Huge DarknessSam T. Oates

Stonefish civilisations

Have we actually looked for aliens?

Falling Star (1884), Witold Pruszkowski

The universe seems empty. Modern telescopes can peer into the depths of our galaxy, survey millions of its fellows, and image the beginning of time. But we see no clear sign of peers anywhere.

Would we expect an interstellar civilisation to give off distinctive signatures we could see right now? How much can we say about the presence of such signatures, given our current instruments? Would civilisations shine like beacons in the void, or would they be more like a stonefish, hiding in plain sight by blending into the background?

NeuroDojo: April 2011
There is a stonefish in this picture.

These are important questions. One of the ultimate goals of the AI industry is expansion into the rest of the universe: consider how much more credible this narrative would be if we could see evidence of interstellar traffic. Conversely, this is challenged if we can rule out basic elements of other interstellar civilisations.

Ideas like the Grabby Aliens hypothesis are also premised on the idea that expanding interstellar civilisations would be somehow distinctive (“clearly change the look of their volumes” in the words of the flagship 2021 paper).

The best recent estimates for the Drake equation’s parameters like this one also exhibit uncertainty around observability. fc, the fraction of intelligent civilisations that are detectable or contactable, lacks good estimates of bounded uncertainty. In the paper’s own words: “…there are no clear arguments for the range of uncertainty of fc: the range of estimates in the literature from 10-2 to 1 give LU[fc] = 2 [LU = Log-Uncertainty, the number of OOMs covered by current uncertainty], but it is clear a broader range is intended.”

It seems like there are two basic expectations we would have of such an interstellar civilisation:

  1. They would use a lot of energy.

  2. They would send things between stars sometimes.

So, could we see these?

Energy use

The popular picture goes something like this: civilisations grab as many stars as they can, surround them with Dyson swarms, do computation with the energy, and emit waste heat as infrared radiation at ~100-300 Kelvin.

Wouldn’t you want to emit (T) at close to zero Kelvin - waste not, want not? Firstly, radiator mass M scales with T-4. Secondly, the Landauer limit means we’re limited to L/(kT ln(2)) irreversible operations per second, where L is luminosity, so throughput is limited to roughly M1/4: doubling radiator mass only gets us about 20% more compute, which is not very attractive compared to investing in expansion. Thirdly, we need to leave some mass aside for the actual computing infrastructure.

Assuming an even split between computer and radiator mass, the roughly achievable emitting temperatures are as follows:

  1. Using a single rocky planet like Mercury: 100 Kelvin

  2. Using all of our rocky planets: 40K

  3. Using all of our planets: 15K

You can probably get close to the CMB floor of 2.7K by using stellar husbandry to mine metals directly from the star. This has the benefit of greatly extending its life at the cost of some dimming. I think the practical limit is around 4K for a star like the Sun due to available mass. Imagine a halo of opalescent gossamer the size of the Oort Cloud, its tips hanging in perfect clarity and coldness, thinking with the parallel inexorability of a glacier…

So we can expect a range of emitting temperatures. I would expect a new project to range from the high hundreds to 100K as it encloses its star and then progress toward the CMB as it matures. Rough timescales would be a century to disassemble the system, going from 100K to 15K, and a millennium to complete the stellar lifting process and push all the way down to 4K. For a mature interstellar civilisation tapping the energy of hundreds of stars, I would expect structures that are hard to distinguish from the CMB; stellar husbandry makes sense for other reasons like lifespan and efficiency, and there’s not much else to do with the mass you get from it.

All this is to say that we have done explicit Dyson sphere surveys, but we’d only expect these to capture new projects or young civilisations that have not yet reached maturity in stellar exploitation, because these surveys focus around the 100K range. This is not because of a preference on behalf of the researchers. WISE, the best infrared observatory for these purposes, hits a floor around 100K at the kiloparsec distance. We have other instruments and data archives;1 these either haven’t received much attention or have their own limitations that make them unsuitable.

For example, we have looked at our local chunk of the galaxy (this WISE survey checked out to 5 kiloparsecs, about a fifth of the Milky Way’s diameter) and… it’s not like we find zero candidates, but only about one in every ten thousand stars is compatible with a near-complete Dyson enclosure. Our instruments limit us to that 5 kpc area, and the survey’s conclusions focus on a smaller, higher-confidence 100 pc area with only around 26,000 stars - a small sample size.

Still, with hundreds of billions of stars in the Milky Way this is hardly a showstopper, and we have only properly surveyed our neighbourhood.

At the other end of the spectrum, this search of the Planck space telescope archive looks for enclosed galaxies radiating 1K above the CMB and rules out anything within 700MPc (Andromeda is 0.8 MPc away). We also know that no one is turning galaxies into mid-infrared emitters around 100 Kelvin.

This interesting paper from August 2026 lays out a research program for finding cold Dyson swarm candidates using archival data. It terms my gossamer haloes ‘Slysh haloes’ after the first advocate for cold Dyson swarm searches, and comes to many of the same conclusions I do above. They have yet to produce their headline results, but this is definitely something to watch.

However, a mature civilisation would probably not limit itself to Dyson swarms around main sequence stars.2 For instance, focusing on black holes makes sense; they’re attractive for their higher energy and compact size (which means low latency, which is on the order of hours for an AU-sized computer).

It seems our current observations are inconclusive. We can rule out entire galaxies being exploited, but not less developed interstellar civilisations.

Progress could come from several areas. A few examples: investigation of archives like the above,3 discovering some reason to expect that computation isn’t that important,4 or a good reason to emit at higher temperatures.

Travel

We can expect interstellar travel to be dominated by beams, antimatter, and fusion.

Beams are things like lasers or particle beams, which are great because you don’t have to carry your fuel with you. But beams aren’t so good for braking - you need someone at the other end to slow down. And it’s expensive to get above 0.5c.5

Antimatter engines solve these problems. But antimatter is very finicky and expensive to produce; it’s a battery and not a fuel, as the saying goes. For lower speed travel up to around 0.15c, fusion engines seem superior because they use fuel that’s mined and not manufactured, don’t have fuel containment problems, and don’t emit crazy amounts of gamma radiation. However, let’s focus on antimatter because it’s clearly better if you need braking or want to approach God’s speed limit.

Would we notice any of this?

Well, we’d obviously notice (briefly) if a stray fuel pellet hit us at 0.4c. But the main problem is that we’d need to be right in the axis of the beam or exhaust, or the emission would be basically invisible. But we could probably see a beam pointed in our general direction: lasers are an open area in technosignature research, a propulsion laser could temporarily be the brightest object in its system, and antimatter engines would leave radiation signatures.

However, received radiation reduces with the square of distance and there are huge numbers of other natural radiation sources like pulsars and accretion jets that need to be distinguished from drive signatures. SETI work in this area remains high-level and theoretical. Current work does not rule out extensive use of antimatter or laser propulsion, but it does put bounds on what we can expect - for instance, we can rule out huge amounts of nearby traffic or a galactic network extensive enough to rival natural energy sources in the aggregate. Nearby antimatter engines that happened to be pointed at Earth or be ‘leaky’ would be particularly distinctive, as natural gamma sources tend to be associated with stars.

Once again, from a high-level survey it seems as though we cannot draw strong conclusions with current evidence, but we can say that naive scenarios are not something we observe.

Travel signatures are naturally hard to detect and we lack the sensing capability to rule out interstellar traffic.

For reasons beyond the scope of this post I don’t expect to see any peers around us. The universe is still in its first blush of youth, and there is even more uncertainty around the emergence of life than in what we’ve just covered. But it is always good to sanity-check one’s positions with what ground truth is available, even if that process only serves to show you how little ground exists at all.

These two narrow points illuminate a broader one. I have always been averse to overconfidence in reasoning about the deep future. The power of abstract reasoning should not be disrespected, but I feel the skill of astrophysicists in teasing conclusions from low-resolution data can inculcate a sense that we know more about our universe than we really do. To a large extent we simply have not looked. And so blog posts about how YouTube Shorts are the Great Filter or wedding vows about purchasing galaxies seem premature.

The further one gets from the tiny shell of existence we have categorised, the more leviathans of the deep rear their heads, the things undreamed of in our philosophy. A foundation might be of the sturdiest design, but that matters little for a castle built on sand.

  1. Personally I think there’s interesting work to be done on the Herschel Hi-GAL archive and the Planck Galactic Cold Clumps catalogue.

    ↩
  2. Remember that the original Dyson sphere paper was written as a satire of SETI, not a rigorous look at the best possible configuration for an interstellar civilisation.

    ↩
  3. Exercise for the reader with API limits to burn: Point Claude at some of these archives over a free weekend.

    ↩
  4. For instance, dualism.

    ↩
  5. Lasers have issues with diffraction, charged particle beams get deflected by the galactic field, neutral particle beams get stripped to nothing by collisions with the interstellar medium, and ‘mass beams’ - pellet launchers - have their advantages eroded as the pellets have to be launched faster and faster.

    ↩