I Am Not A Biologist (IANAB). When you see IANAB, I am reminding you that if an imaginative biologist (say, George Church) is sure that what I say is impossible BS, they are right and I am wrong. OTOH, many very-narrow-focus biologists do not think technologically, or beyond the limited horizons of the tools in their labs in 2016. I work on silicon; we make 14 nanometer lines and spaces with 193 nanometer light. We are developing tools for extreme ultraviolet lithography using 13.5 nm light; perhaps 1 nanometer devices will become possible in a decade or two. I do not pretend to guess what 7 billion clever people can do with such tools, beyond the likelihood that it will be awesome. I want to make it easier for clever biologists to do cheap awesome safely.
Experiments with radically engineered microbiology may be too dangerous to perform on Earth. Whether current fears of genetic modification are overblown, it is possible that very advanced future experiments may be extremely risky. Even a tiny chance of reproducing biological material escaping confinement could have ecosystem-destroying consequences. Again, it is very unlikely that we will be able to perform such experiments soon, but we should develop safe places to perform them soon.
Two safer places to perform these experiments are in lunar orbit (for "unintelligent" experiments) and deep Jupiter orbit (for potentially malicious "intelligent" experiments). These locations are "shielded" from Earth by gravity wells and "delta V", requiring high energies and reaction mass for dangerous objects to return to Earth.
ΔV to Earth
1 km/s around Earth
13 km/s around Sun
"ΔV to Earth" is the estimated velocity change needed to reach Earth from a low orbit around either body. Assume that malicious AI can rebuild itself with the materials at hand, and provide "spoof" telemetry, but also that it cannot violate physical law, create atoms from nothing, spoof multiple high-resolution distant radars about its location, or resist a nuclear interceptor arriving at high speed.
All I'll say about Jupiter: An AI might escape to the lunar surface and build a means of escape, so we should not build potentially malicious AI there. If a Jupiter AI "escapes" into the Jovian atmosphere, that seems like a kinetically unsurvivable trip, and an unlikely place to build objects capable of escaping into the solar system.
While nothing is 100.00% escape proof, lunar and jovian orbits seem more escape resistant than a lab on Earth, and a better place for such extremely dangerous experiments. I will not belabor the Jupiter case, or attempt to construct concrete examples; I will leave that for philosophers and other adolescent individuals to argue about among themselves. Please do not bother me with your speculations unless you can provide detailed CAD drawings, material suppliers, and construction procedures; I have more important things to do.
From here on, I will focus on lunar labs and the opportunities for mid-term (within a century) experimental biology, for bugs purposely designed to be dependent on externally-supplied computation and resources, and cryptographically prevented from surviving mutation and engaging in Darwinian evolution. Indeed, the purpose of distant orbiting labs will be proving that with observational and mathematical certitude.
Lunar Orbit Labs
A lunar orbit microbiology lab will be capable of synthesizing DNA and inserting it into a synthetic microbe "platform", then observing the results. It will have solar panels, some power storage, minimal shielding for radiation protection and thermal inertia, and high bandwidth laser communication to and from Earth. Transfer vehicles from Earth may add new modules, replenish materials, or remove, arc-sterilize, and deorbit old modules into a designated "dumping ground" on the Moon.
Extreme caution is the first priority - the only thing leaving the lunar lab will be information, which can be used for similar experiments on Earth, after they have proven safe in the lunar lab.
Orbit: The radius of the Moon is 1740 km, and the Earth-Moon L1 Lagrange point is 58000 km towards Earth. While orbits closer than 100 km, may be unstable due to uneven gravity from mass concentrations ("mascons"), orbits with inclinations of 27°, 50°, 76°, and 86° are stable "frozen" orbits. A microlaboratory at 200 km altitude (1940 km radius) and 27° inclination will be semi-stable for a long time, above the usual Lunar mission parking orbits, and have relatively low closing velocities with other missions to and from the Moon. The orbital velocity is 1590 m/s, and orbital period is 97 minutes.
Beginning tasks for this laboratory:
- 1) Synthesize DNA de-novo, and insert it into artificial cells.
2) Synthesize DNA-2.0, recoded DNA with a new set of transfer RNA that use different DNA codon triplets to select the same amino acids. Such lifeforms will be recombinant- and virus-incompatible with normal Earth DNA.
3) Synthesize DNA-3.0, recoded DNA which uses extra amino acids not found in nature, produced by reactions (temperatures, pressures, reagents, elements) not compatible with biological life. Such lifeforms cannot replicate without artificial support.
4) Synthesize Checksum DNA-4.0, DNA that synthesizes RNA indirectly, using the DNA-3.0 sequences scrambled with a checksum of the entire DNA string. Any changes to the DNA string will result in garbage proteins; these artificial lifeforms would not be able to successfully mutate; their sequences must be generated with cryptological calculation. They can evolve, but only with the help of powerful computers.
4a) Self-repairing DNA-4.0, with tools to detect and correct DNA errors. This DNA can be conditionally immortal, and with additional tools for repairing cells, or cellular disassembly and rebuilding, the cells can be too.
5) Robot Cells with External Repair: A "ROCER" is the cellular equivalent of an ant in a colony; it cannot reproduce itself, but can detect damage, stop doing its task, and call for help. A damaged ROCER signals for transport to a fully capable ROCER repair cell. A ROCER can be much smaller than a cell, perhaps the size of a large virus, and performs simple tasks, relying on repair cells and other ROCERs for many functions that a natural biological cell performs internally and independently.
Starting with these tools, non-Darwinian artificial lifeforms that cannot evade their programming, we can start working on important tasks for human health and space travel. These molecular devices are designed and manufactured de-novo to perform specific tasks in specific environments - such as individual human bodies.
The initial set of tools may be synthesized from molecules produced by artificial genes inserted into natural organisms. They will be assembled with 10 nanometer-scale electronic tools, brought together with electrostatic fields. This will probably require a vast number of trials before it succeeds, at cryogenic temperatures to minimize thermal agitation. A mind-bogglingly complex task, but at such a small scale that we can replicate the assembly environment a millionfold and perform hundreds of experimental assemblies per second per assembly site until we achieve success. A vast amount of computation will find the most promising paths to assembly.
What we will make, and add to our bodies
HLA-compatible repair robots: The human immune system detects invaders by looking at Human Leukocyte Antigens on cell surfaces. Leukocyte immune cells engulf and destroy objects lacking the right molecular "passport". Every individual human (and animal that we wish to protect) will have their HLA systems completely characterized. Computer models of every individual will be used to create individualized systems of ROCERs, ROCER transports, and ROCER repair cells that can (under external control) be used for various measurement and repair tasks within the body.
Cell garbage removal. Cellular molecular processes make mistakes, and the results of the less common mistakes are molecules that the cell cannot process, some of which accumulate as lipofuscin within cells, or amyloid plaques in brains. While evolved individuals lack the processes to identify and remove these materials, we can learn to do so by artificial means.
Viral removal. We can also identify and remove viruses from the body. Note that children resequenced to a canonical form of DNA 2.0 (along with their helpful gut bacteria) will be completely immune to DNA 1.0 viruses, and will not be interfertile with DNA 1.0 humans, but will otherwise be identical in behavior and biochemistry.
DNA/methylization damage repair: Individual cells store information in DNA, but also in the methylizations that program each particular cell for a particular task. Free radical damage (mostly from oxygen, some from radiation) damage the DNA, which can cripple or age a cell. Given a "canonical" DNA sequence for an individual, plus a table of methylizations for each cell type in a particular location, we can learn to rebuild nuclear DNA. This will eliminate cancer and many degenerative diseases.
Neural repair: These tissues are vastly more difficult than other forms of cell repair, because they are precisely specific. A neuron connects to specific synapses with precise weightings, storing information structurally. If a neuron dies, its structural information dies with it, beyond recovery; adding a new neuron cannot regenerate the information that was stored in the old neuron. With a precise map of each neuron and its synapses, we can replace it, or restore a damaged neuron to health, but a hypothetical repair system must have that map before the neuron is damaged. While some cells in the brain (glia, astrocytes, etc.) can and do regenerate naturally, and some neurons in some portions of some brains (like rat olfactory bulbs) can also replicate, the best evidence is that human cortex neurons are not replaced after development (gestation plus a month or two after birth). Note also that almost all brain cancers are replicating support cells, not neurons; if something can reproduce, it can reproduce incorrectly, so the absence of incorrect reproduction points to an absence of normal reproduction.
This makes sense to me as an engineer. Wires corrode on circuit boards, and those circuit boards can fail. Tossing in a bunch of wires doesn't help, and will probably short out something else, rather than fortuitously bridge only the corroded gaps. Unless dying brain cells contain map information that they can transfer to new brain cells, which can leap into the vacancy created by the dead cell after it is scavenged and removed, there is no plausible way for an added neuron to restore the function of a dead one.
But what nature does not provide, artificial biology (using external maps and a much larger range of tools) may be able to (IANAB). In 2016, our computers are far less capable than molecular biology, but they are already "global"; an individual cell cannot sense what a whole organ or body is doing, beyond the chemical messengers it is programmed to respond to. We can bypass the slow Darwinian process, and add "error handling systems" that individual cells cannot.
If we want our brains to survive and grow beyond a century, or survive harsh environments (like television-advertised diets), we cannot do it with pharmaceutical and dietary cures alone. Neurons die, and are not replaced naturally. We must tinker with the neurons themselves, measuring and characterizing each neuron while it is working, and replace it with an artificial duplicate (biological or biorobotic) when it inevitably fails.
Muscle repair: Muscle cells also do not regenerate, though some rare myosarcomas develop from immature muscle cells that remain after development (note: remove these and store them!). Muscle cells can develop or atrophy, but they are stimulated by neurons, so a muscle cell without a neural synapse attached is useless. Creating new muscle cells with an externally driven plan makes sense; nature does not provide the whole-system knowledge to implement such a plan. When we can rebuild neurons, we can extend them to new muscle cells.
Other tissues: While nature's construction and repair systems are marvelous, they are limited to "cellular local" knowledge, chemical messengers, and for the specific environments encountered during evolution - which do not include television, sedentary living, junk food, alcohol, tobacco, etc. They are not teleological - they do what they do, not what we want. Human systems are not even very good at surviving some conditions encountered in nature; starvation, extreme cold, epidemics.
Biostasis: Cells ingest energy and expel entropy. Homeothermic (warm-blooded) animals require a continuous input of energy and emission of waste to survive; lowering body temperature only 2% (from 37°C to 31°C, from 310 to 304 Kelvin) causes moderate hypothermia, a 6% lowering to 18°C or 291 Kelvin is usually deadly. IANAB, bu5 my guess is that some temperature-excited biomolecular processes have different activation energies and slow down more than others. When processes do not slow in synchronization, cell machinery misbehaves, and the expulsion of entropy fails. Cells accumulate damage and die.
Can specially designed ROCER repair machines continue to operate in a cooled environment, structurally maintaining cells after normal metabolism ceases? This is an important question. If such ROCERs can be made, and supplied with a source of nonbiological energy (such as molecules easily manufactured with solar energy), then a human can be kept in suspended animation without food, and continuously repaired, at temperatures of perhaps 5C, above the freezing point of blood. This can be a low maintenance alternative to cryonics, without differential thermo-mechanical contraction problems (likely a big problem for nanometer synaptic gaps attached to centimeter axons), with liquid water permitting ROCER transport. A great deal of energy will be needed for computation and control, but much of that can be distant and redundant.
Direct Neural Connection: Humans are natural-born cyborgs. Human brains developed in tandem with tool use, and language uses the same left-brain neural pathways. We can reprogram our own neural circuits very adaptively, to do crazy stunts like typing on computers while imagining the web pages they create, and guessing the reactions of those reading them (in this case, I am imagining confusion and annoyed impatience, where's the pictures of buxom celebrities? ).
Humans have very unbalanced I/O; we filter about 20 megabits of sensory input data per second (mostly visual) into a few hundred bits per second of motor response. Brains make decisions, decisions make muscle movements. So far, our decisions and muscle movements have helped us cover a whole planet in vast numbers, and use up irreplaceable resources in a blink of geological/evolutionary time. At present, our decisions are not well adapted to gigayear survival, and typing is only slightly better at providing exercise than poking at a TV remote.
To interact properly with the universe in the long term, we should conciously filter our inputs (elevating environmental awareness over brand name snack foods) and increase our filtered output capability (hundred-bit to megabit output, with some of the stupid reactivity reprocessed into wisdom). Perhaps we can implant muscle tissue, neural growth factors, and imageable platinum microspheres next to the motor cortex, and teach ourselves to control hundreds of virtual limbs rather than the four most of us are born with. This may be the best way to adapt to strange environments, like space. We can turn our addictive fascination with video games into rapid, fascinating, reality-focused training. With the addition of ROCER, we can resculpt the brain to interface better with these systems, or construct new systems that produce electromagnetic rather than chemical and muscular outputs.
Current attempts at human-machine electrical interface use electrodes, zapping neurons with electrical jolts and sensing the weak electrical byproducts of their operations. While neurons move signals down axons as slow, weak electrochemical waves, they signal across synapses with chemical messengers, not electrons, femtojoule chemical pulses rather than the millijoule electrical hammers. When I think about electrical stimulus of neurons, I imagine a prisoner being electrocuted, and emptying bowels and bladder in a horrible death spasm. Yes, we can electrocute a synapse and make it emit neurotransmitter molecules, but we should use biocompatible microbiological systems to generate those neurotransmitter molecules directly. Connecting to those microbiological systems from outside the skull will require sophisticated, repairable, immune-system friendly signal transport, but IANAB, so I will leave it to future biologists to design those systems.
Space settlements as presently envisioned are suburbia wrapped into a radiation-shielded tube, and spun to create artificial gravity. We've tried experiments with independent biospheres. Some folks (who read a lot of science fiction but have not learned to say IANAB ) claim that it "coulda worked" if the designers of Biosphere 2 were as smart as a typical science-fiction reader. In reality, it seems that actual, viable, self-contained biospheres capable of supporting human intelligence are a difficult problem that took nature 14 billion years to solve.
Land animals are not fish in "exowater" suits; we survive on land because we are adapted to it. My guess (IANAB) is that human 1.0 plus artificial biology can be artificially adapted to the high radiation microgravity space environment:
- 1) Microgravity
Microgravity requires bone repair and transport of materials that are normally sorted out by gravity. ROCER?
- 2) Radiation
Radiation damage requires repair. ROCER?
- 3) Vacuum
- Perhaps the most difficult, adding real-time-adaptive mechanical pressure to the skin in ways that mimic 100 kilopascal atmosphere, and supply and support gas interchange to the living cells beneath, while vapor and gas emissions are captured and recycled.
Vacuum tolerance is optional: People can live and work in pressure vessels, and connect to the space environment through robotics. Indeed, microgravity may be beyond the adaptive range of human mobility, so "machine translation" between natural human movement and appropriate movements in zero gee is probably optimum. We will be encased in a robotic shroud; pressurizing that shroud is a minor annoyance, and convective air might be best for cooling the human-connected robotic components. We will connect, not through muscle movements or even myoelectric surface sensors (muscles produce more electrical signal than the neurons that trigger them), but through chemical/synaptic interface to the brain. One human mind will be able to operate vast space systems at the executive level, mediated through countless computers and machines, the way one human mind controls a hundred-trillion-cell body. We will accomplish far more with our minds than with our hands.
But why will people choose to live in space colonies? We are learning to operate space robotic systems through predictive-adaptive telepresence from Earth; in time, space robot operators will become as adept and capable as those controlling robot submersibles. Why will humans go into space?
Perhaps pioneers will engage in risky biological processes not permitted on Earth; first adopters for the technologies described above. If your choice is death or space, you will chose space ... if you can afford to pay tens of millions of dollars per year to do so. The cost of an unmodified human astronaut on ISS is billions per year, and space travel is too dangerous and health damaging the way we do it now. The crossover will occur when space becomes survivable and affordable for wealthy maverick risk takers. Indeed, those risk takers may choose illegal risks to get there; national programs attuned to bureaucratic inertia and public fears will be restrictive, not audacious.
Space colonies may be founded by biological outlaws and renegades, risking all to escape oppression, while protecting natural Earth-life from their dangerous antics. Nobody said evolution is easy.
Space Travel to Other Planets, and Beyond
The current model of a round trip to Mars is an enormous undertaking, perhaps a thousand times as difficult as Apollo to the Moon. This is not just one, but two high speed trips between planets. A Hohmann orbit from Earth to Mars takes 100 times as long as the orbit from Earth to Moon. Worse, slowing from interplanetary velocities (twice!) is much more difficult than reentry from the Moon. We learned about lunar re-entry from the engineers that designed ICBM warhead reentry; they're dead now.
Faster trips involve much higher entry speeds and gee forces; keeping reentry vehicle in the narrow band of altitudes where drag is high enough to prevent a skip into deep interplanetary space or smashing into the dense lower atmosphere requires a circular orbital path at much higher than orbital velocity. That creates high radial gee forces; if the vehicle has a low lift-to-drag ratio, like Apollo capsule, then the high radial gee forces are created by very high tangential gee forces. While heavily modified human beings may be able to withstand those forces better than a normal human with zero-gee induced bone atrophy, it is probably better still to accept a low energy, slow trip, and use biostasis and repair technologies so that the astronauts arrive in perfect health.
The 25 billion dollar Apollo program (in 1969 dollars) inflates to $160 billion in 2016, and produced 300 mission hours on the moon for two people; that's 2.4 trillion dollars per crew-mission-year. The Moon is 390 thousand kilometers away by Hohmann trajectory; Mars is 380 million. Delta V's are significantly higher, and the costs of rockets go up exponentially with delta V. While our electronics and biological skills have increased considerably since 1969, our rockets are less capable than the Saturn V, and produced in much lower quantities than the V-2.
Would you bet on advances in rockets, electronics, or biology?
While some presume we can recreate a living environment on Mars with current technology, so that a one-way trip is not merely an enormously expensive form of suicide, they may not have noticed that current technology is mostly damaging to the living environment of Earth. We will begin by practicing on adapted space colonies, populated with renegades who have modified their personal microbiology to do so cheaply.
Mars, and later the outer planets, will be visited and later occupied by space-adapted humans. In addition to the microgravity and radiation adaptations above, interplanetary travellers will reduce their biological resource consumption through biostasis. They will arrive on Mars to a biosculpted environment designed and tested in orbiting labs and space settlements, and re-enter biostasis if they outrun their resources, or to await delivery of repair parts from Earth.
But there's an important task to complete before we send humans to Mars. Some paleontologists suggest that the chemistry of Mars four billion years ago were better for the formation of early RNA life than the world-spanning deep oceans on Earth of that time. If so, sending DNA-1.0 life to Mars prematurely may destroy any chance of distinguishing rare Martian biofossils from recently introduced contamination.
Before attempting Mars, we should experiment with many forms of artificial life in lunar orbiting labs. We can empirically establish bounding parameters for the forms life may take; if we can absolutely rule out Martian origins in the lab, we can forgo the paleontological quarantine.
If we cannot rule out Martian origins, we should send only DNA-2.0 humans and supporting lifeforms to Mars, so that any genetic contamination can be distinguished from natural biology.
We should also consider these restrictions on the Moon; there is a possibility that large asteroid collisions at the tail end of the late Heavy Bombardment could have sent Earth ocean floor material to the Moon. One of our first "deep explorations" of the Moon may be sifting through the top twenty meters of the lunar surface for microfossils entrained in that ocean floor material; buried under meters of regolith in the cold of space, some early DNA-1.0 biomolecules may still survive there. By confining incoming lunar orbit lab debris to a small reserved area on the Moon's surface, and without winds to move it around, most of the Moon will remain available for this search.
If Earthlike life is vanishingly rare in the galaxy, we will get only one chance to learn our biological heritage. That heritage has been consumed and recycled by plate tectonics and later lifeforms on Earth; preserving the Moon and Mars for whole-planet microbe-level exploration may be our only chance to learn our own history. That history may be crucial to understanding how to look for very rare fossils of extinct protolife elsewhere in the galaxy.