Wednesday, February 11, 2009

Where To Find Von Neumann Probes - And How They Work

Discussions of von Neumann probes tend, like this one, to be largely enterprises of idle speculation. That can be explained by there being no such career as professional von Neumann probe philosopher. However, if you take the concept at all seriously and you're honest with yourself, reflections on the issue can take on a more unpleasant tone. If you think it might be anything more than science fiction, it's disturbing to contemplate.


"Where Are They?"

The title of this section is the literal manner in which Fermi stated his paradox.[1] Put more explicitly: if the universe contains other intelligences, why do we see no evidence of them? In terms of listening for signals, the absence of evidence we've so far encountered is sometimes referred to as the Great Silence. The same question can be asked for alien artifacts as for signals. The discussion of von Neumann probes, or to be more precise the apparent lack of others' von Neumann probes here, invariably segues into Fermi's Paradox. Given the firmament's apparent sterility so far this is not unreasonable.

There are good arguments to recommend a search for artifacts over a search for signals. Assuming that the aliens are using the same technology that we do (a loaded assumption if ever there was one), we likely wouldn't be able to hear them. If Earth's most powerful radio telescope were to be launched into interstellar space, it would only be able to detect Earth from, at most, two light years away. We would seem silent, even from Alpha Centauri. An even more dubious proposition is that we could tell the difference between an alien transmission and random noise. Distant artifacts, as Dyson proposed, may be easy to miss or misinterpret, but proximate artifacts - in our own star system - are not subject to intensity decay or misinterpretation. Strange though it might be, you have the thing; it's here. If we believe that aliens are likely to leave physical artifacts in our star system, then a failed search for them is actually much more unforgiving (and therefore meaningful) then a failed search for transmissions.

So: why then should we believe, even if we grant that intelligent aliens exist somewhere else in this galaxy, that we should find alien artifacts here in this star system? The argument can be summarized as follows.


1) Based on one case, we know that life is possible. While we only have one sample, to argue that elsewhere life occurs not at all or only very infrequently assumes special conditions for Earth and our star system. This not only flies in the face of Ockam's Razor and the principle of mediocrity, but seems positively pre-Copernican in outlook. If life can occur in at least one place, then in the absence of other data, the simplest assumption is that it occurs frequently.


2) The same argument can be made for intelligence.


3) If other intelligences exist, given the age and size of the galaxy and the sun's position within it, there is no reason to assume that we are likely to be the first intelligence.


4) Given the age and size of the galaxy and the relative speed with which von Neumann probes could spread, if they are feasible, they have almost certainly been here already.


And yet, to put it mildly, it is not obvious that there are von Neumann probes here.

A search for self-replicating artifacts has another strength over a search for signals. Self-replicating artifacts will likely outlast their creators, meaning that all of the depressing attrition factors that the Drake equation[2] introduces as candidate explanations for the Great Silence would have to be very close to unity to explain the lack of artifacts. These candidate explanations (whether they are formally in the original Drake equation or not) include: that intelligent aliens avoid detection; that intelligent aliens kill themselves once they get smart enough to build von Neumann probes and send signals; that intelligent aliens are superpredators and they kill each other; or some combination thereof. Where von Neumann probes are concerned, none of these matter unless 100% of the time they take effect before the invention of von Neumann probes. It would only take one - just one - species to escape these attrition factors just long enough for a single successful make and model of von Neumann probe to propagate itself throughout the galaxy. Then even that species can push The Button on itself, while the rest of them can have all the ecosystem catastrophes or interstellar predation events they want. It just takes one.


Question Your Assumptions. What Are We Looking For?

The interesting character Terrence McKenna has argued that the various SETI efforts are all totally misguided as a result of being hopelessly cluttered with parochial assumptions which are peculiar to our own experience, and which we may not be aware of, or at least may not be able to help making in order to begin a search. McKenna's example is that concluding there must be no aliens because we can detect no interstellar electromagnetic signals is akin to concluding there are no aliens because we've detected no interstellar Italian restaurants. XKCD makes a similar comment.

Indeed it's disconcerting to think of the difficulty even members of the same species have had in understanding each other. Supposedly on one of the Pacific islands that Cook visited, the people insisted they couldn't see the ship he'd come from, even though it was right out on the water. Of course they'd never seen anything like it, and even though he was pointing right at it, maybe they thought it was a far off cloud; it couldn't be a canoe, because no canoe is that big. Granted, these anecdotes are not helpful in providing direction to the search but they serve to remind us of one of very few certainties, our own provinciality in a universe that, in Haldane's words, is surely queerer than we can imagine. The Romans told stories of central African men with faces in the middle of their chests and colonial Spaniards whispered about two-legged curly-tailed dragons in Patagonia, and I have no doubt that both of those fantasies were closer to the truth of Africa and Argentina than are most of our current ideas to what we will eventually encounter out there. Even writers whose imaginings are untethered from engineering and budget realities typically bring us a von Neumann probe befitting the late Iron or early Information Age, looking like a huge menacing slab of iron like a kind of automated space Yamato (as in Saberhagen's Berserker) or smaller, vaguely crab-like things (as in Bear's The Forge of God). These are great works of fiction, but the reality is not likely to resemble these works, which are after all the products of current human expectations. Even a century from now the huge metal ships of our speculations may look silly: we're already producing organic LEDs commercially and self-assembling organic circuit boards.

Having just preached about the unanticipateable strangeness of the universe, it will seem strange for me to now make an argument on where we should look. No doubt due to my own training, my bias is toward chemical replicators. But there's value in recognizing the assumptions we make, and anyway we have to start somewhere.


Probable Characteristics of Von Neumann Probes

Small and Numerous

If the probes can reproduce robustly, they would be much more powerful if they could work in concert. Losing one or a few to mishaps in an alien Kuiper Belt wouldn't be so tragic as having a chance impact ruin a thousand-year mission in its eighth century. Consequently the individual probes don't have to be huge our complicated (if we're assuming the active probe exists as an independent object; more on this later). The idea of many small but highly networked probes was explored coherently in 1989 in the solar system exploration proposal of Brooks and Flynn.[3] If these are robustly reproducing probes, then individual cells in a bacterial mat may be a better analogy than fully independent (animal-like) units.


Assembly in Zero Gravity

Probes assembled in zero gravity would have far more flexibility in terms of design and capabilities. NASA engineers describe space vehicles as "fuel, plus what you absolutely need to include as instrumentation". Escaping gravity wells is incredibly expensive. Think of it this way: with nearly seven billion humans and a full economy, we can between us manage to get maybe one object a month into orbit. Beginning the process outside a gravity well eliminates this major constraint.


Assembled from Abundant Materials

Von Neumann probes, if they are to avoid gravity wells, must be able to build themselves from materials that can commonly be found in low-gravity environments. We're learning that carbon and other organics are more common on asteroids and comets than we might have believed before. We commonly think of von Neumann probes as metal ships, but water ice is very hard at low temperatures, and it's easier to work. Yes, it melts, but if you can always make more probes, you don't mind sacrificing a few on a swing through the inner system that you're visiting.


The most tenable assumption I made in describing the characteristics of a von Neumann probe above is that an interstellar replicator would save energy by staying out of gravity fields - why would they need to "land" - and this is why the asteroid belt is frequently offered as the ideal place to look for them. It's a microgravity environment with nickel, iron and even iridium for building materials, as well as organic molecules. (At this writing, we'll be waiting 2.5 more years for the Dawn Mission[4] to arrive at its first asteroid.)

If gravity is the reason that we should look in the asteroid belt, there is one more place in our own star system where we might have even better luck finding them. The biggest gravity well here is the sun, so the further away you can stay, the better. Are there objects far from the sun with water, organics, and some transition metals?


Can We Really Say With Confidence That They're Not Here?

In 2002 biologists from New York's American Natural History Museum discovered Nannarrup hoffmani, a whole new animal species (and whole new genus) of centipedes. These days discovering a new animal (as opposed to a bacterium) is a fairly big deal. Even more interesting, this particular new animal species (and genus) was discovered in Manhattan's Central Park.

By arguing as I have that it would take just one successful ancestor probe to populate the galaxy when in fact we've found none, I might seem to weaken any argument for the possibility of alien intelligence. Frank Tipler makes exactly this argument: since there are no von Neumann probes in the solar system, then we can confidently say there's no other life in the galaxy. I disagree with Tipler's conclusion. We cannot conclusively say there are no von Neumann probes in the solar system, because we've barely begun to explore the solar system. I will agree with his statement, once we have a) proven than #4 above is feasible, i.e. we have our own proof-of-concept von Neumann probe, and b) once we have reasonably good knowledge of our own backyard; that is to say, a similar level of detail to what we now know about Earth's physical environment. At that point, upon having found nothing, we should conclude as Tipler has prematurely that it's most likely we're a freak occurrence, and we're alone. I would put money on our finding something. I also recognize that it's unlikely we'll be able to make those statements within my lifetime. The point is that it's a little presumptuous at this point to expect that we would already have found any von Neumann needles in a 100-AU-wide haystack, when we're still finding new species in Central Park.


Designed Replicators Will Always Become Selfish, Eventually

Richard Dawkins has boiled down the idea of life to "the nonrandom selection of randomly varying replicators". This principle is substrate-independent. Consequently, no matter what you build your von Neumann probes out of and how well you design them, the Second Law dictates that there will be mutation, and there will be random variation. If that variation is heritable, it can be selected. At this point it's worth reminding ourselves what the purpose of an apple tree is - to make more apple trees; or, even more accurately, to make more apple tree DNA. What are the possible purposes of von Neumann probes? Three likely purposes of their designers are to send information back home about the worlds they visit; to ready those worlds for colonization; or to eliminate threats. But all it will take is one - just one - probe to stop wasting its time sending back pictures of gas giants, and devoting more energy to reproducing itself, to begin the process of becoming a "selfish" replicator; that is, one whose design has become focused more on propagation than on the mission its designers wanted it to carry out. All the programming tricks in the world will not hold back these shifts forever, which is how long your probes will be out there.

For any replicator trying to make more of itself, the phenotype material (in our case, protein) only matters insofar as it protects and spreads the genotype material (in our case, DNA). The information about how to make the replicator is more important than the replicator itself, which is really just a temporary shell. What's the point of an apple tree again?


Credit XKCD.

Right now many readers are thinking of neat tricks to stop selfish von Neumann probes, like having the other probes zap them if they get out of line, or counters that make the probe self-destruct after a certain number of propagations. Interestingly enough, the same game is played out in your own tissue in a kind of somatic natural selection that, given enough time, will always end up producing cancer. For us as large and complex organisms, only from one tissue does DNA get into the next generation - the sex cells. To accomplish this act of coordination your cells must cooperate; it doesn't do your liver cells much good in the long run to act selfishly and try to outreproduce your gametes, since they can't go anywhere and ultimately it will kill the whole organism. They become a parasite that can't leave the host, eventually killing it. This is called cancer. Consequently as you might expect, after hundreds of millions of years of living as multicellular organisms, nature has given us level upon level of controls to keep this from happening, including counters programming the cells to self-destruct after a certain number of propagations, and cells that zap each other if they misbehave. But if you live long enough, one cell somewhere, sometime is going to have just the right mutations to escape those checks and balances, and start growing out of control. The more times you roll the dice, the greater the chance that eventually you'll come up snake-eyes. It just takes one.

Given enough time, all von Neumann probes will drift from whatever task their designers built them for, and become self-interested replicators, assuming they're free to act separately to some degree (like bacteria, rather than an animal's somatic cells). The probes that focus on their own reproduction above all else are the ones you would expect to see more of as time passes. After a long time, they're the only ones you would expect to see.


Comets or Asteroids: Which Are a Better Home For Replicators?

Above: Comet Hyakutake viewed from the SOHO sun observer satellite, starting at about 12 seconds in.
Below: Wild 2, the target of the Stardust Mission, with photo enhanced to see surface jets.



Why might comets be better replicator hosts than asteroids? Space probes have interacted with comets Halley, Hyakutake, Borrelly, Tempel-1, and Wild 2, and interactions with two more comets are planned for 2010 and 2014. Asteroids and comets seem increasingly more similar than previously thought, and it looks as though a comet is just an asteroid that since the birth of the solar system has resided mostly far from the sun, where its ices are not boiled away by sunlight, or boiled away only gradually during comets' rapid dives into and out of the inner system. As a result, comets are silicon-iron rocks covered with water and hydrocarbon ices.

Asteroids do contain organic compounds - most spectacularly the Murchison meteorite, which contains not just (racemic) amino acids but nucleobases (like uracil).[5] While the Dawn Mission will tell us more about the material on the surface of a pre-burn "pristine" asteroid, we already know that the comets we've checked up close have significant amounts of hydrocarbons on the outside, much of it in the form of soot. Counterintuitively, the nuclei of comets (as opposed to their tails) are among the darkest objects known in the solar system (Halley held this record with a 0.04 albedo until it was surpassed by Borrelly with 0.03), consistent with carbon. Some comets are green to the naked eye as a result of diatomic carbon, as with Hyakutake or Lulin, 11 days from its closest approach to Earth at the time of this writing. The European probe Giotto found in 1986 that Halley was ejecting methane and ammonia, along with other trace hydrocarbons, and the dust being ejected was of two types, either mineral or C-H-O-N. On the Deep Impact mission, Tempel-1 was shown to contain ethane.

Although the infrared absorption spectra taken during Halley's last visit did show simple hydrocarbon absorptions, similar spectra
could be reproduced in the laboratory using solid-phase (methane and water ice) synthesis, reproducing the conditions in the frozen-solid outer solar system[6]. Similar results were found for Hale-Bopp (which showed most interestingly that there were no known gas-phase syntheses for some compounds observed).[7] So far, the organic chemistry of comets has been explained with gas- or solid-phase chemistry,[8][9] but it's interesting that the hard impact on Tempel-1 strongly suggested this comet contains clays and carbonates, interesting since these materials typically require liquid water to form.[10]

Comets are both electromagnetically louder and larger than originally thought. Hyakutake was the first comet observed to emit X-rays[11], to the surprise of those who initially observed it - though other comets have now been observed to give off X-rays as a result of the solar wind interacting with the coma. As a result of Hyakutake we have a good idea of the size of comet tails - in 1996 Ulysses unexpectedly passed through its tail, 500,000,000 kilometers away from the nucleus.[12]

Not surprisingly, the actual returned material from Wild 2 from the Stardust Mission, while still being analyzed, has yielded the most interesting results to date. Particles captured from Wild 2 have less carbon than carbonaceous meteorites, but more oxygen and nitrogen; there is also the possibility of experimental artifact, since the epoxy material used to capture the dust may have formed polycyclic aromatics upon reentry heating. That does little to diminish the interest in the returned material, however: the oxygen and particularly nitrogen isotopic ratios suggest an interstellar origin for some of the dust, and the molecules analyzed so far include quinolone, at least one amino acid (glycine), and just about every functional group found in organic molecules on Earth,[13] up to and including (probably) sugars[14]. No analyses of the Wild 2 materials' chirality has yet been published.

Most discussions about von Neumann probes assume they will use these components (above). Von Neumann probes are more likely to be composed of the one set of materials and components that we already know can be used to make viable replicators. They have the benefit of being easier to work and more abundant in the universe as well (organic compounds synthesized in the solid-phase, below).


Why is carbon so much more interesting than metals when we're searching for alien artifacts? Forgive my carbon chauvinism, but reasoning from the one case of matter-organizing replicators we know of - life on Earth - replicators are based on the chemistry of light multivalent atoms dissolved in liquid. From what little we do know, as materials, they're cheaper, more flexible, and more efficient. The complexity-per-volume and kinetics of these kinds of compounds are far more efficient per unit mass than anything we know about so far with respect to transition metals. Here's a back-of-the-envelope for you: for the same energy it would take to copy all of the DNA in an adult human, you could melt just about 7 kilograms of nickel[15] (that's in the balmy tropical region at the sunward edge of the asteroid belt where it's about 200 K).[16] In terms of information per unit mass, with our current technology, a human cell massing 10^-12 grams and containing 10^11 base pairs contains 5 x 10^15 times more information per unit mass than a 2 GB memory stick weighing one ounce. If you're building a von Neumann probe, the choice of materials seems clear. Certainly in the very recent history of our planet, metals have been purified by one species as a tool material useful to large replicators - but let's not let our momentary late Iron Age surroundings and narratives convince us that we should be looking for great metal Berserker ships.

To be clear, nothing we have observed so far should even make us suspicious that we've seen anything but geologic and chemical processes. One problem is that if there is some alien biochemical network happening, then our sample of the disordered material blown off the comet by the solar wind would be less likely to exhibit the functioning in situ pattern.

While researching this article, I called Scott Sandford, the Co-head of the NASA Ames Astrochemistry Laboratory, and asked him if there were algae blowing off Wild-2, would the tests run on the Wild 2 returns have detected something? Sandford said yes, that the analyses would have quickly detected the huge molecules involved in life as we know it on Earth[9]. That said, there's only one other place in the solar system (outside comets and asteroids) with such a rich inventory of organic compounds, and it's Earth. If the Mars Rovers found these same compounds on Mars, it's hard to imagine that we wouldn't be incredibly excited, and that our reaction to finding the same compounds on a comet is tempered by our expectation that a comet is an icy, chemical-ridden dirt ball, but cannot be a home for any kind of replicator chemistry.


Comets and The Origin of Life

The idea that extraterrestrial bodies may have acted as life-seeding agents for Earth is an old one. Panspermia was discussed in its post-Enlightenment form most famously by the Swedish chemist Arrhenius, and Francis Crick has wondered whether life was deliberately seeded in a process he called directed panspermia. Comets have been a prime candidate in these discussions, due to their age and their wide range. It's even been argued that comets deposited most of the water on Earth during its early history, although this theory is losing support as a result of the Wild 2 returns showing different deuterium ratios than Earth's oceans.

Now that we're seeing seemingly biogenic methane emissions on Mars, there is increasing support for what can be called oligospermia, or cross-fertilization between planets in the same star system like Earth and Mars. There are fragments of Martian rock on Earth, deposited here after Martian impact ejecta achieved escape velocity and drifted until it crossed Earth's path in space. This raises the question of whether, given enough time, we could be talking about the possibility of not just interplanetary but interstellar exchange of biological material by natural processes, investigated in the 1978 book Lifecloud by Fred Hoyle and Chandra Wickramasinghe.[17] Certainly some of the Wild 2 particles show nitrogen-15 ratios strongly suggesting an extrasolar origin.[13]

In the case of Mars, I personally prefer that its life not be related to our own, because we'll learn more from novel replicator chemistry rather than from just another relative diverged from a LUCA closer to the base of the tree. So far in comets we haven't found any complex molecule appearing in excess of what we would expect - an early telltale would be any such molecule that exhibits selective chirality. Because comets leave such long tails - it would be difficult to argue that comet material did not salt the young Earth. Establishing possible paths between amino acids and ribonucleotides and any over-represented molecules on comets could lead to a whole new series of Miller-Urey experiments.


How Comet Panspermia Might Work

What I am suggesting as an interstellar replicator is the concept of a comet virus; that is, autocatalytic molecules encountering the compounds on a comet, and organizing them in solid phase chemistry to make more of themselves. However, the resemblance of this concept to viruses as we think of them should not be overstated. Assuming a "virgin" star system, the infectious materials will be encountering a dead comet that is loaded with useful, but disorganized, simple compounds. In contrast, in biology, cells are adhered to by viruses highly specialized to take advantage of an already well-organized chemistry. There are cases of self-assembling virus coat proteins, like tobacco mosaic virus, but this is one step in a larger complex process. If there is a replicator in biology at all analogous to comet virus, it is either a viroid (a naked RNA molecule which enters plant cells and diffuses passively through the plant). Moving into engineered molecules, probably even better analogs that we know of so far are the few cases of engineered proteins that reproduce themselves[18], but this example is still far short of a chemical universal constructor.

The life cycle of a comet virus (or a von Neumann comet) might go like this: an "infected" comet in a neighboring star system falls toward its star and happens to be accelerated to stellar escape velocity in the process, getting shot right out of the system. This is not implausible - comets achieving stellar escape velocity (eccentricies > 1, meaning a hyperbolic trajectory) is not only possible, it's not even unusual - it's happening to Lulin[19], which is visible with binoculars as I write. It will take a long time for the comet to get to the next star - Lulin would take 67 million years if it's aiming right at Alpha Centauri - but what does "long" mean when this galaxy has been here for at least 6.5 billion years? The infected comet passes through the Kuiper Belt and begins falling toward the sun, accomplishing this with little or no need for fuel. (If this helpless nonmotility seems unacceptable for a replicator, keep in mind that in our own ecosystem, few bacteria can move on their own, and hardly any plants and not a single virus.) As the comet falls toward the sun, some of its ices vaporize, some of them melt in deep cracks, and metabolism accelerates. Any effective replicator makes an excess of the molecules that begin the whole chemical cascade - its genes - along with other catalytic molecules which we might call enzymes. It is with these that it "does things". While we're speculating, we can list things like: fashioning lenses out of the silicon minerals in the comet-rocks (primitive? trilobite eyes were merely thin layers of quartz); maybe it stores what it observes in chains of polymers it leaves behind for diffusion back across space; maybe it uses the comets' X-ray emissions to communicate to the network of other infected comets spread out like repeaters for light years behind it; maybe it can even steer using the multiple jets observed on comets, all while falling "for free" toward the inner system where, if it is really a von Neumann probe, it can gather information - if these are still behaving von Neumann probes, and not cancerous (selfishly propagating irrespective of intended mission).

As the infected comet falls toward the sun, chains of its template molecules are blown off into space like sperm or loose viruses, leaving a diffuse 5x10^16 cubic kilometer wake of infectious material - diffuse, but huge. Even if the comet's volatiles are completely blown off the comet, they're now occupying a huge volume of space, and blowing outward with the solar wind toward the system's Kuiper Belt. Sacrificing one "carrier" is not a loss if it means effectively spreading template molecules. Some of these will inevitably sprinkle down onto planets; some of these will inevitably be hanging there when another comet swings by. Landing on the partly melted surface of a virgin comet, the template molecules sprayed off in excess by the first comet begin the process all over again.



This method of diffusion of von Neumann material may seem very passive, but works quite well on Earth, as long as an excess of hereditary material is produced.
Above: an animation of the viral lysogenic cycle, where a virus with no independent metabolism drifts passively until it adsorbs onto a cell membrane. The comet virus lifecycle I propose does not require destruction of the comet as part of the dispersal mechanism. Below: this may be a better analogy for the hypothetical comet virus lifecycle because dispersal does not require destruction of the source.
If the hypothetical comet virus lifecycle above seems overly passive, it should be emphasized that on Earth, many niches emphasize fecundity over longevity (template dispersal over template container lifespan). If source materials are cheap (in this case, cells and soil), this is a good strategy.





The Epidemiology of Von Neumann Comets: A Fermi Problem

Fermi was well-known for tackling problems with very little data, but with a clear conceptual framework of how it would work, and arriving at generally sensible order-of-magnitude answers. Consulting firms often throw these questions at interviewees to see how they think through problems. The Drake equation is one example, though McKinsey probably doesn't use it much. Using the same approach - essentially, to get an order of magnitude idea - I think we can come up with a general picture of whether this kind of process could conceivably spread comet-borne replicators across the galaxy in the time period that the galactic disk has existed, or if the comet viruses would still be languishing in or near the star system where they were designed or born.

The estimates I have seen so far for the spread of von Neumann probes (like Tipler's)[20] seem to be based on a simple radial expansion: the probes will move at about their maximum velocity outward from their point of origin (neglecting periods for acceleration and the time it takes to reproduce). The rate-limiting step in the diffusion of comet viruses would still be the interstellar transit time, not the infection time.

Assume the following for interstellar diffusion: a hyperbolic comet like Lulin would take 67 million years to get to Alpha Centauri. Let's guess 200 million years for the next time it's captured as it skims the Kuiper Belt of a nearby star 5 LY away.

Assume the following for the infection process: Halley's Comet is calculated to have lost 2.8 x 10^11 kg during its 1910 passage.[21] Halley's has a size of about 15 x 8 x 8 km; most comets are smaller, so let's assume our infected comet is a circular one with radius 1.25 km. Using volume to adjust proportionally for mass lost, our comet loses 2.4 x 10^9 kg with each pass.

Assume that only 0.1% of that material is infectious, or 2.4 x 10^6 kg. Assume that the infectious particles are the mass of the DNA in a human chromosome (1.22 x 10-15 kg). That means about 2 x 10^15 particles will be shed each time. These molecules will diffuse through the star system (away from the sun, but assume they're evenly spread). Taking the radius of a star system as 100 AU (Pluto is about 50 at apihelion), we're talking about a volume of diffusion of 1.41 x 10^31 cubic kilometers. There will be only one molecule every 7 x 10^15 cubic kilometers.

Now, along come our periodic virgin comets, with average orbit time 200 years. Say that each comet has (for simplicity) a circular orbit of radius 100 AU, and is the same size (but sweeps out a path along that orbit as a cylinder). This means that each comet sweeps out a path of 6.65 x 10^11 cubic kilometers; it only hits a particle on average every 10,000 orbits. Assume the particles are only 10% efficient at infecting the comets, so it would only be infected every 100,000 orbits. For a 200-year period comet, that's 20 million years! But It's not the only comet in the system; let's say there are a hundred over the course of a year, so there's an infection every 200,000 years, and then the amount of particles goes up.

Assuming there's any meaning in talk of replicator chemistry in a supercold solid phase, the rate-limiting step is still transit time (three orders of magnitude greater than infection time). At that rate, based on a spread of the galaxy's age of 6.5 to 13 billion years, a comet virus could have spread between 150 and 300 LY from home. The galaxy is 100,000 LY across. This suggests that if interstellar replicators rely on passive diffusion by comet, then not finding them in our own solar system would not mean they don't exist. A passive diffusion mechanism would give rise to pockets of replicators, rather than von Neumann tsunami that Frank Tipler expects; finding none would mean that we're just down a galactic side alley.

Let's do a brief comparison with current ion thruster technology. The NASA workhorse ion thruster is the NSTAR. Imagine a small 50 kg probe, including fuel and 4 thrusters. Erosion of grid material is a problem in ion thrusters, but the NSTAR has been fired continuously for 3.2 years and not failed. If this combination fires for 1.6 years, coasts until it has to decelerate, and then fires for an additional 1.6 years, it would take 3,475 years to get to Alpha Centauri. Not so great. If we give our NSTARs more credit and fire them continuously for the first 2.15 LY of the crossing then turn and start slowing down (reaching a top speed of 0.029c), it would take 150 years. One of the top-rated thrusters is the VASIMR, being tested by Ad Astra; I don't have failure data, but it has a nice top thrust of 88.5 Newtons. Firing one of these for the first 2.15 LY and then turning to slow, we achieve a top speed of 0.21c and get there in 41 years. Not bad. No doubt we'd lose a lot of them in the trip at that speed, but if we had an automatic factory turning them out - and they could build another one when they got there - this would be doable. If we find none of these, it means they either aren't viable, or we're the first.


Could We Tell The Difference Between Cancerous Von Neumann Probes and Dumb Replicators?


Some readers will object that I began with a discussion of von Neumann probes, and modulated to a discussion of panspermia. The distinction is whether we're talking about designed tools that use self-propagation to carry out functions intended by an intelligence ("behaving" von Neumann probes) and either "dumb" naturally-evolved replicators (space algae) or "cancerous" or selfish von Neumann probes that have long since abandoned their intended function and have out-reproduced their higher-fidelity cousins. If our probes to asteroids and comets find von Neumann probes, we will probably find selfish ones.

Whether we can tell the difference between selfish von Neumann probes and space algae hinges once again on our ability to discriminate intention from noise. Frustratingly, this takes us back to the initial problem that led us to choose a search for artifacts over a search for signals: can we distinguish alien signal from noise? We can safely assume the answer will be closer to no than for the same question with a human artifact. That said, if I gave you a hand-held computer with embedded (dedicated) software that was programmed entirely in Arabic, assuming you don't read Arabic, could you tell me what it was for? This is why it's not clear that we could tell the difference between an artifact that conforms to an alien's intentions and one that doesn't. We'll have to see a lot more interesting chemistry or complex micro-scale structure in comet samples before it's worth losing sleep over these questions, but the idea of interplanetary seeding is getting increasingly hard to call outright impossible.


AN UNSATISFYING CONCLUSION

People laugh at Star Trek, but if we're honest with ourselves, what we want to find out there are humanoid ridged-forehead aliens, or something we can communicate with, or at least their cool computers. What I've proposed is that even if we do find something, it won't be the aliens - who in any event will be utterly incomprehensible to us - but a fragment of their technology; and furthermore that fragment, by the time it gets to us, is likely to have mutated in such a way that we won't be able to discern whether it's the product of non-human intelligence somewhere else in the universe, or a trick of interstellar chemistry. In fact the comet-panspermia hypothesis, which in effect states that we may be the indirect distant descendants of ancient von Neumann probes or space algae, is oddly the most unsatisfying of all.

I originally began this article as an argument for building our own von Neumann probes now. And we should; in the next few years we will likely be discovering a host of Earth-like worlds. It is quickly becoming apparent that main sequence stars with planets are the rule in our stellar neighborhood rather than the exception. Current detection methods naturally bias our current discoveries to gas giants cooking in the uncomfortable proximity of their parent stars, but the Terrestrial Planet Finder (when it's finally launched) may change that. It will be very surprising indeed if our star system is somehow special, and if we don't start finding Earth-like planets orbiting nearby sun-like stars.

A not-wholly-inappropriate response is "So what?" Even with good instruments and powerful computers there's only so much you can learn without a closer approach. There's no reason to think that we'll have the technology to send people to those places, or even to colonize Mars, in the next few centuries. It's worth remembering that in total, so far, we've achieve about forty landings on other bodies in the solar system (including crashes) and that only a handful of those were manned. But the challenges of space travel are biological ones, not engineering ones, and that's why we should be thinking about automated interstellar missions.


NEXT STEPS

Humans should be designing and launching von Neumann probes in the next two generations. To ignore the costs and politics of such an endeavor is to guarantee that it won't happen. Any support for such a program from the private sector will come only from businesses that expect that the technology developed in the design of self-assembling craft will be profitable. Consideration of support from the public sector (which supported all space travel until this decade) must raise questions of political will: any interstellar exploration program has no prospect of returns during the presidential administration during which it's first funded, and realistically, only a marginal prospect of returns while the funding country still exists at all. But there is at least one positive in the mismatch between the project's returns and the time scale of human experience, which is that it allows us to prioritize. We can focus without distraction on the best, rather than the closest, Earth-like candidates; if the difference is getting data in fourteen centuries instead of twelve, who cares?

In the meantime, we should be looking at incentives and near-term profitable technologies. The RepRap project is probably the closest to realizing von Neumann's concepts in a commercially viable way. We can accelerate the process byputting forward incentives for self-replicating technology tomorrow, as with the proposed Mean Green von Neumann X-Prize:



Looking at the RepRap devices, suddenly it starts to seem more real. If the vehicles can accelerate to 0.01 c (a velocity you can reach by accelerating at g for an hour and thirty-three minutes), our probes could reach the edge of the galaxy in 1.6 million years, and fill the entire galaxy in 8.3 million years. That's a long time compared to a single life but in geologic time it's not very long at all. It's pretty amazing to think that we'll conceivably be able to start this within one or two generations. Granted, that's a strong acceleration for an interstellar craft. Frank Tipler's less ambitious 1980 estimate (with 1980 propulsion technology) was that it would take 300 million years to fill the galaxy.[20]


What Will We Find?

Ideas that other intelligences are seeking to uplift us or invite us to some grand galactic congress or exchange of ideas are hopelessly naive. The record of how humans treat each other should be sobering when we consider that these are beings that are related; how will animals from different biospheres react toward each other? To this end, at the very least we should stop deliberately announcing our presence. If we take the idea of alien intelligence seriously enough to send messages, we should take it seriously enough to stop immediately. There is no reason to assume that the rest of the universe is any friendlier than the small piece of it we've seen so far.

There is a nonzero possibility that we won't always be the only intelligence but that we are, somehow, the first. Even if life and intelligence are common, someone has to be the first one. I recognize that natural selection is not a moral phenomenon, but my mammalian midbrain still insists on some level that if we are indeed lucky enough to be first and we waste the opportunity, we deserve whatever fate has in store when the second intelligence appears and starts to spread.


ACKNOWLEDGEMENTS

Thanks to Scott Sandford for taking the time to discuss these ideas with me.


REFERENCES

[1] Los Alamos Technical report LA-10311-MS, March, 1985.

[2] F. Drake. The E.T. Equation, Recalculated. (Wired, Issue 12.12, December 2004).

[3] R. Brooks and A. Flynn. Fast, cheap, and out of control: a robot invasion of the solar system. Journal of The British Interplanetary Society, Vol. 42, pp 478-485, 1989.

[4] The Dawn Mission.

[5] Z. Martins et al. Extraterrestrial nucleobases in the Murchison meteorite. Earth and Planetary Science Letters, Volume 270, Issues 1-2, 15 June 2008, Pages 130-136.

[6] C. Chyba and C. Sagan. Infrared emission by organic grains in the coma of comet Halley. Nature 330, 350 - 353, 26 November 1987.

[7] S. D. Rodgers and S. B. Charnley. Organic synthesis in the coma of Comet Hale-Bopp? Monthly Notices of the Royal Astronomical Society, Volume 320, Number 4, February 2001, pp. 61-64(4).

[8] S. B. Charnley et al. Biomolecules in the interstellar medium and in comets. Advances in Space Research, Volume 30, Issue 6, 2002, Pages 1419-1431.

[9] Scott Sandford, verbal communication.

[10] C. Lisse et al. Spitzer Spectral Observations of the Deep Impact Ejecta. Science 4 August 2006: Vol. 313. no. 5787, pp. 635 - 640.

[11] C. Lisse et al. Discovery of X-ray and Extreme Ultraviolet Emission from Comet C/Hyakutake 1996 B2. Science 11 October 1996: Vol. 274. no. 5285, pp. 205 - 209.

[12] G. Jones et al. Identification of comet Hyakutake's extremely long ion tail from magnetic field signatures. Nature 404, 574-576 (6 April 2000).

[13] S. Sandford et al. Organics Captured from Comet 81P/Wild 2 by the Stardust Spacecraft. Science 15 December 2006: Vol. 314. no. 5806, pp. 1720 - 1724.

[14] G. Cody et al. Quantitative Organic and Light Element analysis of Comet 81P/Wild 2 particles using C-, N-, and O- µ-XANES. Meteor. Planet. Sci., in press.

[15] C. Minetti. The thermodynamics of template-directed DNA synthesis: Base insertion and extension enthalpies. PNAS December 9, 2003 vol. 100 no. 25 14719-14724.

[16] F. J. Low et al. Infrared cirrus - New components of the extended infrared emission. The Astrophysical Journal, volume 278, part 2 (1984), page L19

[17] F. Hoyle and Chandra Wickramasinghe. Lifecloud : the origin of life in the universe. London: Dent, 1978

[18] R. Issac. Approaching exponential growth with a peptide self-replicator and studies in the dimerization-inhibition of transcription factors E47 and Jun. Doctoral Dissertation, Purdue University, 2002.

[19] Comet Lulin data at JPL Small-Body Database.

[20] F. Tipler. Extraterrestrial intelligent beings do not exist. Quarterly Journal of the Royal Astronomical Society (1980). 21, 267-281.

[21] D. Hughes. The size, mass, mass loss and age of Halley's comet. Royal Astronomical Society, Monthly Notices (ISSN 0035-8711), vol. 213, March 1, 1985, p. 103-109.