Cosmic Anarchy and Its Consequences
Executive Summary: This post aims to contribute towards a better understanding of the cosmic political environment. Specifically, it focuses on latency as a key historical constraint to the size and intensity of governments and shows that at lightspeed the distances between stars are too vast to be practical for deferring decision-power to a central body. This result appears to be quite robust to potential upward corrections in the cosmic speed limit or the extension of governance to digital subjects. Subsequently, sovereign political organizations are highly unlikely to ever control more than one star system and a galactic government is outright impossible. This translates into a potential of more than 100 billion independent unistellar civilizations in the Milky Way and means that the primary political ordering principle in the cosmos is anarchy.
The primary characteristic of anarchy is the lack of an enforcer of community rules, so that civilizations need to rely on self-help and are unable to solve collective action problems amongst each other. This makes the cosmic environment conducive to violent intercivilizational conflict as well as vulnerable to galactic existential risks that require coordination to be reduced or mitigated. There are two main implications of this. First, before our civilization ever launches space settlements outside of our solar system we will have to give a lot of thought to cosmic value alignment and collective action problems. Second, we have to update the likelihoods of different explanations for the Fermi Paradox, which describe how our current cosmic social environment looks like. The impossibility of galactic political organizations is strong evidence against the Zoo hypothesis, which assumes that there is a benevolent cosmic society out there and that we live in a dedicated nature reserve. Conversely, cosmic anarchy increases the likelihood of the Dark Forest hypothesis, which assumes that the galaxy is an uneasy equilibrium between many civilizations that could be described in Hobbesian terms as a “war of all against all”. As a consequence, this post argues that the risks of attempts to message extraterrestrial intelligence (METI) should be taken more seriously and ends with a call for an immediate moratorium and the establishment of proper governance structures.
Table of Contents: 1) Chronic distances; 2) The evolution of states; 3) How slow can you govern?; 4) The cosmic speed limit; 5) How big is space?; 6) Faster-than-light communication and travel; 7) Governance latency for non-human subjects; 8) Space settlement; 9) Cosmic value alignment; 10) Cosmic collective action problems; 11) Implications for the Fermi Paradox; 12) Summary and conclusion
1) Chronic distances
There are several useful lenses to think about distances, the most important ones being metric, chronic and economic. For example, something can be 10 kilometers away, it can be 10 minutes away or it can be 10 dollars away. Metric distances have the advantage of being the most universal. For example, the equatorial circumference of Earth is about 40’075 kilometers. This metric size of the world is the same for everyone and has remained very constant over time. The same cannot be said for its chronic or economic counterparts. Firstly, chronic distances are subject to daily and seasonal fluctuations. For example, from the chronic perspective your home is further away if there is a traffic jam or for Napoleon’s and Hitler’s armies Russia suddenly got bigger in winter. Secondly, and more importantly, chronic distances depend on travel technology and hence have been shrinking massively in lockstep with technological progress. When Goethe travelled to Italy it was the trip of his lifetime, nowadays, he could just do it as a weekend trip by plane. Jules Verne’s classic 1873 book Around the World in Eighty Days revolves around the adventurous bet of the wealthy English gentleman Phileas Fogg to cross the whole world in a mere eighty days. Something that was reasonably challenging, yet still realistic, as inter alia shown by the first isochrone world map created for the Royal Geographic Society in 1881.
John Bartholomew created a similar isochrone world map centered on London in 1914 that shows how the world got smaller over the decades.
Fast forward one hundred years into the present and the distances have shrunk even more dramatically.
Note that the visuals can be deceiving here. Galton’s original map had “within 10 days” as his closest isochrones. The map from Rio2Rome has “over 1.5 days” as its furthest isochrones. Today’s Phileas Fogg could literally stay at home in London until day 78 and then board commercial airliners to make it on time (eg. London-Shanghai-New York-London). And of course it’s not just connection to and from London that have improved. From any bigger airport you can reach almost any populated area of the world within about one and a half days.
All the things mentioned above for chronic distances, such as daily and seasonal fluctuations as well as a longer-term decline of prices due to technological progress, are true for economic distances as well. The only caveat to add is that often there is some kind of trade-off between chronic and economic distance, where you can either pay more for something or someone to arrive faster at a destination or to pay less to arrive slower. While there are no equally nice isocost maps that show how economic distances shrunk over time, the following graph does a pretty good job as well.
As you can see in figure 4 the prices for the transport of information, goods and people have all declined substantially between 1930 and 2005. Average international freight charges per ton decreased by about 80%, the cost per airline passenger mile traveling decreased by about 90% and the costs for a three-minute call between London and New York decreased by about 99.7% (original data from OECD, 2007, p. 187)
Calculating chronic and economic distances requires more information than metric distances, such as what object or person travels with what technological means at what point in time. However, this additional information is not just a gimmick. Specifically, the chronic and economic distances can tell us much more about the feasibility of travel and transport than the metric distance. For example, if you look up car routes, public transport schedules or flights I would predict that you pay more attention to travel time and travel cost than about the number of meters you travel.
2) The evolution of states
“For 99.8 percent of human history people lived exclusively in autonomous bands and villages. At the beginning of the Paleolithic [i.e. the stone age], the number of these autonomous political units must have been small, but by 1000 BC it had increased to some 600,000. Then supra-village aggregation began in earnest, and in barely three millennia the autonomous political units of the world dropped from 600,000 to 157.” (Carneiro, 1978, p. 219)
The first states, defined as „political organizations with a centralized government that maintains a monopoly of the legitimate use of force within a certain territory” (Cudworth, Hall & McGovern, 2007, p. 95), emerged around 5000 years ago. These first states generally used to be city-states like the Greek polis and international politics used to be a very regional affair. Athena’s diplomacy focused on the neighboring city-state of Sparta, not on Ankara or Berlin. For the first 4’500 years most states had very little interaction with each other and were often not even aware of each other’s existence, let alone part of the same global capitalist economy.
None of the earliest states have survived as sovereign entities. The oldest state that has continuously existed until today is San Marino, dating back to 301 C.E. In the middle ages most other European city-states could only protect their interests by joining leagues of sovereign city-states, such as the Hansa, or deferring their sovereignty to a larger entity, such as France. However, the institutional logic of large sovereign states gave them an advantage over coalitions of small sovereign entities. Central authority reduced transaction and information costs and created a unified economic climate, whereas the cooperation in the Hanseatic system with decentralized sovereignty were plagued by continual defections. Subsequently, natural selection gradually made sovereign city-states disappear. (Spruyt, 1994, p. 185)
However, if large sovereign countries were a strictly superior form of organization to smaller sovereign groups, towns or tribes, why did they not succeed in the first 299’000 of the approximate 300’000 years of human history? Political organizations are embedded in a technological context and different types of organizations are suited for different environments. Spruyt (1994, p. 184) identifies expanding trade as the decisive factor that began making larger sovereign entities more efficient. Yet, this still fails to explain why trade expanded in the first place. The two basic drivers that have increased trade as well as the optimal size of governing structures over time have been advances in communication and transportation technology. Seen through the lense of isochrone and isocost maps (figures 1–4), it’s not the political organizations that have gotten bigger over time, it’s the world that has gotten smaller.
The global village envisioned by media theorist Marshall McLuhan has increasingly turned into reality. Thanks to the Internet, Donald Trump can type an angry tweet about something Emmanuel Macron said to a French radio station and mere seconds later even a teenage boy in a remote village of Indonesia has access to it. At the same time, our global village continues to have 193 chieftains, who recognize no higher authorities than themselves. On the one hand, we can see this as a policy puzzle. Why are nation states so persistently dominant if a hypothetical central world government could communicate with most of their citizens within seconds and physically reach almost all parts of their territory within a few days? Moreover, public support for world federalism peaked shortly after the Second World War and with the end of colonialism the number of states increased rather than decreased. On the other hand, we can see the technological potential for communication and control distances as an early precursor for the actual political development. Especially those factors related to identity such as culture, language and genetics are changing at much slower rates than technology. For example, the Internet is still very much an emerging technology in the grander scheme of things, yet in many ways it has already drastically aligned global attention spheres. It may be an unpopular thing to say in the current political moment but globalism is still winning pretty hard. Sure there are some hiccups and serious obstacles along the way, yet wherever you go on the planet, the global middle and upper class drinks Starbucks, browses Facebook, watches Netflix, speaks basic English and makes fun of Donald Trump. Just try to imagine for a moment what the cultural effect of 500 years of Internet access for every human on the planet would be. The crucial question is not so much whether Earth civilization would develop a political roof eventually, but whether we will be able to build this roof before the collective action problems in the current semi-anarchic condition lead to an existential catastrophe.
In short, the speeds of communication and transport have been key constraints to the metric size of political organizations. A reasonably low decision latency is a necessary but not a sufficient condition to exert centralized control over a territory. We also know that the global chronic and economic distances have shrunk so much that a world government is easily within the feasible spectrum. The question that we want to turn to now is where the limit is. In other words: What is the chronic size of the biggest possible sovereign political organization?
3) How slow can you govern?
Unfortunately, there is no good data on the chronic size of historical governments; however, the metrically largest governments provide a good starting point.
The British Empire was the largest political organization in history in terms of territorial control overall and like other colonial empires it contained some very remote parts, such as Australia and New Zealand. These regions were most distant from London in the time between the arrival of the first settlers and their connection to England via steamship (1831), telegraph (1871) or plane (1919). The First Fleet of sailboats that established the penal colony in 1788 took about 250 days, but faster exchange was certainly possible as Second Fleet ships already made it in close to 150 days. The Australian National Maritime Museum puts journey times before steamships at about 109 days or about 3.5 months, which would be the time required for communication as well as moving an army. As London to Australia is about as far apart as two points can be on a map and as all European colonial empire relied on sailboats to reach their most distant territories, none of them will be significantly above that number (eg. Columbus’ first journey to Hispaniola in 1492 took about 2 months, later ones were even faster).
The 13th century Mongol Empire stretched from Eastern Europe to Vladivostok and was largest contiguous land empire and the second largest overall. The Mongolian messenger system called Yam relied on horses was the most advanced messenger system of its time and one of the reasons the Mongolian empire could be stretched so far in the first place. It consisted of a system of stations with spare food, shelter, fresh horses and fresh messengers. From its capital of Karakorum the Mongolian empire extended furthest West. In 1241/42 the Golden Horde invaded large swaths of Central and Eastern Europe. During this invasion, on the 11th of December 1241, the second great khan Ögedei died on a drinking binge in his capital. According to the account of Giovanni da Pian del Carpine the news reached the Mongol forces under Batu Khan in Central Europe within 4–6 weeks, in January 1242. However, not all historians trust this number. The normal messenger speed between central Europe and Mongolia was closer to 3 months and Carpine’s own travel with a Mongol party from Kiev to Karakorum took about five months. Either way, the Yam was an impressive system and is put at speeds of 200 to 300 kilometers a day. It was kept alive in Tsarist Russia until the early 20th century and its speed amongst animal-assisted systems might only be beaten by the Umayyad’s Barid system that relied on camels.
For comparison, the Roman equivalent called Cursus Publicus was only able to relay messages at about 61–100 km a day. The maximum extension of the Roman Empire was eastwards, specifically in winter, when messages could not travel by sea from Rome. When Pertinax became the new emperor on January 1st, 193 CE the message had to take more than 3’000 kilometers by land to Byzantium and around 1’800 kilometers by ship to be announced 63 days later in Alexandria, Egypt. Of course, Alexandria is still on the coast and there were more remote settlements, so it probably took about 3 months before all Roman subjects would know about their new leader.
As soon as countries started to replace horses and ships by telegraphs it seems pointless to look for any records in maximum communication time. On the other hand, there are still some examples that show how long it can take to move significant amounts of military material and men to the outer regions of a country. For example, in the war between Russia, the third biggest empire of all times, and Japan at the beginning of the 20th century, Russia sent its Baltic Fleet to back up its blockaded Pacific Fleet at Port Arthur. The ships departed in October 1904 and arrived a bit more than seven months later in May 1905. By that time Port Arthur had already fallen for four months and the Russian fleet made its long journey only to be defeated within two days in the Battle of Tsushima.
ICBM’s or cyberattacks have substantially decreased chronic distances for military responses as well in recent times; however, these are often improper means to respond. So, when Argentina made a surprise attack on the British Falkland Islands in the South Pacific in 1982, it still took the British Task Force a while to get there, which lead to the strange situation in which people around the world could follow at home on their TVs for about a month how the British fleet was approaching war.
Overall, I would put the reasonable upper size limit for a government at about 3 months for one-way communication, resulting in 6 months latency. Similarly, I would put the reasonable maximum time to transport a significant amount of goods and humans in the form of an army at about 6 months, resulting in 9 months latency. These are ballpark figures based on the British(1.), Mongol(2.), Russian(3.) and Roman(25.) empires. There is no exact line before which a government is totally sound and after which it suddenly stops to work. Rather, the more time it takes for a central government to communicate with its outer regions, the less decision-power lies with the central government and the more sense it makes to have two or more separate governments.
For example, wherever we put the fastest possible delivery time of messages between the Mongolian capital and its outermost regions in central Europe, it’s already clear that Karakorum had little direct control over the Golden Horde. If Batu Khan’s army were attacked by an European army, the Mongolian leader could hardly ask his boss in Karakorum how to respond. Even if we assumed that Yam could deliver his message to Mongolia within 5 weeks, the response from the Great Khan would be immediate and took only 5 weeks to travel back to Central Europe, the Mongolian Army would be directed with a latency of 10 weeks and a bandwidth for input and output of a couple of kilobytes. Such an army would lose to snails. I don’t think anyone today can imagine living under a government with months of latency and my judgement from the history books is that the Mongols were already pretty close to how far you can stretch a government before it bursts into parts with an unusually fast information system, unusually fast armed forces and high levels of autonomy for military as well as political leaders. For example, the taxation of conquered territories from the central government was occasional and mostly geared towards providing supplies for the Mongol army and their information system (Smith, 1970).
Nevertheless, we haven’t reviewed all historical governments and counterfactual governments could have governed with even bigger latencies. In order to increase our confidence that central governments above a certain latency threshold are so impractical as to be infeasible, let’s double the initial estimates to 6 months for one-way communication and 12 months for moving an army. If anyone thinks that’s still not conservative enough, feel free to use a bigger number. I am also happy to offer 100 CHF to the first person that can show that any of the more than 1 million sovereign political organizations throughout human history had a regular (as in not due to unique circumstances such as navigation mistakes; control sustained for at least 5 years and 5 interactions with capital) minimum (using the fastest means available) one-way communication time of more than 6 months between its capital and a significant portion of its citizens (>1%) within its territory.
4) The cosmic speed limit
Ever since Einstein’s theory of special relativity it has been the consensus of physicists that our Universe has a clear speed limit in the form of the speed of light (c=299'792’458 m/s). Conveniently, radio waves and lasers travel at the speed of light. Hence, communication at the speed limit is not just possible, it’s already reality. However, it’s impossible to move something with mass, such as an army, at the speed of light, as this would require an infinite amount of energy. How close a future civilization could come to the speed of light is hard to say. The recently launched Parker Solar Probe is planned to reach a record-breaking maximum of 0.064% the speed of light, providing a lower boundary. All things considered something like 50% the speed of light seems reasonably optimistic and would for example be the same number Nick Bostrom (2014, p. 101) used when he calculated our cosmic endowment.
Combining the maximum speeds of communication and transport with the maximum latency of governance we can put the maximum radius for a government at about 6 lightmonths.
5) How big is space?
Space is big. Really big. Which is why metric distance is usually not labeled in kilometers but either in astronomical units and parsecs or in the time it takes light to cover a distance, which happens to be very convenient for our purposes.
Earth’s circumference at the equator: ca. 0.13 lightseconds
Distance Earth-Moon: ca. 1.3 lightseconds
Minimum distance Earth-Mars: ca. 3 lightminutes
Minimum distance Earth-Sun: ca. 8 lightminutes
Distance Sun-Neptun (outermost planet): ca. 4 lighthours
Distance Sun-Kuiper Cliff: ca. 7 lighthours
As we can see the limits of technology do not just allow for an Earth government but also for a government encompassing our whole solar system. A 14-hour communication latency is no insurmountable obstacle to a functioning government. What other celestial objects can we govern in a sphere with the radius of a lightday, a lightmonth or six lightmonths? Well, unfortunately, the answer is none!
Sun-Proxima Centauri (closest star): ca. 4.2 lightyears
Sirius (brightest star in the night sky): ca. 8.6 lightyears
Radius of the Milky Way Galaxy :ca. 50'000 lightyears
Sun-Andromeda (closest galaxy): ca. 2'500'000 lightyears
It’s not even close. Even under the pretty unrealistic assumption that the government would be based in a spaceship midway in-between our closest star and us the latency would be way too high for a common government. Galactic government structures such as the ones portrayed in Star Wars, Star Trek, Stargate, Dune or Foundation are even more unrealistic. Just to give some sense of how ludicrous such arrangements are under the restriction by light-speed, let’s assume for a moment the Milky Way had the same type of senate that Star Wars has. If there was some war on the outskirts of the galaxy it would take about 50'000 years for the news to reach the capital. The chancellor could then call a special session, where all senators have to come, so another 50'000 years for that message to get out and 100'000 years more for all representatives to arrive. Also, in Star Wars the chancellor is limited to one four-year term, so there would be 37'500 intervening chancellors, between the one calling for the session and the one holding it, except that there is no one there to vote for a chancellor in the first place. Then the special session may decide to send an army, which takes about 100'000 years more. Adding everything up the government latency to respond to a crisis would be about 300’000 years or roughly thirty times the development span of humans from stone age to singularity.
Consequently, Type III civilizations on the Kardashev scale, which would control the energy of a whole galaxy, are impossible. Anything higher than Type II, meaning controlling the energy of a whole solar system, is highly unlikely. Subsequently, outside of our own solar system, the popular term “space colonization” also seems quite misleading. The word “colonization” is primarily used to describe “a process by which a central system of power dominates the surrounding land and its components.“ However, space is simply too big for interstellar central systems of power. Similar terms with less misleading connotations would be space settlement or space emigration.
6) Faster-than-light communication and travel
We happen to live during a brief and interesting period of discovery, however, in the long-term some kind of sigmoid development curve seems a much more reasonable assumption than eternal progress. The Universe is based on strict rules and after some progress civilization simply reaches the limits of what’s possible within those rules. Given our current scientific understanding our default assumption should be that faster-than-light communication and travel are not possible. However, it is of course true that there’s quite a bit of uncertainty whether this cosmic speed limit is really as absolute as we tend to think. Firstly, our understanding of the laws of physics may be wrong and faster-than-light travel through space could be possible. Secondly, Einstein only prohibits faster-than-light travel through space, so people have speculated that it may just possible to deform space instead with something like an Alcubierre drive.
So, what if faster-than-light speed communication were possible? Well, it really depends on how much higher the true cosmic speed limit would be. Let’s say for example it would be possible to communicate 50% faster than light. That would change next to nothing. It still clearly wouldn’t suffice to form a bistellar government between Proxima Centauri and our sun. For that we would need something closer to 400% the speed of light. The requirements for a galactic government seem even more ridiculous. To reach the ultra slow communication speed of 6 months per way from the center of the Milky Way (while also ignoring for a moment that there’s actually a black hole there), we would have to be able to communicate at 100’000 times the speed of light.
So, faster than light communication and travel would not be enough. We would need massively faster than light communication and travel for the problem of space governance to change significantly. Not to mention that it would also have to be safe, reliable and affordable. Maybe it would be theoretically possible to massively deform space but it would damage something in the fabric of space or it may just require energy on the scale of whole suns for a trip. While I do think it’s worth it to explore the implications of counterfactual cosmic speed limits as well, we cannot simply disregard our current understanding of physics for cosmic sociology.
7) Governance latency for non-human subjects
We have deduced the maximum governance latency from the historical governance of humans. Yet, who are we to simply project the maximum latency of human governance onto the space of all possible minds? In other words, what are more universal factors that determine the maximum latency of governments?
An intuitive first answer may be lifespan. A dayfly would certainly want lower government communication latency than a year. Conversely, a digital brain that can live for millions of years might still find it useful to hear back from the government in 100 years. However, I don’t think lifespan is the causal factor here. A longer lifespan does provide better incentives to think more, and more longterm, which correlates with more interest in problems or tasks that are compute-hungry and time-insensitive. So, a longliving organism may still have something to gain from a “government” with a latency of one hundred years, yet, it could probably gain much more from a government with lower latency. If something would threaten your life in 10 minutes you would only find government help useful within that timeframe. No matter how old you can get.
On second consideration the speed of thought that a mind has and the related speeds of perception and action seem more crucial for governance speed. Put simply, the opportunity cost for deferring decisions at a fixed latency arguably increases as those speeds increase. Our current human society is quite exceptional in that the internal communication speed of our brains is significantly slower than our external communication speed across vast distances. Ceteris paribus this favors centralization. Machines on the other hand can run at the speed of light and tend run at much more “thought cycles” per second (currently around 2–4 GHz) than humans (ca. 40 Hz). Judged from that perspective machine governance would have to be around a billion times faster, which has its own interesting implications as it shrinks the governance radius well below what’s needed for a world government. However, there is also a trade-off between brain size and thought speed, so if digital minds were much bigger than our brains they could become equally slow. Bostrom calculates that a digital brain of the size of a smaller planet would be about equally slow in terms of round-trip latency as a human brain (2014, p. 59). Hence, at least a full-blown Matrioshka brain might think in longer intervals than humans.
There are many more factors that have to be considered for governance speeds such as bandwidth and the difference in information and computing power between central and edge decision-nodes. Putting all those factors in a coherent model is beyond the scope of this text. However, deferring decision-power to a central body with high latency really only makes sense if that body has superior computing capacities and the problem is not time-sensitive. For example, Sirius has about 25 times the luminosity of our sun. Assuming equal rates of turning the energy emitted by stars into computing power, equal availability of that computing power for a specific problem, that Sirius civilization could be fully trusted and that bandwidth between the sun and Sirius was not an issue, it only would make sense to defer problems that are not time-sensitive and would require at least 1800% of Sol’s yearly computing power (as 17.2 years are lost on latency).
Overall, my intuition is that the overwhelming majority of tasks or problems relevant to digital minds would have to be governed at much lower latencies than those for humans. Especially tasks related to security, maybe the most fundamental pillar of a government, would likely require low latencies. If unistellar civilizations could fully trust each other, which is doubtful if they cannot collaborate on security, there would be room for a dialogue on „grand questions“, but that’s about it. Having said that, I am happy to admit that there is significant uncertainty on this point.
8) Space settlement
So far, we have looked at why governing more than one solar system is likely impossible, whilst also discussing some uncertainties that could possibly challenge this assessment. However, moving forward we will assume that our initial hypothesis is correct and that anything above a Type II government on the Kardashev scale is impossible, in order to tease out what consequences this would have.
First off, we need to be very clear. The inability to “colonize” space does not mean that we cannot settle space. To provide an analogy, early Homo sapiens were able to move out of Africa and spread all across the globe to Europe, Asia, Australia, North America and South America without horses, let alone modern means of transport. Governance requires regular interaction, for settlement it’s enough to just keep moving in a direction. So, despite having finished settling the globe around 10’000 BCE, homo sapiens lived in small hunter-gatherer tribes rather than city-states, nation states or a world government.
In fact, nowadays, we can be highly confident that space can be settled as a variety of papers show that the settlement of our galaxy is doable within a relatively short time and that even an unistellar civilization like ours can launch intergalactic settler probes (Sandberg & Armstrong, 2013; Sandberg, 2018). For the purposes of settlement we can very broadly divide space into five zones:
The governable Universe: In this sphere with a radius of about 6 lightmonths from the center it is at least somewhat feasible to have a central government. Of course, decisions can be delegated to different degrees and feasibility does not mean that it is optimal, necessary or even probable that a government will exist on that scale. Still, within this sphere the concept of space colonization is a coherent notion and coordination to solve collective action problems is possible.
The returnable Universe: This describes the current sphere of all objects to which a space probe or a communication could be sent out at lightspeed in the present to make a round-trip and return to the center.
The reachable Universe: This describes the current sphere that is reachable (one-way), if we were to send out probes at light speed at this moment.
The observable Universe: This describes the current sphere whose objects would have been able to emit light or other information at some point in the past that can reach the current Earth. It has a radius of about 46’500’000’000 lightyears (46.5 gigalightyears) and a volume of about 408 trillion cubic light years. In the earliest days of the Universe, before the age of recombination, there was no visible light, so the visible Universe is a bit smaller than the observable Universe with a radius of about 45’500’000’000 lightyears (45.5 gigalightyears). Even though most of the observable Universe is forever unreachable by now it is of course possible that a distant civilization has already sent out probes billions of years ago that will reach us at some point.
The Multiverse: We don’t know for sure how big our Universe is or if it is even finite, however, based on the observed flatness of the observable Universe the unobservable Universe is at least 250 times bigger. Furthermore, many astrophysicists nowadays believe in the existence of an infinite amount of Universes in physical reality.
For practical purposes we do not have to care about most of our Universe and the multiverse as we will never interact with it in any way. If we as a civilization would plan space settlements it’s really only the first three zones that come into question. Settling within the governable Universe is very attractive as we’re talking about close distances and quite controllable political risks in exchange for more civilizational resilience to many existential hazards such as climate change or asteroids. Conversely, sending out settler probes to the returnable but non-governable Universe comes with some amount of political risk. Specifically, we would spawn civilizations outside of our political control that could potentially decide to invade or annihilate our solar civilization at some point in the future. Sending the probes out further on a long intergalactic journey into the reachable but non-returnable zone should be almost risk-free again, as those settlements will not be able to influence our civilization’s timeline.
The adequate level of paranoia about space settlements in the returnable but non-governable Universe mostly depends on the feasibility of cosmic value alignment and the severity of cosmic collective action problems. Based on that optimal strategies may involve waiting until the expansion of the Universe puts more galaxies in the reachable but non-returnable zone or it may at least include reducing information hazards. If Earth civilization were to send out hundreds of probes to solar systems in the Milky Way it may be better to make sure that they have no memories or false memories of where they came from. Not only would all new civilizations otherwise only know one other solar system that should be habitable with high certainty (ours), but they would also know that our solar civilization would be the only civilization to know their location. If for some reason even just one of those civilizations would decide that this knowledge poses an existential information hazard to them, the settlements of our civilization may come back to haunt us. Either way, the mere fact that space civilizations seem to follow r-selected animals (“You’re bugs!”) in that they have the ability to reproduce relatively quickly with a quite massive multiplying factor means that we have to think long and hard before sending any probe outside of our governable zone to spawn a new civilization as we will irretrievably lose direct control over the settlement process thereafter.
9) Cosmic value alignment
Once the settlers in Robinson’s (2009) Red Mars are beyond the control of their home planet, they rebelliously declare not to “pay any attention to plans made for us back on Earth!” (p. 77). Luckily, for extrasolar settler probes such spontaneous acts of mutiny against Earth-originating settlement plans will be of little concern as they will most likely not contain obscurant packages of wet-ware but software designed to be aligned with the interests and values of Earth civilization. However, it would also be a bit too intellectually lazy to assume that solving the AI value alignment problem between humans and AI would be sufficient to also solve the value alignment problem between Earth-originating space settlements and us.
First off, the methods that we will likely employ to try to align superhuman AIs with our values won’t be available in deep space. Specifically, most people assume that some kind of (inverse) reinforcement learning will be needed to not just reflect the current human utility function but to allow for moral progress and the co-evolution of AI along that path. In other words, the current models rely on observation or feedback from humans and settler AIs will simply lack access to these humans. While impractical, it may not be necessarily impossible to ship some humans along with the settler AIs, yet, that wouldn’t stop those humans to diverge from Earth’s human population.
Secondly, the value alignment would have to be ridiculously robust to not be subject to any drifts anywhere and without governance there’s simply no correction mechanism. In other words, if we were to send a thousand identical seed AIs with a thousand identical IKEA plans to assemble a civilization on a thousand space probes to a thousand star systems around the galaxies we would most definitely not end up with a thousand identical civilizations a few million years later. Reasons for this would include:
· Extremely high variance in solar environments and impossibility to predict the exact respective environments at send-off. Environments will differ in factors such as the relative abundance and availability of elements, gravity, radiation, electromagnetic storms and temperature.
· Due to the high variance in environments and mass constraints at send-off, probes cannot succeed if they can only follow very narrow instructions. They need to have a general ability to learn and grow
· Successfully matured settlements will have vastly different computational capabilities as the stars in the Milky Way produce highly varying amounts of energy per second. Giving the exact same algorithm massively more or less computer power leads to different outcomes.
· Settler probes will be subjected to highly varying degrees of space dust, radiation and electromagnetic interference on their long journeys that will lead to a variety of small-scale and large-scale damages.
· The sheer number of star systems means that sending out settler probes would have to be an iterative process with 1st gen settlements creating 2nd gen settlements with their local resources, which would go on to create 3rd gen settlement and so on. Even in a high fan-out scenario the vast majority of settlements will be removed from Earth by at least two intervening civilizations.
· Chaos theory states that small differences in initial values of complex and highly interdependent systems can compound over time and result in massively different outcomes. Hence, the proverbial butterfly in Brazil can cause a tornado in Texas, or, a 1-bit difference in year 0 can result in a dramatic value difference in year 1 million.
The more value mutations the above factors cause, the stronger the effect of natural selection will be, which will not select for the most noble values, but those most suited to spread and multiply in the cosmic (social) environment.
10) Cosmic collective action problems
Let’s assume for a moment that a civilization solves the cosmic value alignment problem well enough, so that value differences amongst the billions of unistellar civilizations in the Milky Way would be relatively miniscule and stable. Would that make-up for the lack of a central government? Unfortunately, the answer is no. The impossibility of interstellar governance and the resulting lack of an enforcer is not a mere detail. It means that civilizations are inherently incapable of solving interstellar collective action problems.
Let’s take humans as an example. We do have our differences. However, seen from the perspective of the space of possible minds we are all very close aligned. The most recent common ancestor of all living human beings is likely only a bit more than 2’000 years away, two randomly selected humans would share about 99.9% of their DNA and I have yet to come across a member of our species that does not love pizza, sex and oxygen or hate loneliness, pain and smog. Yet, humans have been in violent conflict with each other since the dawn of history. We could naively ascribe this violence to things such as “too much hate” or “not enough love”. However, Hobbes was probably much closer to the truth when he focused on the incentives of the overall system rather than the inherent evil or good of individuals. Without the ability to solve their interpersonal security coordination problems Hobbes decried the lives of early humans as “nasty, brutish and short” and as a „war of all against all“. Conversely, he suggested that the state (“Leviathan”) could offer its citizens security in exchange for some aspects of their individual freedom. Indeed, excavations have shown that stateless humans from all across the globe died at much higher rates due to violence than 20th century citizens of states, and that despite two World Wars (Pinker, 2011, fig. 2–3). Furthermore, we can see an effect of introducing police into previously unpoliced areas. For example, the historic crime rates in Canada show a clear negative correlation with the respective distance from the next Mountie fort (Restrepo, 2015). In short, Leviathan works.
From a game-theoretic perspective this is not surprising. In unpoliced environments justice is a self-help system and people need to rely on their reputation as someone not to cross, deterrence in the form of mean looks and weapons as well as a second-strike capability in the form of family vendettas to make sure that no one dares to harm them. Even if everyone would share an interest in upholding community norms the enforcement often lacks as third party individuals that challenge a transgressor have often little immediate benefit for considerable risk that they have to take on. Today, we have mostly solved the Hobbesian anarchy between individuals thanks to police and the justice system that control and enforce community rules as neutral third parties. However, the anarchy in the international system between states as described by Kenneth Waltz (1979) is still quite pervasive and conflicts between them are unfortunately still often solved by self-help rather than global courts and global police.
Even worse, as already mentioned in the discussion of world government in a chapter 2, the consequences of the failure to solve these collective action problems may not just be widespread violent conflict but extinction. Nick Bostrom’s vulnerable world hypothesis states that, „if technological development continues then a set of capabilities will at some point be attained that make the devastation of civilization extremely likely, unless civilization sufficiently exits the semi-anarchic default condition.“ (Bostrom, 2018, p.6), and he lists three types of threats that can create such a vulnerable world:
Type-1 vulnerability: „There is some technology which is so destructive and so easy to use that, given the semi-anarchic default condition, the actions of actors in the apocalyptic residual make civilizational devastation extremely likely.“ (p. 9)
Type-2a vulnerability: „There is some level of technology at which powerful actors have the ability to produce civilization-devastating harms and, in the semi-anarchic default condition, face incentives to use that ability.“ (p. 12)
Type-2b vulnerability: „There is some level of technology at which, in the semi-anarchic default condition, a great many actors face incentives to take some slightly damaging action such that the combined effect of those actions is civilizational devastation.“ (p. 14)
Of course all these same types of concerns that apply to Earth and collective action problems can be applied to bigger scales. Something, we could call the Vulnerable Galaxy Theory and the Vulnerable Universe Theory respectively. For example, it could be that some activities, e.g. the acceleration of space fleets close to c, creates tears in the fabric of space or generates black holes that individually may be of limited concern (and possibly not even noticeable at the time of initial space settlement), but collectively lead to catastrophic outcomes. Or, it may be that any technologically mature civilization can create technology to unilaterally destroy the Universe, such as false vacuum decay. Such prospects of a vulnerable galaxy or a vulnerable Universe are quite scary. Even within our galaxy cosmic anarchy is much more Hobbesian than Waltzian with billions rather than 193 sovereign entities and no communication or shared forum between most of them. Overall, a vulnerable galaxy or a vulnerable Universe seem less probable than a vulnerable Earth, however, contrary to problems on Earth, these cosmic collective action problems would be on scales that are inherently ungovernable. In the most extreme circumstances, such as easy false vacuum decay technology, this could mean that the increased Type IV existential risk would dominate the astronomical waste argument and that it may not be rational to build settlements at all outside of our governance radius.
11) Implications for the Fermi Paradox
In 1950 physicists at the Los Alamos National Laboratory joked about a New Yorker comic over lunch, which suggested that aliens were to blame for disappearing trashcans in New York. The discussion had already moved on when Enrico Fermi, famous for his ability to make good estimates on complex problems, suddenly exclaimed, „Where is everybody?“ This puzzlement over the lack of any indications for the existence of extraterrestrial life is nowadays known as the Fermi Paradox.
The basic argument goes something like this: There are about 250 billion stars in the Milky Way and about 2 trillion galaxies in the visible universe. With recently developed methods to detect exoplanets, we nowadays also know that planets are a very common feature of solar systems. Based on data from the Kepler space mission there could be as many as 40 billion Earth-sized planets orbiting in habitable zones in the Milky Way alone. Furthermore, most stars in habitable zones of our galaxy are older than our sun. If Earth even remotely follows the mediocrity principle intelligent life should have developed in many, many places across the galaxy and hundreds of millions of years ago. Even at the slow pace of currently envisioned methods for interstellar travel the Milky Way could be completely settled within a few million years. What’s more, even unistellar civilizations like us have the means to send out settler probes to other galaxies. So, again, where is everybody?
Due to the feasibility of galactic and intergalactic space settlement in reasonable cosmic timeframes the initial puzzle along the lines of „Why can’t we detect any radio signals from other (unistellar) civilizations?“ has shifted towards the question „Why are we not part of an (inter-) galactic empire?“ Robin Hanson (1998) has theorized that there must be some great filter, a development step that is very hard to achieve or survive, that stops almost all civilizations before they reach the stage of space settlement. Either Earth has already past this great filter or it will have to face it between now and massive space settlement. Some great filter candidates in the past may for example be the formation of RNA, prokaryotic life, eukaryotic life or tool-using animals with big brains.
It’s needless to say that we as a civilization would highly prefer to find out that the great filter is behind us than that it is in front of us.
Subsequently, Bostrom (2007) has argued that finding any evidence of extraterrestrial life would be bad news for our species as it implies that all the filters that the extraterrestrial life has past are less likely “great filter” candidates and that the great filter therefore is more likely to lie ahead of Earth civilization.
Are the limits of governance a possible great filter ahead of us? Yes and no. Yes, in the sense that it makes Type III civilizations de facto impossible. No, in the sense that it doesn’t „filter out“ any civilizations before they could become space-settling civilizations as envisioned in Hanson’s original paper.
Note that figure 8 is not meant to be accurate in terms of the relative frequency of civilizations. However, assuming that the origin of life and other developmental challenges such as eukaryotic cells are non-trivial filters most advanced civilizations should assumed to the offspring of other advanced civilizations rather than the direct descendants of prokaryotic life in their territory. Put simply, we could call the limits of governance a great filter on the Kardashev Scale but it doesn’t dissolve the Fermi paradox as Type II civilizations can easily settle the Universe. Hence, a term like the “great ceiling” might be a more adequate description.
Having said that, this doesn’t mean that we can’t learn anything about the Fermi Paradox from the limits of governance. Many of the hypotheses that have been put forth to answer the Fermi question do not rely on great filters either. However, in this paper we will only look at hypotheses whose likelihood is impacted by cosmic anarchy. If you want a more comprehensive overview, see inter alia Urban (“Wait But Why”)(2014), Wikipedia, Sandberg, Armstrong & Cirkovic (2017), Sandberg, Drexler & Ord (2018). Cosmic anarchy should probably lead us to slight update against the simulation hypothesis as computing power is more likely to be wasted and to be used in smaller fragments under these conditions. If our Universe is representative of possible base Universes there is likely less computing power available for a simulation like ours. In the case of the aestivation hypothesis cosmic anarchy seems to increase the relative utility of aestivating as more galaxies are separated by safe distances at later points in time. Conversely, we might also want to slightly update against it in terms of feasibility. While deadly probes require much less autonomy than a seed probe, it’s still very hard to coordinate their comprehensive surveillance over long timespans with randomized failure rates. This list of small possible updates based on the assumption of cosmic anarchy could go on for long. However, realistically, most of it is quite speculative and hard to remember. Instead, I would encourage readers to focus on following three main take-aways from cosmic anarchy:
11.1 Strong update against the Zoo hypothesis
The idea that some kind of benevolent cosmic society has created designated nature reserves and just „waits“ for us to develop naturally before it would contact and welcome us has been put worth in different form such as Ball’s Zoo hypothesis, Star Trek’s Prime Directive or Terry Bison’s They’re Made Out of Meat. While the appeal of this idea is understandable, it always seemed highly unlikely that advanced civilizations would still fall for the naturalist fallacy and justify large scale suffering like the Holocaust as „natural“ development. Neither would it seem to make a lot of sense for a technologically mature society to build such a “nature reserve” in space and not in a simulation. That’s not just easier to observe and control but could be made much more energy-efficiently. Taking into account space governing restrictions the Zoo hypothesis now seems downright impossible to me. There simply is no organized cosmic society.
11.2 Strong update in favor of the Dark Forest hypothesis
The Dark Forest hypothesis is taken out of Liu Cixin’s Remembrance of Earth’s Past trilogy and assumes that the sociology of the universe is brutal, as there is a general offense advantage. Civilizations are anxious not to send signals to the outside world and are often not stationary. The Dark Forest hypothesis and similar dark explanations of the Fermi Paradox such as the Berserker hypothesis, which assumes that one civilization has sent out masses of deadly probes, have been dismissed by people mainly due the to two types of arguments.
· Natural Selection for Benevolence
A number of scholars have suggested that only highly ethical and peaceful civilizations manage to become Type I civilizations and higher either due to resource constraints or existential hazard technology:
„The limits of growth in a finite system, which will be imposed on all stellar civilizations by the colossal distances that separate the stars, will affect the natural selection of these civilizations. Those that manage to overcome their innate tendencies toward continuous material growth and replace them with non-material goals will be the only ones to survive this crisis. As a result the entire Galaxy in a cosmically short period will become populated by stable, highly ethical and spiritual civilizations.“ (Papagiannis, 1984, p. 309)
“Those civilizations devoted to territoriality and aggression and violent settlement of disputes do not long survive after the development of apocalyptic weapons. Civilizations that do not self-destruct are pre-adapted to live with other groups in mutual respect. This adaptation must apply not only to the average state or individual, but, with very high precision, to every state and every individual within the civilization. […] In any case, the result is that the only societies long-lived enough to perform significant colonization of the Galaxy are precisely those least likely to engage in aggressive galactic imperialism.“ (Sagan & Newman, 1983, p. 120)
The argument for natural selection in favor of these traits en route to become a Type I civilization is quite sound, however, it doesn’t have the implications that some wish it would have. It’s like saying all life evolved in water; therefore all life must have gills. Whereas correctly one would say, all life evolved in water; therefore most life had gills at some point in their ancestral history. Yet, life on land does not have gills anymore because it’s not useful to survive on land. On land there’s actually natural selection against gills. What matters most to predict cosmic sociology are the natural selection pressures within the cosmic system, not those within a different environment in scale and time. My assumption would be that most civilizations that are intelligent enough to venture out into space would also understand by themselves that intracivilizational conflict on the planet and intercivilizational conflict in the galaxy are two very different strategy spaces with different requirements for success. General intelligence is exactly the ability to succeed in many different environments and if a mere monkey can figure out I’m sort of confident that a Jupiter brain will be able to do so too.
· Fast Space Colonization
The assumption that space can be settled fast has been used a strong argument against some fragile dark forest equilibrium with many different civilizations. However, as I have tried to highlight throughout this text common ancestry is not sufficient to form some kind of semi-coherent notion of a civilization, you need common governance. All these space “colonization” models do not take into account that settlement will be independent political units in a self-help system without any higher authority that defines, communicates and controls community rules and enforces them by punishing any transgressors. As an example let’s examine the explanation Sandberg gives in his most recent Fermi Paradox paper why even settlers coming from different civilizations would not fight over resources:
„We will assume claimed resources are inviolable. One reason is that species able to convert matter to energy can perform a credible scorched earth tactic: rather than let an invader have the resources they can be dissipated, leaving the invader with a net loss due to the resources expended to claim them. Unless the invader has goals other than resources this makes it irrational to attempt to invade.“ (Sandberg, 2018, p. 2)
This assumption is insofar understandable, as the simulation of space settlement would get very complicated otherwise. However, it’s also clear that it doesn’t hold up to scrutiny. Deterrence fundamentally relies on signaling and the first massive problem in space is that can be very hard to tell from afar if a solar system is settled at all, let alone to credibly signal across thousands of lightyears that a civilization possesses some sun-exploding switch and is determined to use it. Of course, having such an easy off-switch at the ready is itself a huge risk for the defending civilization in terms of accidents or terrorists. Neither would it be a very credible deterrence as the defending civilization has very little to gain from destroying its own sun rather than fighting or fleeing and the distances of deep space travel are so big that no fleet of attacking probes would have fuel for a journey back. So, once the attackers would emerge out of the dark at their destiny there would simply be no other course for them than to attack. Lastly, the acceptable rate of failed invasion can in fact be pretty high if the size of an effective attacking fleet can be small versus the size of the solar system.
Given that selection pressures for peacefulness on the planetary level are a bad predictor for cosmic behavior and that a dark forest is not precluded by quick space settlement but may rather be its default outcome, we should take this theory more seriously. Of course, a lot hinges on an offense-defense balance that we currently know very little about, yet, cosmic anarchy seems at least quite conducive to such a messy „war of all against all“.
11.3 Strong update against the use of METI
The Search for Extraterrestrial Intelligence (SETI) is about passive listening and looking for signs of extraterrestrial life across the Universe. It’s counterpart active SETI or Messaging Extraterrestrial Intelligence (METI) is about actively sending out signals and messages to star systems and galaxies that may contain extraterrestrial life. Whereas SETI is relatively uncontroversial some of the most distinguished voices in astrophysics have warned about the dangers of METI. For example, the most famous astrophysicist of recent times, Stephen Hawking, warned very clearly that “If aliens visit us, the outcome would be much as when Columbus landed in America, which didn’t turn out well for the Native Americans.” Even Carl Sagan, the famous astrophysicist and co-author of the paper cited above that assumes that there is a natural selection for benevolence, criticized METI as “deeply unwise and immature,” and recommended “the newest children in a strange and uncertain cosmos should listen quietly for a long time, patiently learning about the universe and comparing notes, before shouting into an unknown jungle that we do not understand.” (Brin, 2013)
Now, given the potential catastrophic consequences of messaging our location in a Dark Forest scenario, you might think there would be a global civic debate or even a democratic vote before any such haphazard project would even be considered, coordinated by some internationally legitimized body preferably within the United Nations. Yet, that’s not at all what has happened. The first serious exploration of METI was done by NASA in 1971 in its Project Cyclops Report, which listed the risks of invasion, exploitation, subversion and cultural shock. The report also noted that “by revealing our existence, we advertise Earth as a habitable planet“ and recommended that before a decision to send anything „the question of the potential risks should be debated and resolved at a national or international level.“ (pp. 31&32) Nevertheless, there were two initial projects in the 1970s without prior international agreement, the Arecibo Message and the Voyager Golden Record. On the positive side, these projects still took risks into account, insofar as the message had been sent to a quite distant region in the Milky Way (making it harder for the recipient to guess where exactly it came from and giving Earth more time) and the Voyager Golden Record could conceivably still be destroyed by future generations of Earth if they perceive it as an information hazard.
Later on, governments largely forgot about METI and a small group of researchers (unsurprisingly not a single one of them has any political science background) that likes METI just sort of lost patience and unilaterally decided to move ahead without properly addressing concerns and has continued to send out our position to new star systems. So far, we have sent out 31 (!) messages to different star systems and most of them will arrive there between 2017 and 2069. That’s pretty soon and yet still late enough so that future generations and not those sending out the signals will bear the risks of a potential reaction. Just to give you a hint as to what a bad parody of a civilization we are living in: In 2012 the magazine National Geographic decided that it would be a cool promo for its new TV series on UFOs to send out radio signals of all tweets with the hashtag #ChasingUFOs and the message of a comedian to the star system from which SETI has picked up the mysterious “Wow! Signal”. Our non-governance of METI risks is so ridiculous that one would want to laugh, but has to cry because we possibly have sentenced all future generations of Earth to death just because some “creative” marketing guy for a magazine thought that it might be a cool promo to shout tweets to the stars from which an unexplained radio signal has been detected.
The only real argument in favor of METI brought forward by its advocates has been the Zoo Hypothesis. Yet, as discussed above it would be weird for a highly advanced and highly moral galactic society to fall for the naturalist fallacy. Neither does it seem obvious at all that such a cosmic political organization would change its mind about non-interfering once we shout at stars and that it would start to shower us with far-advanced technology rather than put the dinosaurs back in place. Most importantly, however, the distances in space are simply too big for any benevolent cosmic society to exist that could decide and enforce to turn whole swaths of the galaxy into a dedicated “nature reserve”. Hence, the Zoo hypothesis is impossible or at least one of the least likely hypotheses of the ca. 100 different answers that have been proposed to explain the Fermi paradox. The few other arguments brought forward by METI are all fallacious whataboutism. First, it’s true that Earth has emitted radio signals into space by accident anyways. However, the notion that aliens would suddenly see pictures from Hitler as depicted in the movie Contact is fiction. While surely not optimal, the weak signals that are meant for consumption on Earth disperse and fade into the cosmic background noise relatively quickly. Not even the huge Arecibo radio telescope would be able to detect Earth’s TV transmissions, if they were broadcasted from our nearest neighboring stars. Theres a millionfold difference between this and sending high-energy directed messages to specific star-systems. It’s a bit like saying: “Well, we already slightly rustled some leaves by accident so we might as well scream on the top of our lungs into a megaphone”. Second, it’s also true that Earth’s biosignature, most notably oxygen, may have been giving us away for a while already, but then again you can only reliably detect biosignatures if you’re reasonably close and they do not provide certainty as there are still many false positives and false negatives. Third, SETI and strong radio transmitters are not regulated, so it’s possible that an extraterrestrial signal would be met with an uncoordinated response from Earth. However, this is not an implicit endorsement of METI or an argument for shouting into the jungle pro-actively at all, it’s an argument for adequate information hazard protocols regarding SETI and regulation over strong radio emitters, exactly so that not any idiot at a magazine can unilaterally decide to send tweets to space regions from which we pick up strange signals.
It’s not that METI advocates would not be aware of the risks or of the fact that they are deciding the fate of future generations without any due process, when we are still in the very early stages of figuring out our cosmic environment. It’s really just that they decided to move forward anyway. Governments, which unfortunately tend to think in election terms and more often address risks with reactive rather than preventive action, have so far utterly failed to regulate METI. A moratorium on METI seems to be the very least that the precautionary principle would demand in a situation with such high stakes and no adequate academic and public evaluation. Accounting for cosmic anarchy, which strongly updates against the already unlikely Zoo hypothesis (best-case for METI) and strongly updates in favor of a Dark Forest scenario (one of the worst-cases for METI) means that METI needs to be called out in even stronger terms for what it really is: An arrogant and unnecessary gamble with the lives of our children and the security of our civilization with no benefits expect for boosting a few egos.
12) Summary and conclusion
Whereas there has been a decent amount of deliberation in the academic literature on governance forms of individual space settlements and specifically on the question of whether the harsh conditions in outer space make authoritarianism inevitable (eg. Cockell, 2013), there has been very little effort to understand the general cosmic political environment. Space is an endless desert of nothingness with oases that are very far apart from each other. As we have shown it is de facto impossible to centrally govern multiple star systems as the latency of central decisions would be too high to be practical for almost any tasks, including security, which historically has been the foundation for further types of cooperation. This observation seems quite robust to upward corrections in the fundamental limits to the speeds of communication and transport in our Universe as well as to the extension of governance to digital beings. Consequently, the ideas of space “colonization” outside of our solar system and of future galactic governing structures are misleading. Space settlement is feasible, but it will lead to many sovereign organizational units.
Subsequently, the best lens to examine the cosmic political environment is through the concept of anarchy. Specifically, the type of pre-state anarchy described by Hobbes seems a closer analogue to cosmic reality than the rather semi-anarchic environment between states described by Waltz that still involves many common institutions. Cosmic anarchy implies that civilizations need to rely on self-help and are unable to solve intercivilizational collective action problems. Hence, even if cosmic value alignment were successful, galaxies could still end up in a “war of all against all” and possibly even create unmanageable galactic or universal existential hazards.
We currently only know very little about our macrocosmic environment and the puzzle of the great silence out there. The most likely explanations are probably still that we are either alone in the observable Universe, that we are part of a simulation in another base reality or that advanced extraterrestrial civilizations are in summer sleep. However, understanding the cosmic political environment should at least bring us to strongly update in favor of the Dark Forest hypothesis and to strongly update against the Zoo hypothesis. Even under the assumption that we are currently alone in space, we should still intensely study Dark Forest dynamics as our control over far distant future settlements would not be as absolute as some assume it to be. The argument for longtermism is strong enough even with a bit more modest assumptions on our influence on the far future. However, the argument for METI, which already has been highly questionable before, seems outright insane under cosmic anarchy. Therefore, this paper ends with a call for action. METI poses a real existential risk to human civilization and should be governed accordingly. The first priority should be an immediate moratorium on all METI activities. Secondly, interstellar communication on behalf of our entire civilization cannot continue to be the hobby of a small self-appointed group with no political legitimacy or accountability (the SETI & METI institutes are both privately funded). Rather SETI and METI research should be institutionalized within the United Nations under proper oversight.
