Implications of a search for intergalactic civilizations on prior estimates of human survival and travel speed

S. Jay Olson∗ Department of Physics, Boise State University, Boise, Idaho 83725, USA

Toby Ord† Future of Humanity Institute, University of Oxford, Oxford, UK (Dated: June 28, 2021) We present a model where some proportion of extraterrestrial civilizations expand uniformly over time to reach a cosmological scale. We then ask what humanity could infer if a sky survey were to find zero, one, or more such civilizations. We show how the results of this survey, in combination with an approach to anthropics called the Self Indication Assumption (SIA), would shift any prior estimates of two quantities: 1) The chance that a technological civilization like ours survives to embark on such expansion, and 2) the maximum feasible speed at which it could expand. The SIA gives pessimistic estimates for both, but survey results (even null results) can reverse some of the effect.

I. INTRODUCTION and P (v), we show how they are updated in two steps. First, merely adopting this cosmology in the context of The standard motive for studying the possible be- SIA re-weights P (q) and P (v) to lower-than-expected havior and technology of intelligent extraterrestrial life values of q and much lower-than-expected values of v. P (q) P (v) is to generate search targets for SETI. This has been Specifically, P (q) → q and P (v) → v3 , properly true for the development of “expanding cosmological normalized. civilizations” [1, 2] (aka “aggressively expanding civi- In the second step, we show how P (v) is changed again lizations,” [3] “loud aliens,” [4] etc.) — a behavioral by the results of a search. If expanding civilizations are endpoint for expansionistic civilizations in the universe, observed in a cosmological survey, v could be inferred based on conservative physics assumptions, but ambi- from their visible geometry [7]. But even if n = 0 civiliza- tious technologies. The resulting cosmology is precise tions are detected, P (v) is shifted back in the direction enough that one can use anthropic reasoning to esti- of one’s original prior estimate — the extent to which it mate the appearance rate of such expansionistic civiliza- is shifted back will depend on the certainty of the search tions [5, 6], and their visible geometry [7], as guidelines result. for future searches — a form of extragalactic SETI. Here, we take a step in a different direction, using the We organize the paper as follows: Section II discusses same basic model of expanding civilizations. We use an- the inputs to our analysis, reviewing the basic assump- thropic reasoning to estimate two additional parameters tions and cosmology of expanding cosmological civiliza- in the cosmology — the maximum practical expansion tions, assumptions about our prior pdf’s, and the Self- speed in the universe, v, and the probability that any Indication Assumption. Section III derives our initial given human-stage civilization will eventually become ex- predictions based on adopting the cosmology, while sec- pansionistic, q. These estimates have a far more personal tion IV derives the update from search results. Section meaning than most in cosmology. They can be inter- V reflects on the meaning and power of these results, and preted as predicting aspects of humanity’s own near-term summarizes our conclusions. future1. What we show is how one’s prior estimates of q and v change when adopting an anthropic school (the Self- Indication Assumption [8], or SIA) and the world model arXiv:2106.13348v1 [astro-ph.CO] 24 Jun 2021 of expanding cosmological civilizations. If one’s prior es- II. INPUTS timates are given as probability density functions P (q) The basic ingredients for this kind of analysis are a world model (depending on a set of parameters), a set ∗ [email protected] of prior assumptions about the world (manifested by a † [email protected]; This research was supported by probability density function over the parameters), and an the European Research Council (ERC) under the European anthropic school that describes how we should account Union’s Horizon 2020 research and innovation programme (grant for observer-selection effects. agreement No 669751). 1 “Near-term” in the sense that humanity will likely settle the We review these here — the “world model” is the cos- question of our own cosmic expansion, and the maximum prac- mology of expanding cosmological civilizations, and the tical spacecraft speed v on short (non-cosmological) timescale. “anthropic school” is the Self-Indication Assumption. 2

A. Expanding cosmological civilizations

If a small set of technologies are practical for an ad- vanced civilization, then a form of extreme expansionist behavior becomes possible. The essential technology is fast spacecraft that can travel between galaxies [9, 10], reproduce themselves from available resources [11], and carry out general instructions. If this technology is phys- ically attainable, then a single actor can release a wave of intergalactic settlement that expands (over cosmo- logical time) in all directions at speed v, fully saturat- ing/harnessing every galaxy along the way. The cost of such a cosmological project is just the cost of launching the first spacecraft. The reward is control of perhaps FIG. 1: A model for the cosmic time-dependence of the thousands of superclusters of galaxies. relative appearance rate of advanced life, F (t), which is If the appearance of such expansionistic civilizations effectively a model of the planet formation rate of the happens rarely and at random in the universe, this be- cosmos, with a lag of several Gyr to allow for the havior results in a clean geometry at the cosmological evolution of intelligent life. scale, where the universe is homogeneous — in fact the geometry is almost identical to nucleation and bubble growth in a cosmological phase transition [3, 12]. The the distant future [13] (strictly limiting all domains following assumptions give rise to the basic geometrical to a finite maximum size, depending on the time of picture we use: their appearance). 1. The background cosmology is standard ΛCDM — We do not assume that all technological civilizations a spatially flat Friedmann-Robertson-Walker solu- initiate such expansion. We introduce a parameter q to tion2, in which we have appeared at cosmic time represent the fraction of human-stage civilizations that t0 = 13.8 Gyr. We denote the scale factor as a(t). go on to cosmic expansion. Failure to expand could come from any cause, from voluntary abandonment of tech- 2. Appearance events are a spatially random process nology [14] to “late filter” extinction [15]. The value of in the universe (a Poisson point process), with an q only tells us the fraction of human-stage civilizations appearance rate denoted as f(t), with units of ap- that result in cosmic expansion, or in other words, q is 3 pearances per Gly of co-moving coordinate volume the probability that a random human-stage civilization per Gyr of cosmic time. goes on to expansion. 3. Each appearance event is followed by expansion in The cosmic appearance rate of expanding civilizations, all directions at constant speed v in the co-moving f(t), can be expressed as f(t) = α F (t), where α is an frame of reference, resulting in a growing domain of uncertain parameter, setting the scale of the appearance fully occupied galaxies3. The net expansion speed rate, and F (t) is a model for the relative time dependence of the domain will be close to the speed of the in- of the appearance rate of Earthlike planets (normalized tergalactic spacecraft, since the timescale for travel to a maximum value of unity), with a lag of several Gyr between galaxies dominates all other timescales in to account for the slow process of biological evolution. While the F (t) we use (displayed in figure 1) should not the process (e.g. reproduction, launch, decelera- 4 tion, etc.). This type of galaxy-to-galaxy expansion be too controversial [16] , α sets the scale and is radically can easily be modeled with a simulated distribution uncertain over many orders of magnitude. of galaxies, and it results in a growing spherical do- We can further relate α to human-stage civilizations by expressing α = γ q, where γ F (t) represents the cos- main with a sharp boundary, above the homogene- 5 ity scale [1]. mic appearance rate of human-stage civilizations . That is, the appearance rate of expanding civilizations is the 4. We assume that, once initiated, the process of cos- appearance rate of human-stage civilizations, times the mic expansion does not stop unless constrained by probability for a human-stage civilization to expand. the presence of other expanding domains, or the causally isolating effect of cosmic acceleration in 4 Substantially different models for F (t) have recently been pro- posed [4] by assuming that the process of evolving intelligent life would typically involve the interaction between multiple habit- 2 −5 We assume the following solution: ΩΛ0 = .692, Ωr0 = 9 × 10 , able systems — panspermia models of evolution. −1 5 Ωm0 = 1 − Ωr0 − ΩΛ0, H0 = .069 Gyr . A “human-stage civilization” is a still-advancing technological 3 We take units in which c = 1, and thus we express v as a number civilization that does not yet have the ability to embark on cosmic strictly less than one. expansion. 3

When an ambitious civilization appears at cosmic time B. Prior probability t0 and initiates cosmic expansion (an appearance event), it will occupy a growing sphere with co-moving coordi- Before any anthropic calculations, we should specify nate volume given by: our prior assumptions about the parameters {γ, q, v}. We will assume an independent prior pdf for each, so that P (γ, q, v) = P (γ) P (q) P (v). Z t 3 Our results in the following sections will concern how 3 0 3 4π 1 00 v V (t , t) = v 00 dt . (1) to update P (q) and P (v) using our anthropic model, for 3 0 a(t ) t any prior one might adopt. Thus, we keep these general, for now. But we do specify a particular form for P (γ)— Here, V (t0, t) is the co-moving coordinate volume of 0 it is thus an important ingredient of our final results. a sphere of light at time t that was emitted at t , and The rate of appearance of human-stage civilizations in thus multiplying by v3 gives the coordinate volume of an 6 the universe is radically uncertain over many orders of expanding civilization . We assume that all ambitious magnitude. In a prior sense, γ appearing in any order civilizations expand with a similar speed (the maximum of magnitude from 1010 to 10−100 (appearances per Gly3 practical speed), and thus we call v the dominant expan- per Gyr) is equally plausible, but even this could be criti- sion velocity, treated as another uncertain parameter. cized as too narrow, with other estimates in the literature With the assumption that ambitious civilizations ap- spanning as many as 10122 orders of magnitude [18]. pear at random and independently in the cosmos, the This degree of prior uncertainty suggests beginning fraction of all space that is not occupied by them is de- with a log-uniform prior [19], i.e. P (γ) = 1 1 . ln(γmax/γmin) γ noted by gα,v(t), given by [3]: Indeed log-uniform behavior can be seen to emerge in a related context, when compounding many sources of modest uncertainty to make estimates with the Drake −αv3s(t) gα,v(t) = e . (2) equation [20]. The endpoints of our prior, γmin and γmax, should be separated by many orders of magnitude. We t will find that if γ is sufficiently small (below about Here, s(t) = R F (t0) V (t0, t ) dt0, i.e. the integral over min 0 0 10−3), and γ is sufficiently high (above about 104), all space within our past light cone, weighted by F (t)— max pushing the endpoints further out will not effect our final for our cosmological parameters and model for F (t), we 3 results. get s(t0) ≈ 1268 Gly Gyr. The average/expected number of expanding civiliza- tions on our past light cone (and thus in principle visible C. Self-indication assumption to us) is denoted Eα,v, given by [5, 7]: The classic statement of the Self-Indication Assump- tion is that we should reason as though we are ran- 3 Eα,v = α(1 − v )s(t0). (3) domly selected from the set of all possible observers [8]. In the context of the world-model above, this means A survey that covers only a fraction frac of the sky, that if we have some prior pdf P (γ, q, v) over the pa- or for any other reason expects to detect only a fraction rameters (before anthropic arguments), the SIA im- frac of civilizations on our past light cone would thus plies that we should re-weight the prior, proportional expect to detect, on average, frac Eα,v. to the total number of observers appearing in each A cosmological expansion scenario is set by the param- parameter-universe (per unit co-moving volume, over all eter pair {α, v}, and previous calculations have focused time [21]), Nobs(γ, q, v), giving a “SIA prior” propor- on estimating α, assuming a given value of v. Since we tional to Nobs(γ, q, v)P (γ, q, v), and re-normalized. are interested here in predictions that are relevant to hu- That is, the “SIA prior” says we are more likely to live manity’s future, we will work with the parameterization in a parameter-universe {γ, q, v} that contains many ob- {γ, q, v}. All equations above expressed in terms of α can servers, before we incorporate any further knowledge. It be obtained in the new parametrization with the replace- is probably hopeless to find a good model for Nobs(γ, q, v) ment α → γ q. — the total number of observers would presumably be dominated by members of vast and long-lived civiliza- tions in the distant cosmic future, and thorny questions of the “observer reference class” are inevitable. But, the story is not complete — we should include 6 In an expanding universe, a coasting spacecraft will eventually our particular anthropic information. That is, we hu- be slowed relative to the co-moving frame [17], over cosmic time. However, since expansion will presumably generated by galaxy- mans are not just any observers, we are observers from to-galaxy hops, with the speed reset at each galaxy, we expect the a human-stage civilization that appeared at cosmic time net expansion speed v to be constant and close to the spacecraft t0 = 13.8 Gyr, in an otherwise empty corner of the cos- speed at the time of the boost [1]. mos (devoid of expanding civilizations). 4

Incorporating this anthropic information will take the we keep the information in the update general, simply form of a Bayesian “update” to the SIA prior — for now denoting it P (γ, q, v| info). According to Bayes’ theorem, this is given by

L(γ, q, v|info) Nobs(γ, q, v) P (γ, q, v) P (γ, q, v| info) = RRR 0 0 0 0 0 0 0 0 0 0 0 0 (4) L(γ , q , v |info) Nobs(γ , q , v ) P (γ , q , v ) dγ dq dv where L(γ, q, v|info) is the likelihood function, i.e. the probability that a random observer in world {γ, q, v} sees “info.” This is given by L(γ, q, v|info) = Nobs(info|γ, q, v)/Nobs(γ, q, v). Substituting this form of L(γ, q, v|info) into the previous equation, the factors of Nobs(γ, q, v) cancel out, giving:

Nobs(info|γ, q, v) P (γ, q, v) P (γ, q, v| info) = RRR 0 0 0 0 0 0 0 0 0 . (5) Nobs(info|γ , q , v )P (γ , q , v ) dγ dq dv

Unlike the SIA prior, which depended on the unknow- That is, the number of human-stage observers in this able Nobs(γ, q, v), this expression only requires a model universe is proportional to their appearance rate γ in for Nobs(info|γ, q, v), the number of observers with an- empty space, times the fraction of space (at the current thropic information like ours. This is something we can time) that has not been saturated with ambitious civi- hope to model — indeed, we have already introduced the lizations. world model to do it. This gives us the machinery to calculate posterior From equation 5, one can see the SIA now says that probabilities, i.e. the SIA updates to any chosen prior. we should favor parameter-worlds containing many ob- servers with anthropic info identical to our own. It no longer blindly favors more total observers throughout all III. PREDICTIONS PRIOR TO of spacetime, just more like us. In our case, the “an- COSMOLOGICAL SEARCHES FOR LIFE thropic info” is that we are members of a human-stage civilization, at cosmic time of arrival t0 = 13.8 Gyr, in As expressed above, Nobs(t0|γ, q, v) ∝ γ gγ,q,v(t0), i.e. a region of the universe that is not saturated by expan- the cosmic number (density) of human-stage observers sionistic life. who note they have spontaneously appeared at cosmic In the world model of the previous section, the number time t0 in unoccupied space, is proportional to the frac- of human-stage civilizations in parameter-world {γ, q, v} tion of unoccupied space at the present time, gγ,q,v(t0), who see this anthropic info (“info” denoted now by t0) is times the appearance rate of human-stage civilizations proportional to: appearing in unoccupied space, γ. Thus, by equations 5 and 6, the SIA posterior over Nobs(t0|γ, q, v) ∝ γ gγ,q,v(t0). (6) {γ, q, v}, according to this world-model, is:

γ g (t ) P (γ) P (q) P (v) P (γ, q, v|t ) = γ,q,v 0 . (7) 0 R γmax R qmax R vmax 0 0 0 0 0 0 0 γ gγ0,q0,v0 (t0)P (γ )P (q )P (v ) dγ dq dv γmin qmin vmin

3 −γ q v s(t0) Substituting our model gγ,q,v(t0) = e (equation 2) and the log-uniform prior for P (γ), then integrating over γ0, this becomes:

3 −γqv s(t0) s(t0) e P (q) P (v) P (γ, q, v|t0) = 0 03 0 03 . (8) R qmax R vmax e−γminq v s(t0)−e−γmaxq v s(t0) 0 0 0 0 0 03 P (q )P (v ) dq dv qmin vmin q v Marginalizing over γ gives:

3 3 (e−γminqv s(t0) − e−γmaxqv s(t0)) P (q) P (v) P (q, v|t0) = 0 03 0 03 . (9) 3 R qmax R vmax e−γminq v s(t0)−e−γmaxq v s(t0) 0 0 0 0 q v 0 03 P (q )P (v ) dq dv qmin vmin q v

3 3 1 −γminqv s(t0) −γmaxqv s(t0) The above SIA posterior depends on the behavior of the quantity q v3 (e − e ). There 5

0 are (at least) three cases in which it simplifies greatly, R qmax P (q ) 0 −1 with Cq = ( q q0 dq ) , and leading to a clear interpretation. min P (v) P (v|t ) = C (13) A. The “conventional SETI” limits 0 v v3

0 R vmax P (v ) 0 −1 with Cv = ( 03 dv ) . Suppose our prior assumptions are dominated by a vmin v conviction that either v → 0 or that q → 0. That is, for Equations 11, 12, and 13 are our first major conclu- whatever reason, we are absolutely sure that the maxi- sions. By adopting a world-model in which expand- mum practical speed of expansion is close zero, or that ing cosmological civilizations are possible, SIA tells us human-stage civilizations almost never embark on cosmic we should update our priors over q and v according to P (q) P (v) expansion. This would be reflected by a P (v) or a P (q) P (q) → q and P (v) → v3 , properly normalized. with support arbitrarily close to zero. SIA wants to tilt our assumptions to a lower-than 3 1 −γminqv s(t0) In either limit, the value of q v3 (e − expected probability that human-stage civilizations ad- 3 −γmaxqv s(t0) vance to cosmic expansion (lower q), and with an expan- e ) approaches (γmax − γmin) s(t0). And in that case, equation 9 reduces to: sion speed slower than we first supposed (lower v). Why? It allows for more observers with anthropic information identical to our own. Keeping q down (the probability P (q, v|t0) = P (q) P (v). (10) to expand) means the appearance rate of civilizations like ours can be high, while leaving plenty of free space That is, our anthropic information makes no change to in which to appear. Keeping v low means the universe 7 our prior assumptions about q and v. saturates with ambitious life more slowly, leaving more This can be regarded as a basic consistency check on opportunity for spontaneous appearance of human-stage our framework. Assuming from the beginning, with com- life here at t0. plete certainty, that expanding civilizations are irrelevant These predictions do not depend explicitly on the de- (being either too slow or too rare to matter) ensures they tails of the background cosmology, or even on our partic- play no role in observer-selection effects. ular appearance time, as s(t0) falls out of the calculation in equation 11. This is a consequence of the domain of γ we assumed B. The regime of expanding civilizations in our approximation, but it is not very restrictive. The 1 requirement of γmin << means, for our back- If we stay away from the previous limits, so the support s(t0) ground cosmology, that γ << 10−3 (appearances of of both P (v) and P (q) is away from zero, then we can min human-stage civilizations per Gly3 per Gyr), which is make another simplification. Suppose we take γmin to 1 by no means outside the realm of possibility — human- be tiny, in particular γmin << . And we also take s(t0) stage life could be a cosmic rarity. The requirement of γ to be large, so that γ >> 1 . Then 1 −4 max max q v3 s(t ) γmax >> 3 , if vmin ≈ 0.1 and qmin ≈ 10 , min min 0 qminv s(t0) 3 3 min −γminqv s(t0) −γmaxqv s(t0) 4 we can approximate (e − e ) ≈ 1 would mean γmax >> 10 (appearances of human-stage over the entire domain of interest, and equation 9 reduces civilizations per Gly3 per Gyr), which is also quite a to: pedestrian number8. Finally, we note that if q (the probability for a human- P (q) P (v) stage civilization to expand) is further decomposed into P (q, v|t ) = C (11) n independent probabilities, so that q = q q ··· q , with 0 q,v q v3 1 2 n priors P (q) = P (q1) P (q2) ··· P (qn), then the SIA update 0 0 P (q ) R vmax R qmax P (q ) P (v ) 0 0 −1 is the same for each probability, giving P (q ) → 1 , where Cq,v = ( 0 03 dq dv ) is a nor- 1 q vmin qmin q v 1 P (q2) P (qn) malizing constant. P (q2) → , P (qn) → , etc. q2 qn Marginalizing, this gives: For example, in a two-step decomposition, q1 could represent the probability that a human-stage civilization P (q) survives until the necessary technologies become avail- P (q|t ) = C (12) 0 q q able, and q2 could represent the probability to expand, given survival. The SIA result tells us to become more pessimistic about both q1 and q2. We must therefore con- clude that estimates of all independent sources of near- 7 Conceptual questions can emerge in these limits. What is the interpretation of q in the limit that v → 0? Presumably, q would be the probability for a human-stage civ to become expansionistic at a subgalactic scale, or even the scale of a single solar system. Effectively, this limit becomes the world-model of 20th century 8 The Laniakea Supercluster alone contains approximately 105 SETI and science fiction. galaxies, stretched over a distance of about 0.5 Gly. 6 term human existential risk are likewise amplified, pro- panding domains [7], and thus get an excellent idea of vided they are universal risks, i.e. risks that all human- the practical speed limit we are likely to encounter in the stage civilizations must face on the road to technological future. But suppose the survey detects n = 0 such civ- maturity. ilizations. This too will update our assumptions about v. Since appearances are a Poisson process (random, IV. UPDATES BASED ON SEARCH RESULTS independent events), with expected number Eγ,q,v = 3 γq(1 − v )s(t0) on our past light cone (and frac Eγ,q,v SIA has shifted our prior assumptions about q and v, expected to be visible to our survey), the probability to but we can go a step further — suppose we perform an see n = 0 is given by: exhaustive galaxy survey, covering fraction frac of the 3 −frac Eγ,q,v −frac γq(1−v )s(t0) sky (or for any reason expecting to detect fraction frac pγ,q,v(n = 0) = e = e (14). of ambitious civilizations on our past light cone), looking for expanding civilizations. Now, assuming a survey detects n = 0 civilization, we If one or more expanding civilizations were detected, can invoke Bayes’ theorem to update our previous SIA we could infer v from the visible geometry of their ex- estimate from equation 8:

pγ,q,v(n = 0)P (γ, q, v|t0) P (γ, q, v|t0, n = 0) = RRR 0 0 0 0 0 0 (15) pγ0,q0,v0 (n = 0)P (γ , q , v |t0) dγ dq dv 3 3 s(t ) e−qγs(t0)(v +frac(1−v )) P (q) P (v) = 0 . (16) −qγ s(t )(v3+frac(1−v3)) −qγ s(t )(v3+frac(1−v3)) RR (e min 0 −e max 0 ) P (q0) P (v0) 0 0 q0 (v03+frac(1−v03)) dq dv

If we take the same condition on γ as that preceding The effect of confirming exactly n = 0 expanding civi- equation 11, and marginalize, we obtain the following lizations exist on our past light cone is to perfectly cancel 1 estimates: our earlier SIA-induced shift of v3 . That is, our estimate of v goes back to what it was in the very beginning, be- P (q) fore we had invoked any anthropic arguments at all. An P (q|t , n = 0) = P (q|t ) = C . (17) ideal search cancels the SIA effect, where v is concerned. 0 0 q q How can seeing n = 0 civilizations have such a pow- That is, a survey resulting in n = 0 detections does not erful effect? Although we began with independent, un- update our previous SIA-based estimate of q. However, correlated prior assumptions about v and γ, P (v) and it does update our knowledge of v: P (γ), they became highly correlated in the SIA posterior, equation 8. The predicted lower values of v are associ- P (v) ated with higher values of the appearance rate, γ. Thus, P (v|t , n = 0) = C . (18) 0 surv v3 + frac (1 − v3) the first SIA result (reducing our estimates for v) has the effect of greatly amplifying the number that should be 0 R P (v ) 0 −1 visible, on our past light cone. Seeing n = 0 is then a where Csurv = ( v03+frac (1−v03) dv ) . This is our sec- ond major result. surprise, and we get a large reversal to our estimate for Two limiting cases are of immediate interest. In the v. limit that frac → 0, i.e. when our survey never had any As noted above, if n > 0 civilizations are detected, we chance of detecting anything, then we retain our earlier could infer v directly, but what happens to our knowledge result (equation 13) — a useless survey changes nothing: of q? In that case, due to the Poisson property, we have:

P (v) lim P (v|t0, n = 0) = Cv 3 . (19) frac→0 v n (frac Eγ,q,v) p (n) = e−frac Eγ,q,v . (21) However, in the limit that frac → 1, i.e. when our γ,q,v n! survey was guaranteed to detect every expanding civi- lization on our past light cone (but still sees none), then we get: Inserting into equation 15 and marginalizing over γ yields the same result as before for q, reproducing equa- tion 17 (in the high-γmax, low-γmin approximation). Sur- lim P (v|t0, n = 0) = P (v). (20) frac→1 vey results do not update q, in this limit. 7

V. CONSEQUENCES AND CONCLUSIONS of v, the results of a sky survey do not reverse the shift to lower estimates of q. Second, while a low value of q could have many causes (some hopefully benign [14]), The SIA re-weighting of prior assumptions can be the conclusion of section IIIB shows that all independent, shockingly powerful. As an illustration, suppose you at- universal sources of existential risk [22] are unavoidably tended two lectures, back-to-back, by two equally con- amplified by our result as well, making late filter extinc- vincing experts — the first discussing insurmountable tion [15] more probable than before. difficulties of relativistic space flight, and the second dis- One important question we have not attempted to an- cussing the resources available to a technologically ma- swer here is the relevancy of current astronomical data. ture civilization. We have, in fact, looked at the sky and detected n = 0 Emerging from the lecture hall, you find yourself con- expanding cosmological civilizations. But what fraction flicted, splitting your credence equally between two views frac of expanding civilizations on our past light cone of the universe. One in which the maximum practi- should current astronomical data have been expected to cal expansion speed of an ambitious civilization is about notice? Finding an appropriate answer is key to inter- v = .1, and in the other view somewhere around v = .9. preting the results of section IV, in light of current and Then, invoking the SIA together with the possibility of future data. expanding civilizations implies you should be 99.9% con- Finally, we note this is not the first time alien life vinced of the low-speed scenario (because it leaves room has been combined with the SIA to adjust our intuition for ≈ 1, 000 times as many human-stage civilizations to for humanity’s survival. Grace noted [23, 24] that the have originated at the present time). SIA supports a Milky Way Galaxy that is packed with That is, until you return to your day job of analyzing human-stage life (i.e. with anthropic info like ours), but millions of unsurprising galaxy spectra. If your survey combined with negative SETI results, this implies a late data is sufficiently deep and detailed, and it covers, say, filter awaiting human-stage civilizations. This is based 30% of the sky, you might return to the conclusion that on a different world model than we have used, but it is the low-speed scenario is only about a 73% favorite after interesting to note how general such conclusions can be. all. ACKNOWLEDGMENTS The lowering of estimates for q (the probability for human-stage civilizations to advance to cosmic expan- We are grateful to Milan Cirkovi´c,for´ his helpful com- sion) is troubling for two reasons. First, unlike estimates ments.

[1] S Jay Olson, “Long-term implications of observing an ex- and its relevance to setl,” Journal of the British Inter- panding cosmological civilization,” International Journal planetary Society , 491 (1988). of Astrobiology 17, 87–95 (2018). [10] Stuart Armstrong and Anders Sandberg, “Eternity in [2] S. Jay Olson, “Expanding cosmological civilizations on six hours: Intergalactic spreading of intelligent life and the back of an envelope,” (2018), arXiv:1805.06329 sharpening the fermi paradox,” Acta Astronautica 89, [astro-ph.CO]. 1–13 (2013). [3] S Jay Olson, “Homogeneous cosmology with aggressively [11] Robert A Freitas Jr, “A self-reproducing interstellar expanding civilizations,” Classical and Quantum Gravity probe,” Journal of the British Interplanetary Society 33, 32, 215025 (2015), arXiv:1411.4359. 251–264 (1980). [4] Robin Hanson, Daniel Martin, Calvin McCarter, and [12] Alan H Guth and Erick J Weinberg, “Cosmological con- Jonathan Paulson, “If loud aliens explain human earli- sequences of a first-order phase transition in the s u 5 ness, quiet aliens are also rare,” (2021), 2102.01522 [q- grand unified model,” Physical Review D 23, 876 (1981). bio.OT]. [13] Toby Ord, “The edges of our universe,” (2021), [5] S. Jay Olson, “Estimates for the number of visible galaxy- arXiv:2104.01191 [gr-qc]. spanning civilizations and the cosmological expansion of [14] Bill Joy, “Why the future doesn’t need us,” Nanoethics– life,” International Journal of Astrobiology 16, 176–184 the ethical and social implicatons of nanotechnology , (2017). 17–39 (2000). [6] S. Jay Olson, “On the likelihood of observing extragalac- [15] Robin Hanson, “The great filter-are we almost past it?” tic civilizations: Predictions from the self-indication as- preprint available at http://hanson. gmu. edu/greatfilter. sumption,” (2020), 2002.08194 [astro-ph.CO]. html (1998). [7] S Jay Olson, “On the visible size and geometry of aggres- [16] Abraham Loeb, Rafael A Batista, and David Sloan, sively expanding civilizations at cosmological distances,” “Relative likelihood for life as a function of cosmic time,” Journal of Cosmology and Astroparticle Physics 2016, Journal of Cosmology and Astroparticle Physics 2016, 021 (2016). 040 (2016). [8] Nick Bostrom et al., Anthropic bias: Observation selec- [17] Frank J Tipler, “Ultrarelativistic rockets & the ultimate tion effects in science and philosophy (Routledge, 2002). future of the universe,” in NASA CONFERENCE PUB- [9] Martyn Fogg, “Feasibility of intergalactic colonisation LICATION (NASA, 1999) pp. 111–120. 8

[18] Brian C. Lacki, “The log log prior for the frequency of [21] Nick Bostrom and Milan M Cirkovi´c,“The´ doomsday extraterrestrial intelligences,” (2016), arXiv:1609.05931 argument and the self-indication assumption: reply to [astro-ph.EP]. olum,” The Philosophical Quarterly 53, 83–91 (2003). [19] M. Tegmark, Our Mathematical Universe: My Quest for [22] Toby Ord, The precipice: existential risk and the future the Ultimate Nature of Reality (Knopf Doubleday Pub- of humanity (Hachette Books, 2020). lishing Group, 2014). [23] Caitlin Grace, “Anthropic reasoning in the great filter,” [20] Anders Sandberg, Eric Drexler, and Toby Ord, “Dis- Unpublished manuscript (2010). solving the fermi paradox,” (2018), arXiv:1806.02404 [24] Katja Grace, “Sia doomsday: The filter is ahead,” blog [physics.pop-ph]. available at https://meteuphoric.com/2010/03/23/sia- doomsday-the-filter-is-ahead/ (2010).