR studies, something I have done - only some of those states are useful and the state transitions are semi random.

[SomeGuy123]

5 hours ago, K^2 said:

We already have technology for self-replicating matter-consuming robots. It's called life. It's absurd to think that robots are going to spread more rapidly, or consume more matter than living things, because a self-replicating robot is just another living thing. And it will follow all the same rules.

Back to the topic at hand. 

Let me restate your argument.

It's absurd to think that airplanes will fly more rapidly than birds or have more range, because an airplane is just another flying thing built by living beings.

Similarly, a robot that can self replicate is going to use nuclear energy or vast swathes of solar fabric less than a millimeter thick for power.  It would use high temperature radiators (for vacuum) or powerful fan to pump cooling air through (in atmosphere) it's internal radiators.  It would "digest" it's food by converting it to plasma first and separating it into individual elements.  It would be smart enough to understand it's own internal processes and would be capable of redesigning them if needed.  

How, exactly, do you think such a robot would be the same speed at consuming matter as living beings?

Edit : I thought about it, and I may have figured out your reasoning error.  I thought it was due to ignorance - you genuinely thought that evolution was sure to find the best possible arrangement of matter for any self replicator.  

But actually what you're thinking is that if we had to make flying machines that work the same way as birds and were made of the same materials (bone, feathers, muscles), our "ornithopters" would probably be inferior in overall performance and maneuverability to this day.  

Ditto if we had to make machines that are made of amino acids, must work in salty water, and have to be able to self replicate with cellular sized subunits - if we can beat nature's solution it won't be by very much.

And you've made an argument from authority.  I'm telling you that if a robot is a bunch of mechanical parts and circuit boards and wiring, well, you can make most of that stuff with CNC machines and fully automated assembly lines.  We don't have self replication yet but we could.

But since nobody has quite succeeded yet, and some of your bioscience guys like to go on about just how complex current life is (much of which must be to survive in the hostile competitive environment of earth's biosphere...) you like to feel superior in thinking self replicating robots are far away and will be no better than life is.

 

Edited February 14, 2016 by SomeGuy123

[K^2]

It's a question of constraints. Yes, I'm absolutely sure we could engineer a machine that will briefly be capable of reproducing at far greater rate than any living organism. Hell, we could probably find a way to just throw an existing living cells into overdrive and get a factor of ten at least on their reproduction rates. But living cells don't do that, despite increasing in number being the primary goal of a living thing. Why?

You are mostly right about it being the matter of how building materials and energy are supplied. Where you are completely wrong is suggesting that all we need to solve is build a conveyor. Living cells are perfectly capable of doing that. You seem to focus on simplest living organisms, that are entirely at a mercy of environment and random processes. But eukaryotic cells have overcome that billions of years ago. They have literal conveyor belts, cables, pipes, and other delivery systems built within the cells. Ones that take necessary materials and send them to where they are needed fast. Everywhere it made enough of a difference to bother with, it has been done.

What it really comes down to is adaptability. Living things need to be adaptable to survive changing environment conditions. A lot of evolutionary "decisions" seem like they are a waste of resources, but really, they are critical in making sure that the organism doesn't survive just today, but a million years from now. And a Von Neumann probe is going to be the subject to same restriction. If some alien species builds machines that are simply better than living things at making copies of themselves by sacrificing adaptability, yes, they'll be able to spread rapidly. Then they'll get wiped out by the first unforeseen event, change in environment, or even an external agent. In order to actually survive and keep spreading through the stars, they have to adapt. They have to evolve. And that's just life.

Also, if you think airplanes are more efficient than birds, I suggest you mention that to an aeronautical engineer and watch his response. If there was ever a biological need for birds to fly faster, they would have. As it stands, they fly quieter, stay in the air longer, and consume far less energy doing it. Because these were the specs.

[SomeGuy123]

3 hours ago, K^2 said:

And a Von Neumann probe is going to be the subject to same restriction. If some alien species builds machines that are simply better than living things at making copies of themselves by sacrificing adaptability, yes, they'll be able to spread rapidly. Then they'll get wiped out by the first unforeseen event, change in environment, or even an external agent. In order to actually survive and keep spreading through the stars, they have to adapt. They have to evolve. And that's just life.

You're neglecting vast swathes of details.  Vast swathes.  Evolutionary algorithms are slow.  The robots could be, and probably would be actually sentient and would be capable of self redesign.  DNA doesn't store very much data - you could pack in far more contingencies and "adaptability" code segments and blueprints into a robot that has on the order of terabytes or petabytes of onboard memory instead of a few gigabytes...

Details matter, not just theory.  That's where I think we really disagree.  I have a pretty clear mental model of what I think a von neumann machine would actually look like.  (TLDR, it's a sentient robot spacecraft with redundant modular systems and some of those systems are machines that can digest chunks of asteroid, convert it to pure feedstock of various gasses, and then the gasses feed into the nanoassembly systems that are able to make the parts for any component of the spacecraft.

The spacecraft's brain would be loaded with the algorithms - the fundamental thought processes and tools and ideas of the engineers who designed it.  At some level of tech we'll be able to have something like an encyclopedia that doesn't just store facts and procedures, it stores snippets of logic ripped from the actual brains of people capable of using those facts and procedures skillfully at an expert level.  So with such an encyclopedia and the right hardware, you'd be able to literally do anything already known...

Nature has nothing comparable at all.  

 

[micr0wave]

12 minutes ago, SomeGuy123 said:

DNA doesn't store very much data

1mg DNA = 1 billion CD's

[fairytalefox]

So what this discussion is about? FTL traveling doesn't seem to be possible, I can say it right now. Light speed definitely isn't an arbitrary barrier, it's a part of what time and space are. I really don't see how this can depend on DNA capacity, grey goo growth rate or number of angels dancing on a needle.

[SomeGuy123]

5 hours ago, micr0wave said:

1mg DNA = 1 billion CD's

It doesn't store much data because in actual living multicellular organisms, all the DNA is just shared in common between the cells (with certain minor edits during differentiation).

So there's a lot of total data but it's only a few gigabytes of unique information.

[SomeGuy123]

5 hours ago, fairytalefox said:

So what this discussion is about? FTL traveling doesn't seem to be possible, I can say it right now. Light speed definitely isn't an arbitrary barrier, it's a part of what time and space are. I really don't see how this can depend on DNA capacity, grey goo growth rate or number of angels dancing on a needle.

Because we don't know that, maybe our understanding of space-time is wrong or incomplete.  There could be "tricks" we have not discovered as a species.  But we do know the basic trick of "self replication given resources, and those resources are mainly commonly available elements".

 

[K^2]

8 hours ago, SomeGuy123 said:

So there's a lot of total data but it's only a few gigabytes of unique information.

And just how much do you need? And how are you going to do error-checking on all that data during replication?

8 hours ago, SomeGuy123 said:

A true factory has a direct supply chain and active transport of every reactant and product.

Eukaryotes have that. They have active transport and direct supply. Every single living cell in your body works exactly the way you describe, and you are still trying to prove to me that it's impossible to evolve. Unless your thesis is that original Eukaryote was created by God, you already lost this argument.

[SomeGuy123]

27 minutes ago, K^2 said:

Eukaryotes have that. They have active transport and direct supply. Every single living cell in your body works exactly the way you describe, and you are still trying to prove to me that it's impossible to evolve. Unless your thesis is that original Eukaryote was created by God, you already lost this argument.

That's like saying birds have wings and muscles, therefore they are just like jet engines and airfoils.

Edited February 15, 2016 by SomeGuy123

[K^2]

37 minutes ago, SomeGuy123 said:

That's like saying birds have wings and muscles, therefore they are just like jet engines and airfoils.

They do have airfoils, and they have exactly the same aerodynamics limitations as airplanes. There is no magic in physics.

Again, you are not addressing the fact that every Eukaryote is literally a factory. Your entire complaint from the start was that a factory can't replicate without any one component, and therefore, cannot evolve. A Eukaryote cell is literally that. I am waiting for a detailed rebuttal, where you explain the difference between Eukaryotic cell and a factory.

[fairytalefox]

8 hours ago, SomeGuy123 said:

Because we don't know that, maybe our understanding of space-time is wrong or incomplete.  There could be "tricks" we have not discovered as a species.  But we do know the basic trick of "self replication given resources, and those resources are mainly commonly available elements".

 

You see, speed squared plus "time speed" squared is always c squared. You can't actually change your full speed, you can only split it between moving faster in space and moving faster in time. For me, it looks like a conservation law, or like another face of the conservation law maybe. We don't try to trick conservation laws, because it never works. It's how I see this.

[Bill Phil]

Here's the problem: 

A: we've only seen a tiny portion of the universe, and we haven't even mapped out entire galaxy yet.

B: a Fast FTL might be limited and will likely require large amounts of energy, no free lunch after all.

You're also assuming FTL is the only way to get around in the universe. If we can overcome the energy losses in a Bussard Ramjet, a Tau Zero like scenario could occur. Accelerate for a few hundred trillion years.... Of course there's not uniform density, however you could potentially keep a plasma shield in front of you at all times, with enough power generation. Magnetic bottling and such...

Now we may not be able to overcome the energy losses, but time is a resource as much as anything else. And even if a starship takes a millenium, it still gets there. But it's still a huge engineering hurdle.

Edited February 15, 2016 by Bill Phil

[SomeGuy123]

3 hours ago, K^2 said:

They do have airfoils, and they have exactly the same aerodynamics limitations as airplanes. There is no magic in physics.

Again, you are not addressing the fact that every Eukaryote is literally a factory. Your entire complaint from the start was that a factory can't replicate without any one component, and therefore, cannot evolve. A Eukaryote cell is literally that. I am waiting for a detailed rebuttal, where you explain the difference between Eukaryotic cell and a factory.

I already gave one.  Let me reiterate.

The factory I'm talking about uses hard carbon or metal components, everything tightly integrated like a watch.  Thousands and thousands of steps are integrated together and the thing is hyperfast.  Pieces of the factory do fail, and digital watchdog systems detect this and order the failed components recycled.

Like this.  

A floppy eukaryotic cell is nothing at all like this.  At all.  Again, it's like the difference between a mitochondrion in a bird and a jet engine.  The parts are different, the principles are different, the energy storage mechanisms are different - the only thing in common is both are combusting hydrocarbons for energy.

The cell is floppy, it won't fail if a single atom gets out of place, it isn't cooled to near cryogenic temperatures, it isn't made of metal and diamond, it doesn't have digital sensors and exact control logic (mRNA regulation is statistical), it isn't in hard vacuum, it can't be intelligently redesigned, it can't operate using most of the elements on the periodic table, it won't have digital error correction between generations to reduce the probability of mutation to zero...

I'm talking about machine phase chemistry.  You're talking about goop in a bag.  There are about alike as jet engines and mitochondria.

And yeah, just like jet engines guzzle hydrocarbon fuel by the ton, nano machinery factory would burn through energy and feedstock at incredibly voracious levels.  They would burn far more energy to, say, make an ingot of steel than a steel forge uses for the same task.  (but the ingot would be perfect).  Which doesn't matter - living cells don't have nuclear reactors or vast arrays of solar panels plugged in to them via power cables...

Also, something else you might have missed : the minimum "replication subunit" of a nanomachinery factory is still a bunch of equipment.  It's a plasma furnace to digest matter.  A bunch of chemical reactors to get it to the atomically pure feedstocks.  Probably multiple stages of filtering.  (all sorts of high end equipment to do that - lasers, calutrons, etc)

There are still 3d printers - they make the bulk stuff like frames and tank walls and casing and things where atomic precision isn't needed.  And then the core machine, the nanofactory, is hundreds of thousands of separate assembly lines that converge on one another. Some assembly stations in the factory can be state adjusted - ON/OFF/A/B - which is how the factory can make multiple outputs.  (different final products have modular components added or missing in sockets on them)

The whole "nanofactory" isn't nano at all, once you include all the assembly lines and all the ports for input gases and coolant and power and all the duplicate backup assembly lines, the thing is desktop printer sized or larger.  Any smaller and it's not functional.

There are then robots as well.  And a control computer.

This whole machine weighs over a hundred tons.  Remove any major part, and it will not replicate itself.  It is too complex for nature to design via evolution, at least not without more time than the sun has left.  It doesn't compete with living cells in the same environment and the "nanomachinery" isn't separate free running cells, everything must be welded down and connected to power leads and be heat sinked...

I think maybe you had some sci fi nanobot the size of a eukaryotic cell roaming around free in mind.  I will agree with you there - the environment at that level doesn't have enough energy to do things this way.  The "nanobot" would never get enough energy to copy itself at any decent rate since it is having to synthesize diamond, etc.  Also, it would just jam and fail since it's outside the vacuum chamber it would be intended to operate in.  So you'd have to make it more flexible to survive in the living biosphere on earth, and after you do all that you end up with something that is marginally better than existing life at best.

Edited February 15, 2016 by SomeGuy123