Self-replicating Solar Factories
In our last two threads on green energy, it was proposed that we seed the Moon with self-replicating solar factories. Is anyone seriously attempting to do this? There was the NASA study in 1981, and a more recent (very technical) study in 2004.
Of course, Earth is full of self-replicating solar factories. They’re known as plants.
Consider an orchid, such as the Phalaenopsis I’m growing from the local grocer; it subsists on ice cubes. A stalk with three boring leaves, I thought it would be done after the blossoms fell. But surprise, it sent out two shoots each of which developed three huge buds, oddly asymmetrical. Over two weeks, the buds swelled, and you could see faint patterns of spots underneath the outer greenish sepals. Then one day, over the course of the day the sepals opened into a perfectly formed flower, complete with the little pink pedestal for the insect it needs to pollinate. A self-replicating factory for orchid flowers.
If anyone thinks Earth needs yet one more good reason to save its biosphere, surely an orchid is one. Nowhere else in the universe is there an entity that makes something as amazing as an orchid flower.
At the same time, the orchid reminds us that a couple of ice cubes a week can sustain a solar replicator. What other examples from biology could give us clues for future sustainable technology?
Comments are closed.
Ultraphyte 
Why ice, rather than water for your orchid?
Regarding self-replicating factories, we are not even close to being able to build one. The nearest I’ve read about is a desktop fab than can make all the mechanical parts for another fab. No electrical or electronic parts and human assembly still required.
Although we think of biology as complex, they have relatively few parts. A yeast has around 6500 genes with several multiples of that number of proteins due to alternative splicing. Humans have less than 30k genes with perhaps 10x as many proteins. True, we have DNA to build, and RNA too for transcription and regulation, but as machines we are less complex than a space shuttle by far.
If we can consider simpler machine systems, I could see autonomous robots building solar PVs for as long as they can last. They would be topped up with new shipments to replace failed ones.
Yes, can I second this question. I don’t grow orchids myself but my mum has a fairly impressive collection of them and they’re not watered much but do get a fair bit of spray when it’s hot. Afaik some/most are parasitic in their native habitat and like moisture so perhaps ice cubes are a kinda cheap timed release mechanism. Either way I’d like to know if it’s a titbit of knowledge I can pass on to my far more knowledge mother!
Orchids are grifters. They are parasitic in the sense that all species parasitize fungi to grow from an embryo to a plant (unless you’re raising one on cultural medium). Thereafter, some are free-living, some are intermittently or permanently parasitic on fungi. I don’t know of any orchid that has a mutualistic relationship with a mycorrhizal fungus, where the plant gives anything back to the fungus. Given the diversity of fungi, I wouldn’t be surprised if there are some, particularly among the most primitive orchids.
Beyond that, orchids often parasitize their pollinators, in the sense that they trick pollinators into moving their pollen without providing a reward, through things like fake sex, sticking the pollen onto a bee’s head with a hammer, and so on (in the last case, the male and female flowers look different, as bees learn very quickly not to put their head under the pollen hammer). Thing is, orchids transfer pollen in big lumps rather than individual grains, so if they’re lucky once, they produce thousands of seeds.
This is one reason why there are so many orchid species–they only have to be lucky once, and there are potentially enough offspring for a new species. They need that diversity too. It helps keep the fungi and pollinators from developing counter-measures to all their myriad scams.
Ice–because the ice melts slowly, so the potting medium hydrates gradually, rather than all at once. A good example of extremely low-tech solution to an age-old problem, how to avoid overwatering an orchid.
About the biology as complex, actually there are more parts than we think, because the “junk DNA” is full of control elements, maybe a dozen per gene. And the interconnections are incredibly convoluted. It’s not just the number of parts, it’s the number of connections.
It’s not just the number of parts, it’s the number of connections.
That is true of machines too. A watch isn’t a pile of pieces, it is an assembled object that must have all the parts assembled correctly. Steam and combustion engines must also have various parts tuned, e.g. regulators, fuel injectors. I will accept that the “junk DNA” is involved in gene regulation, but how much more complex does this make the device? A carpenter might argue that choosing the right piece of wood, from te right tree and the right part of the tree, plus his skill in working the work in a certain way is necessary for the roduction of a finished piece, e.g. a musical instrument.
I’m not even sure the number of connections captures the complexity, I suspect it’s the number of paths through the graph made by those connections.
The space shuttle has a lot of parts, but by design its subsystems are largely segregated, and the subsystems of subsystems are segregated, and so on, on many different levels. This keeps the complexity of engineered systems manageable in a way that the complexity of biological systems is not.
Did anybody ever have a satisfactory discussion with a proponent of intelligent design about why their designer hadn’t taken Dijkstra’s “goto considered harmful” to heart? I tend to smugly think evolution doesn’t hold a candle to engineering, but maybe that’s hubris. Sometimes I am haunted by a fear that this is actually important: that a system that isn’t segregated can be better than any system I have any hope of understanding.
Alex is correct about the state of the art in sintered component printing. There’s no reason why we shouldn’t step back a bit and make PCBs using some of the older metal-spraying techniques rather than present lamination and etching techniques. Similarly, other than “no-one’s done it yet”, there’s no reason why a robot shouldn’t assemble sintered print componentry.. Which leaves the main stumbling block being electrical and electronic parts and adding same to the PCBs. With present surface mount components, this is also sometimes done by robots.
Conclusion – it’s not been done yet, and forget self-replicating nano-machines, but I think most of the technologies exist even if they’ve not been put together.
Circuit boards are one thing, integrated circuits are quite another. I really don’t see a robot fabbing the microcontrollers needed to make a duplicate of itself any time soon.
Which is analogous to organisms not creating all their compounds, but ingesting them as a food. Vitamins and some amino acids are well known examples for humans.
I agree. We’re either going to need to keep sending up high-value components like microcontrollers or need to develop an entire ecology with hundreds or thousands of different species of robots replicating different types of parts.
I’m left wondering whether or not you’re actually aware that an IC is a highlly miniaturised example of a PCB. I’ve not addressed the issue of making, say, transistors, but present-day ICs are only 70 years from the first transistors.
I am well aware of the history and technology of semiconductor device manufacturing. And, aside from nitpicking about the difference between a PCB with discrete circuit elements and an integrated circuit, I’m left wondering whether we disagree or not. I suspect we don’t, actually. You seem to consider 70 years as a short time, in the context of this discussion. In that sort of timeframe, lots of things become possible, perhaps even fully self-replicating robots, integrated circuits and all. In the spirit of the report linked in the original post, if not quite analogous to the self-replication ability of an orchid, I was (rather optimistically) thinking of robotic materials processing facilities we could start designing today, to be launched in 5, maybe 10 years’ time. In that timeframe, I think structural components, as well as crude conductors and semiconductors, all made on the moon from lunar raw materials, should be technologically possible. The extraordinary mechanical precision, material purity, and crystal perfection achieved in a modern chip foundry, or even in the comparatively primitive chip foundries of 50 years ago will almost certainly not be possible on the moon within 10 years.
The other issue with current machines is that humans are needed in the system to do quality control on a per-hour basis. This includes things like adjusting cutting blades, diagnosing complex faults, and so forth.
I don’t know of any good way to automate QA/QC, and the complexity of the medical industry (which effectively does human QA/QC) makes me think that automating it will be difficult, especially on systems where the complexity approaches that of a living body.
Does “the number of connections captures the complexity”?
The power of connections is that you *multiply* the complexity, rather than adding. Also you get factorials and powers. If you have a thousand genes each with twelve regulators, then each gene gets multiplied by twelve factorial (12!) number of different states. If a thousand genes can be interconnected with each other, it’s something like a thousand factorial (1000!). The numbers boggle the mind.
Can “a system that isn’t segregated can be better than any system I have any hope of understanding”? That’s already been done. Software evolution has produced programs so complex no one can understand them, yet they perform better than what the software engineers could write.