Note: Please use the hashtag #ea0106 when discussing this piece on Twitter. For Facebook discussion, check out and follow the Centre for Science Fiction & Fantasy page.
The generation starship trope has provided a rich seam for science fiction writers. From Heinlein and Aldiss in the Golden Age of SF, the idea of a starship bound on course for generations, with its Noah’s ark of human cargo, has fascinated writers and readers. Yet, recent scientific advances have revealed how difficult it would be to build and sustain such a ship throughout its long voyage. More importantly, when we are eventually able to build it, will we still need it?
What is a generation starship?
Any realistic approach to interstellar travel must confront the vast distances to even the nearest stars. If faster-than-light starships are ruled out, as they must be under Einsteinian physics, journey times become extremely long. If we cannot even approach the speed of light—and our current space probes travel at around 1/10,000th light speed—then journey times become even longer. At 10 percent of light speed (a 1,000-fold improvement on our current performance), a journey to the TRAPPIST-1 system, which has a number of potentially-habitable planets, would take about four centuries. Hence, scientists, some of whom were also SF writers, came up with the idea of a ship in which generations would be born, live, and die, with their successors eventually making planetfall in the new system.
The generation starship immediately raises a host of problems. For one, although the first-generation astronauts would volunteer for their fate, they would condemn their children and their children’s children to that same fate—essentially, lifelong imprisonment in a metal box without hope of reprieve. It is hard to see how that could ever be justified. Even if we cross this ethical hurdle, the ship’s population will face traumatic psychological problems maintaining morale and sense of purpose over the generations, with depression, mutiny, suicide, and breakdown of social order all too likely.
J. D. Bernai wrote an essay on the generation ship called ‘The World, the Flesh & the Devil’ in 1929. The first fully-fledged fictional treatment of a generation ship was likely Don Wilcox’s ‘The Voyage that Lasted 600 Years’ (Amazing, 1940), in which the captain wakes every hundred years to check on progress, and finds increasing social disorder. Terms such as ‘interstellar ark’, ‘world ship’, and ‘colony ship’ entered SF parlance, and writers tackled the sociological problems head-on. Robert Heinlein’s Orphans of the Sky and Brian Aldiss’s Non-Stop are perhaps the classic explorations of the breakdown in social order. Indeed, so many stories exploring the trope were written that, in 1966, Algis Budris wrote, somewhat jadedly,
The slower-than-light interstellar spaceship, pursuing its way through the weary centuries, its crew losing touch with all reality save the interior of the vessel. . .Well, you know the story, and its unhappy downhill round, its exciting struggles between the barbarian tribes which develop in its disparate compartments. . . .
Whether the vein had been exhausted, or science—both technological and sociological—revealed the concept to be too fraught with complications, the trope has been used only sparingly since the 1970s. Gene Wolfe’s four-novel The Book of the Long Sun series of the mid-1990s and Ken MacLeod’s Learning the World: A Novel of First Contact of 2005, provide notable exceptions.
What would a generation starship be like?
The size of a generation ship would be determined by the size of the population required to colonise a foreign planet. From a genetic diversity viewpoint, the number of people needed for a self-sustaining, healthy community is now thought to be relatively small. Given effective genetic screening, some 160 people would be sufficient, even before taking frozen human sperm and eggs, which would be taken along to boost the genetic mix, into account. Genetically, a small village’s worth of people should be enough.
A more severe constraint on the lower boundary for starship size is set by technology’s effects on society. An 18-century village was largely self-sufficient, but a 21st-century village most certainly is not, relying on myriad goods and services from the outside world. The modern economy is extremely specialised. For instance, let’s look at health care. There are at least 20 specialties that one would like to have on board a starship: ear, nose, and throat; heart; obstetrics; eye; paediatrics, etc. And just to have one specialist of each sort would be cutting it fine—suppose the ship’s sole heart surgeon died or were incapacitated? Working through the various domains of the modern economy—manufacturing, transport, energy production and distribution, education, legal, accounting, finance, etc.—it’s easy to come up with several thousand essential roles. Of course, by the time we are able to build a starship, artificial intelligence (AI) may have reshaped our economy (more on this below). Putting that aside for the moment, however, our starship is beginning to look more like a cruise liner than a village.
The world’s largest cruise liner, Harmony of the Seas, has the capacity for some 5,500 passengers with a crew of 2,300—about 8,000 people altogether. This is likely the necessary minimum to populate the various specialisations of the modern economy. The cruise liner format probably also provides the necessary minimum for quantity and variety of living space.
Our cruise liner/starship will also have to rotate. Long periods of weightlessness degrade bones, muscles, and eye tissue. Perhaps genetic engineering will have found a solution by then, but absent that—and absent a magic ‘grav’ button—the centripetal force provided by rotation is a good substitute for true gravity. Rotation has problems of its own, however. If the whole ship rotates, imperfections in the initial spin or in the distribution of mass would lead to a tumbling effect over time. That points to a design with a stable core and a rotating carousel. However, over time, the angular momentum of the carousel would transfer itself to the core. So it would be necessary to have a counter-revolving counterweight. These moving parts would be vulnerable to wearing out over the long years and centuries of the voyage.
Assuming our 22nd-century engineers can overcome these problems, we would be looking at a pencil-like ship’s core, with double ‘doughnut’ carousels rotating in opposite directions around it. This odd structure would have to be assembled in space. How would it fare over the voyage?
The first task would be to accelerate the ship to cruising speed—and there we meet perhaps the most daunting of the problems besetting interstellar travel: the colossal energy budget. A cruise liner is massive (the Harmony weighs 220,000 tonnes). To accelerate a one-tonne mass to 1/10th light speed would take 1/1,000th of current world annual energy consumption. To accelerate the Harmony to that speed would take 220 years of world energy!
Although a space craft would be made of lighter materials than a cruise liner, a couple of factors would tip the scales the other way. One is fuel—the generation ship would have to carry enough fuel to accelerate and, at the other end of its journey, decelerate. The other is cladding. Interstellar space is not empty but rather sporadically populated with dust particles and the occasional larger object. Even a dust particle travelling at relativistic speed would have tremendous destructive force; significantly larger objects would be terminal. Cladding would provide some protection against the smallest particles. Larger objects would have to be detected sufficiently in advance to be destroyed by lasers.
The bulk of the cladding would be ‘on top’ of the space ship, to protect against incoming objects in the direction of travel. However, the whole ship would have to be clad to some extent in order to provide protection against cosmic rays (high-energy particles), which come from every direction, as well as dust, etc., coming in ‘obliquely’. And even with all of this, there is no guarantee that a fatal collision can be avoided.
The only presently-known propulsion system that could provide sufficient thrust to accelerate a multi-100,000-tonne mass to relativistic speed is nuclear fission. The most developed concept to date is Britain’s Project Daedalus, from the 1970s, which featured Helium-3 mined from Jupiter as the propellent.
Renowned SF author and futurist Kim Stanley Robinson has reflected at length on the generation starship concept, here. In addition to the physical problems outlined above, he highlights the biological, ecological, sociological, and psychological dimensions, and—often overlooked—the problem of what do to on arrival at the destination solar system.
In terms of biology, Robinson points out that humans are not distinct creatures but rather symbiotic associations of human and bacterial material. In the island world of the generation ship, bacteria could evolve rapidly in unexpected directions, with perhaps life-threatening consequences. In terms of ecology, it would be extremely difficult or impossible to achieve complete recycling. Some essential chemicals would end up bound to different substrates and be lost to the ship’s population. And, although not mentioned by Robinson, there is also the problem of the dead. Unless the living eat the deceased, a large amount of organic material would be lost via the 13 to 15 generations that will pass away before arrival.
The social and psychological concerns, however, may not be as grave as Robinson and Golden Age SF writers suggest. Sociologically, the regimen on board ship would have to be more or less totalitarian to ensure adherence to the ship’s mission, mitigation of stability-threatening behaviours, and achievement of reproduction goals. Some, or many, individuals would have limited-or-no choice of partners, or, indeed, careers. Yet, like it or not, totalitarian social structures are widespread in human society today—either nationwide (as in China or North Korea), in the disciplined services, or, indeed, on board ships.
Psychologically, the generation ship environment appears at first glance to be stifling, potentially leading to a host of emotional problems. Yet, in reality, it is barely more limited than that experienced by most of mankind prior to industrialisation. Given a choice between a life spent aboard a generation starship and a life spent in a medieval monastery—or even a medieval village—which would we choose? The starship would surely offer a richer, more diverse, and more fulfilled life—certainly a longer, healthier one. Moreover, the availability of AI and virtual reality (VR) would enable starship occupants to expand their horizons beyond the limits of their physical environment.
In sociological and psychological terms, then, life on the starship might not be such a bad deal. But Robinson scores a major point when he draws attention to the problems on arrival.
Far from bringing the problems of the voyage to an end, arrival in the foreign star system just adds new ones. It is highly unlikely that the target planet would have a human-breathable atmosphere; indeed, if it did, it would very likely have its own developed flora and fauna, bacteria and viruses. Intervention in an existent ecosystem would bring all kinds of challenges which the colonists would have to manage with the resources brought with them. They would have to live in sealed habitats on-planet (or, more likely, still on-ship) until they had either genetically engineered themselves to suit the new environment or re-engineered the environment to suit them.
If, as is perhaps more likely, the destination planet is lifeless rock, the challenge becomes one of terraforming. Sources of water, nitrogen, oxygen, and other human-essential chemicals would have to be found, either by mining and processing materials in the planet crust or by steering comets planetward. A planet of Earth-like gravity would be of Earth-like size, i.e. very large. Terraforming might take centuries (as it does in Robinson’s novel 2312).
And when, finally, colonists and planet have adapted to each other, and the community settles down, beginning to build the institutions and infrastructure of a civilised world, do the problems even end then?
Problems of Colonising
Our colony comprises several thousand persons with expertise in every field of human activity. However, their expertise would not run very deep; several thousand people is nowhere near enough to cover the deep specialisms of the modern economy. The services the colonists provide to one another will tend to be of intermediate or non-specialist standard. For example, a heart transplant might be conducted by a general surgeon, yoga class taken by a sports instructor, an algorithm upgrade written by an engineer. , a . Evidence from stressed communities suggests that people can step up and perform wonders. But they can just as easily collapse or make errors that could be fatal to a tight-knit colony.
On the demand side, too, there will be shortfalls. A ‘customer base’ of several thousand is not enough to support even a moderate level of specialisation. The services of some experts might not be called upon for long periods, leaving skills to atrophy. And some services, such as resource extraction, design, and construction—vital on arrival—would not be needed at all during the voyage, and would somehow have to be nurtured through the generations and the centuries.
However, AI and VR could help, allowing engineers to design and build virtual structures en route, even to ‘practice’ building infrastructure for the target planet, while surgeons could get VR practice on uncommon operations. Robot assistants could cover gaps in human service provision, and 3D printing could provide components that today are produced in large manufacturing runs. AI could also teach humans skills that would be needed upon arrival.
AI, then, could be the saviour of the colonists. But before looking more closely at AI, we should first step back and ask what sort of society will we be when we have generation starship capability?
Beyond the Singularity
First, as should be clear from the discussion above, many dimensions of knowledge and technology have to advance a great deal before a generation starship will be possible. And although advances are happening, they are not happening quickly, meaning that the moment of ship launch may be a couple of centuries ahead.
Before even considering travel to another star system, we must become masters of our own, having established permanent successful colonies on Mars, the Moon, and possibly other bodies in our solar system. Terraforming a barren world such as Mars would more or less qualify us as a Kardashevian Type I civilisation (Nikolai Kardashev, a Soviet astronomer, was referring to energy use, a Type I civilisation being able to utilise the energy of a planet, but the concept seems broadly applicable to terraforming).
These goals are being pursued, but slowly and without a definite roadmap. Bigger budgets and more political urgency would speed up projects, but there is no sign that governments of developed countries are going to devote a much larger share of already-squeezed budgets to such endeavours.
Where governments hesitate, the private sector may step in, and indeed has done so in the form of Mars One and Elon Musk’s ventures. But it seems doubtful that the private sector could manage an entire off-world colony, in all its manifold dimensions and with all the security concerns it would raise, entirely by itself.
What is happening—at speed, and with the full force of governmental as well as private sector participation—is the development of AI. The median view of surveyed experts is that Artificial General Intelligence (AGI, i.e. human-level intelligence) is only some 45 years away, with Artificial Superhuman Intelligence (ASI) only a decade or so behind that. Ray Kurzweil predicts the ‘singularity’ (the advent of AGI) by 2030.
AI experts have, of course, been wrong before about the imminence of AGI. We don’t yet even know what AGI would look like. What is different this time is the scale of resources being applied—including by leading private sector firms such as Google and Amazon—and the awesome power of the AI applications being rolled out today. True, these applications are still narrow AI; few people are working on AGI – although some are, and maybe the combinatory algorithms of applications such as AlphaGo Zero and Watson will be capable of more than is presently apparent. Intelligence does not necessarily imply consciousness (whose nature we are also not sure of), but merely the ability to solve complex problems in complex real-world environments.
Suppose Kurzweil et al. are wrong about the date of AGI. Suppose it takes not 12 years, not 45, but 100? Or 200? Or 600? These timescales are all well within that of our putative generation starship mission to TRAPPIST-1. It could easily take a couple of centuries to develop the (non-AI) technology and gather experience of solar system colonisation, four for the starship voyage, and yet more for the terraforming.
AI to the rescue?
I suggest that, long before the six-to-10 centuries are up, we will almost certainly have serviceable AGI, good enough to ‘man’ a starship. It will be relatively easy to design such a ship, since its mass would be insignificant (just the substrate for the AGI circuitry), it could endure super-high acceleration, and the AGI can be sent in multiple expendable copies with no need for lasers-and-cladding. The trip could be much faster than four centuries, since a high proportion of light speed could be achieved for a much smaller energy budget. The round trip could even perhaps be done in less than a century, allowing accelerated progress on other projects yet to be conceived.
That, of course, leaves us humans on Earth, unable to venture so far afield (and, I suspect, even relinquishing our solar system colonies as too unhealthy for human habitation). Yet, in the long life of our species, this would not matter. There may eventually come a day when the AGI, now ASI, god-like, having mastered the unimaginable energies of the central galactic black hole (Kardashev’s benchmark for Type III civilisation), would come and lead us, now less than children, less than ants, into the some portal of unimaginable delights, including the TRAPPIST-1 planets and a cornucopia besides—the generation starship consigned to the Heath-Robinson category of elaborate devices superseded before they were born.That, at least, is my hope.