Who is asking, and how it works

Searches 129 nearby galaxies for the waste heat that swarms of starlight collectors would shed, fitting each galaxy's light from ultraviolet to infrared with and without such a swarm; tests with 1,419 injected fake swarms show the method finds swarms taking as little as 4 to 5 percent of the starlight in old, quiet galaxies. None of the 129 galaxies prefers a swarm. The per-galaxy 95 percent upper limit is under 0.3 percent of the starlight for the median quiet galaxy without a bright active nucleus, and the zero detections bound the population: fewer than 2.6 percent of such galaxies can host swarms taking 25 percent of their starlight, and fewer than 14.6 percent swarms at 300 kelvin taking 5 percent (each at 95 percent confidence). The authors say the dusty and active galaxies cannot be separated from waste heat by survey colours alone, and propose cheaper archival screens and a future far-infrared mission to follow up.

Olivia Curtis, Aidan J. Rowland, Jason T. Wright, Caryl Gronwall, Jakob M. Helton, Joel Leja. arXiv:2608.12458 (astro-ph.GA, with astro-ph.SR), 34 pages, 16 figures, 4 tables; submitted to The Astrophysical Journal (preprint version 1, 2026-08-12). Read it.

Bears on: The share of intelligent species that make themselves detectable; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect; Aestivation: advanced civilisations are asleep, waiting for the universe to cool so that their thinking will go further on the same energy; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by hand on 2026-10-18, not by a reading: the audit of the 2026-10-17 reading found it missing from the log. Read on arXiv's own page on 2026-10-18 (abstract, discussion, conclusions). The title is as arXiv lists it; the paper's own title line drops the word Infrared. A limit on other galaxies' waste heat, not on any of the seven factors: it bounds the product of all of them and moves no range on its own (the detectable record's test says so).

The record is data/papers/curtis-2026.toml in the public repository.

James Webb Space Telescope infrared images and spectra of two of the seven red dwarf stars that the 2024 Hephaistos search flagged as possible Dyson-sphere hosts (candidates D and E) show that the infrared excess comes from background galaxies about one arcsecond away, not from anything at the stars: candidate D's is at redshift about 0.9 and looks like a hot-dust-obscured galaxy, candidate E's at redshift about 0.4 and looks like a dusty starburst. The authors find no excess from regions a few astronomical units from either star, which rules out both a Dyson sphere and an extreme debris disk there. They conclude that the 2024 estimate of how often such a chance alignment would happen was too low, and that future searches for Dyson spheres by mid-infrared excess need sharper images than the all-sky survey provides.

Erik Zackrisson, Arjan Bik, Anita Ali Asgar, Olivia Curtis, Jason T. Wright, Tongtian Ren, Roberto J. Assef, Andrew Blain, Alexis Brandeker, Michael A. Garrett, Uma Gorti, Andreas J. Korn, Priyatam K. Mahto, Armin Nabizadeh. arXiv:2607.09460 (astro-ph.GA), 13 pages, 9 figures; a preprint (version 1, 2026-07-10), not yet found in a journal. Read it.

Bears on: The share of intelligent species that make themselves detectable; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by hand on 2026-10-18, not by a reading: the audit of the 2026-10-17 reading found that the log carried the follow-up that mentions this result (Korn and others, 2026) but not this paper. Read on arXiv's own page on 2026-10-18 (abstract, introduction, sections 5 and 6). It removes two candidates; it gives no number for any of the seven factors, so no range is argued for.

The record is data/papers/zackrisson-2026.toml in the public repository.

A review of what the Kepler mission and other surveys show about how planets are spread among stars, concentrating on orbits inside about the Earth-Sun distance, and of what those statistics say about how planets form and move. Neither its abstract nor its full text concludes that the share of stars with planets is close to 100 percent. Its summary says about 30 percent of Sun-like stars host planets inside about the Earth-Sun distance (down to Earth's mass and size), averaging about three such planets each; it also warns that the average number of planets per star and the share of stars with planets are different quantities. It is logged without an estimate.

Wei Zhu, Subo Dong. Annual Review of Astronomy and Astrophysics 59, 291-336 (2021); arXiv:2103.02127. Read it.

Bears on: The share of stars that have planets.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 6 (I accepted in part: logged as a publication, added as an estimate if the full text gives the figure). Authors, title, venue, pages and date checked on 2026-10-17 against the arXiv record and Crossref. Read in full 2026-10-17 (arXiv HTML version of 2103.02127): the outside reviewer's statement that the review concludes the share of stars with planets is close to 100 percent is not in the paper and is not made in this record. The paper's first summary point gives about 30 percent of Sun-like stars with inner planets (inside about 1 AU, down to Earth mass), its section 3.1 gives about 10 percent with cold giant planets, and its section 1.1 explains why a planets-per-star figure is not the share of stars with planets. A figure for part of the parameter space is not an estimate of this factor, so none is added. Flagged for me in the changelog: whether the 30 percent belongs in the factor's notes.

The record is data/papers/zhu-2021.toml in the public repository.

Runs the Drake Equation as the 2018 Oxford paper did, with a range for each factor, and splits the detectable factor in two: the chance that an intelligent biological species can communicate across space (taken as that paper's 0.01 to 1) and the chance that it goes on to build a machine intelligence that can (taken as 0.0001 to 1, by assuming the extra step is as selective as the first, so the chance is the first one squared). Finds that detectable machine intelligences are about as numerous as detectable biological ones, and that machines dominate the universe as a whole.

Alex De Visscher. International Journal of Astrobiology 19(5), 353-359 (2020); arXiv:2001.11644. Read it.

Bears on: The share of intelligent species that make themselves detectable; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 16 (I accepted). The reviewer gave the arXiv number as 2001.11822, which is a paper on a different subject; the right number is 2001.11644. The 0.0001 to 1 in the paper is the range for the machine route, an assumption, and not an estimate of the share of intelligent species that become detectable, which for biological species the paper takes as 0.01 to 1; so it did not meet the condition I attached to item 15 (see the detectable factor's notes). Venue, volume, pages and the two ranges checked on 2026-10-17 against the arXiv full text and Crossref.

The record is data/papers/de-visscher-2020.toml in the public repository.

Rebuilds the history of star formation in a region of about 2,000 parsecs around the Sun from the European space telescope Gaia's measurements, and finds three bursts of star formation, about 5.7, 1.9 and 1.0 billion years ago, which the authors tie to the Sagittarius dwarf galaxy passing through the Milky Way's disc. It is a history of the region round the Sun, not a rate in stars a year for the whole galaxy, so it gives no number for the star-formation estimate list.

Tomás Ruiz-Lara, Carme Gallart, Edouard J. Bernard, Santi Cassisi. Nature Astronomy 4, 965-973 (2020). Read it.

Bears on: How many stars are born in the galaxy each year.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 3 (I accepted in part: logged as a publication, added as an estimate only if the full text gives a galaxy-wide rate). Authors, title, venue, pages and the three burst ages checked on 2026-10-17 against the Crossref record and the paper's abstract as carried by the Instituto de Astrofisica de Canarias page. Read in full 2026-10-17 (arXiv HTML version of 2003.12577): the paper's figures plot a normalised star formation rate against lookback time and give no absolute rate, in solar masses or in stars a year, for the whole galaxy or for the region; the bursts are dated 5.7, 1.9 and 1.0 billion years ago with durations of 0.8, 0.2 and 0.1 billion years and decreasing strength, and the phrase 'four times' appears nowhere in it. The outside reviewer's remark is therefore not supported and is not made in this record; no estimate is added.

The record is data/papers/ruiz-lara-2020.toml in the public repository.

Asks how many copying molecules random chemistry could assemble among the stars of the observable universe (about ten to the 22) and finds chains of about 32 units, against the 40 to 100 that a copying molecule needs, so life would not arise in the observable universe by chance alone. His way out is the far larger universe that inflation implies, with more than ten to the 100 stars, where it would. Like Koonin's, a bound on one mechanism and a count of stars needed, not a chance per planet.

Tomonori Totani. Scientific Reports 10, 1671 (2020); arXiv:1911.08092. Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 10 (I accepted in part: a publication record linked from the chance-life-starts factor, kept in that factor's notes and not in its estimate list, because the paper gives a count of stars needed, not a chance). Venue, volume, article number and date checked on 2026-10-17 against Crossref and the arXiv record; the figures are as the factor's notes already described them after the check of 2026-10-16.

The record is data/papers/totani-2020.toml in the public repository.

Combines the Kepler mission's final planet catalogue with Gaia's measurements of the stars and finds, for planets of 0.75 to 1.5 Earth radii on orbits of 237 to 500 days around Sun-like stars, a rate of less than 0.27 planets per star (the 84th percentile), which could be lower by a factor of about two because some candidates are false alarms. For planning missions to find rocky planets in or near the habitable zone, the authors recommend allowing for a true rate anywhere from 0.03 to 0.40.

Danley C. Hsu, Eric B. Ford, Darin Ragozzine, Keir Ashby. The Astronomical Journal 158, 109 (2019); arXiv:1902.01417. Read it.

Bears on: Planets per star that could support life.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 8 (I accepted). The reviewer's 'about 0.2' is a fair summary but not the paper's statement; the figures above are the abstract's own, checked on 2026-10-17 against the arXiv record.

The record is data/papers/hsu-2019.toml in the public repository.

Shostak's chapter on the detectable factor from 1961 to the present, which the abstract says considers estimates of the share of intelligent species that can make themselves visible to other societies, using the recent capabilities of terrestrial technology. The chapter's own figure has not been read, so it is logged without an estimate.

Seth Shostak. chapter 12 in Douglas A. Vakoch and Matthew F. Dowd (eds.), The Drake Equation: Estimating the Prevalence of Extraterrestrial Life through the Ages (Cambridge University Press, 2015), pp. 227-240. Read it.

Bears on: The share of intelligent species that make themselves detectable.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 16 (I accepted once the figure is read). That chapter 12 is Shostak's, on the detectable factor, and its title and pages were checked on 2026-10-17 against the book's contents page and the Crossref record. Searched 2026-10-17 for the chapter's text: the Cambridge Core chapter page (https://doi.org/10.1017/cbo9781139683593.015) shows the abstract only, the Google Books interface refused the query and the Internet Archive holds no copy, so no open source gives the chapter's number. The outside reviewer's statement that it gives 1 is not made in this record, which logs the chapter without an estimate. The abstract adds that the chapter's value of fc matters for SETI experiments because radio or other technology is hardly inevitable even where intelligence is present.

The record is data/papers/shostak-2015.toml in the public repository.

Estimates how unlikely it is that a system able to copy itself and translate its code arises by chance, and puts the chance in one observable region of the universe below one in ten to the 1,018, a deliberately generous upper bound. His conclusion is not that life is absent but that, if the universe is infinite and keeps inflating, such a chance event happens somewhere, so life needs a far larger universe than the one we can see. A bound on one mechanism, not a chance of life starting per planet.

Eugene V. Koonin. Biology Direct 2, 15 (2007); arXiv:q-bio/0701023. Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged 2026-10-17 on the outside review's item 10 (I accepted in part: a publication record linked from the chance-life-starts factor, kept in that factor's notes and not in its estimate list, because a chance of one in ten to the 1,018 is not a number the computer can hold). Venue, volume, article number and date checked on 2026-10-17 against Crossref; the figure is as the factor's notes already described it after the check of 2026-10-16.

The record is data/papers/koonin-2007.toml in the public repository.

Argues that a civilisation that does mostly computing would shed its waste heat as cold as it can, far below the 100 to 600 kelvin that searches for Dyson spheres look for, because the cost of a computation falls with temperature; a cold structure needs a vast radiating area, so it would be conspicuous at submillimetre wavelengths rather than hidden. Existing far-infrared surveys could already bound such cold waste heat at about ten to the twenty watts around nearby stars, several orders of magnitude below past infrared limits, and red dwarfs are the best targets. The paper also notes such civilisations may simply be asleep instead.

Michael Garrett. arXiv:2608.31153 (astro-ph.IM), 14 pages; submitted to Monthly Notices of the Royal Astronomical Society. Read it.

Bears on: The share of intelligent species that make themselves detectable; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect; Aestivation: advanced civilisations are asleep, waiting for the universe to cool so that their thinking will go further on the same energy; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one; We are listening the wrong way: an efficient message is indistinguishable from noise, and they may not be using radio at all; The dark forest: everyone is hiding, because announcing yourself gets you killed.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/garrett-2026.toml in the public repository.

Follows up the seven red dwarf stars within 300 parsecs that the 2024 Hephaistos search flagged for unexplained infrared excess, plus three similar objects. Their stellar properties are ordinary and the excess still has no clear explanation; it resembles the glow of a disk around a young star, but the stars show no signs of youth. For two of the seven, new JWST images show a very red background galaxy sitting behind the star, which explains their excess without any structure around the star. No candidate is confirmed as artificial.

Andreas J. Korn, Matías Suazo, Erik Zackrisson, Pía Cortés-Zuleta, Adam D. Rains, Armin Nabizadeh. arXiv:2607.25701 (astro-ph.SR, with astro-ph.GA), 11 pages. Read it.

Bears on: The share of intelligent species that make themselves detectable; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/korn-2026.toml in the public repository.

Proposes that once a civilisation's industry runs itself off its planet, spreading between stars becomes too cheap and useful to refuse, but it spreads quietly: small probes, robotic seed factories and slow replication rather than star-harvesting empires. One such civilisation would saturate its reachable neighbourhood within about ten million years, so the fact that none has reached us means none arose within reach. The traces to look for are weak artefacts and local probes, not galaxy-scale energy use. A preprint, not yet reviewed.

Sergey Ivliev. arXiv:2606.13914 (astro-ph.IM, with physics.soc-ph), 28 pages; submitted to Acta Astronautica. Read it.

Bears on: No interest in contact: they exist, they could reach us, and they simply do not care to; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/ivliev-2026.toml in the public repository.

Reconstructs the gene set of the last common ancestor of complex cells, comparing 186 genomes of complex cells with more than 65,000 bacterial and archaeal genomes and 1.3 million groups of viral genes, and traces where each gene family came from. Besides the two accepted partners, the archaeal host's lineage (Asgard archaea, 3.3% of the ancestral proteins on average across three datasets) and the ancestor of mitochondria (Alphaproteobacteria, 4.0%), it finds two more bacterial sources passing the same tests: Myxococcota (3.9%) and Planctomycetota (2.2%). About 4.5% of the gene families arrived by way of viruses, mostly giant viruses. Relative timing from the gene trees, with no dates, puts the Asgard genes oldest, then Planctomycetota, with the mitochondrion's ancestor and Myxococcota later and overlapping, so some transfers came before the mitochondrion. The authors read the origin of complex cells as a gradual process with several partners in a microbial community rather than a one-off merger of two cells, and say their data cannot tell which partner was the host or whether any lived inside another.

Moisès Bernabeu, Saioa Manzano-Morales, Marina Marcet-Houben, Toni Gabaldón. Nature 656, 399–405 (2026). Read it.

Bears on: The share of planets with life on which it becomes intelligent; Complex cells are rare: life may start easily but then stay simple forever, because the one step that made big cells happened only once.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Read in full 2026-10-20 against the finding (Nature 656, 399-405, issue of 2026-08-13; published online 2026-06-10). The percentages are borne out: 3.92% Myxococcota, 2.24% Planctomycetota, 3.99% Alphaproteobacteria and 3.28% Asgard archaea, each an average across the paper's three datasets. Corrected: (1) 'from thousands of genomes' now gives the counts; (2) 'acquired late in a protracted process' now says what the paper shows, relative order without dates, with the mitochondrion's ancestor arriving after the Asgard and Planctomycetota genes and overlapping with Myxococcota; the paper calls the process gradual, not protracted; (3) added the paper's own limit, that it cannot say which partner was the host or how the partners lived together. Read the other way: Betts and colleagues 2018 (Nature Ecology and Evolution 2, 1556-1562, https://doi.org/10.1038/s41559-018-0644-x) call the earlier evidence for a late mitochondrion an artefact and read their dates as a quick origin driven by the capture, so the late-mitochondrion reading is argued, not settled.

The record is data/papers/bernabeu-2026.toml in the public repository.

Asks how future searches could tell apart the explanations for the silence, above all whether technological civilisations are rare or hidden, and proposes a programme that pairs expanded radio and optical listening with searches for artefacts in the solar system. The artefact searches have a time advantage: they probe evidence that accumulated over the solar system's history rather than needing a civilisation to be active at the same moment as us. In particular, micron-scale technological debris in lunar and planetary soils could constrain past activity in the galaxy.

Ian A. Crawford. arXiv:2606.00463 (astro-ph.IM); for Advancing the Search for Technosignatures, Proceedings of IAU Symposium 404 (forthcoming). Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; They are already here: visitors or their machines have reached the solar system, now or long ago, and we have not recognised them.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/crawford-2026.toml in the public repository.

A long review of every kind of technological trace that has been proposed, ordered by scale from the solar system to the edge of the observable universe: planetary surfaces and atmospheres, stellar energy use, compact objects, communication, travel, and galaxy-scale engineering, including neutrinos and gravitational waves as carriers. It argues for searches that combine biosignatures with technosignatures and for looking for anomalies without assuming a technology. A map of what has and has not been searched for.

Clément Vidal, Benji L. Fields, Damian R. Sowinski, Stuart Bartlett, Richard J. Terrile, Alex Ellery, Daliah Bibas, Armando M. Mastrogiovanni, Niklas Döbler, Manika Singla, Julia DeMarines, Theresa Fisher, Yuri Uno, Jake D. Turner, Evan L. Sneed, Advait Huggahalli, Megan Grace Li, Zhuofu Li, Ramiro Saide. arXiv:2605.21093 (astro-ph.EP, with astro-ph.IM), 118 pages; from a workshop at the Penn State SETI Symposium 2023. Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; We are listening the wrong way: an efficient message is indistinguishable from noise, and they may not be using radio at all.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the arXiv listing (https://arxiv.org/abs/2605.21093, submitted 2026-05-20; a second version appeared 2026-10-02): the author list is 19 names and is now given in full, in the listing's order (the listing splits the last-but-one author's name, Zhuofu Li, across two entries).

The record is data/papers/vidal-2026.toml in the public repository.

A reply to Vazza's 2025 argument that simulating our universe, or even the Earth, is impossible on energy grounds. The authors answer that Vazza refutes a stronger claim than the one Bostrom made: the hypothesis needs only a rendering of coherent experience with no detectable seams, not a bottom-up simulation of all physics, so resource-based refutations miss their target. No new evidence either way and no probability offered.

Eliott Edge, Chad Ashton Brown. Frontiers in Physics 14, 1808725 (2026). Read it.

Bears on: The simulation: the universe we observe is a computer model, and whoever is running it did not bother to include anyone else.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/edge-2026.toml in the public repository.

Simulates ten plausible futures for an Earth-born civilisation over a thousand years with a model that lets it collapse and recover, and finds the share of its life spent technologically active ranges from about 0.38 to 1.00 depending on governance, resource pressure and hazards. The most influential levers are the rate of resource depletion and how much recovers after a collapse. The authors derive an effective detectable duration for a civilisation that is only sometimes active and argue that the silence may reflect many low-duty-cycle civilisations rather than rare ones, so that the lifetime term is better read as the average time a civilisation stays detectable.

Celia Blanco, Jacob Haqq-Misra, George Profitiliotis. arXiv:2604.13774 (physics.soc-ph, with astro-ph.IM). Read it.

Bears on: How long a civilisation stays detectable, in years; Civilisations burn out: the fast growth that makes a civilisation detectable also uses up its planet, and the ones that survive are the ones that slow down and go quiet; Ships in the night: civilisations are common but short-lived against the age of the galaxy, so almost none are alive at the same time.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/blanco-2026.toml in the public repository.

Treats humanity as one data point in a counting experiment and asks what the most pessimistic reading of that one count allows: at least 0.051 communicating civilisations in the observable universe at 95% confidence, or at least 8 in ten trillion for the galaxy. From that bound the chance that the observable universe holds more than one is 97.6%, so solitude in the observable universe is disfavoured, while for the galaxy the bound says almost nothing. It is an argument about what the lower ends of the ranges may honestly be, which is where the headline's answer is decided.

Max Baak, Hella Snoek. arXiv:2603.02372 (stat.OT, with physics.pop-ph), 8 pages. Read it.

Bears on: There is no paradox: once you admit how little we know about each factor, an empty galaxy is the expected result, not a surprise.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/baak-2026.toml in the public repository.

Turns the silence into a bound on the last piece of the equation: if life and intelligence arise as readily as the optimists assume on Earth-like planets, then the absence of any contact means technological civilisations last no more than about 5,000 years in the most optimistic case. The authors stress that this is an upper bound derived from the silence itself, not a prediction of how long civilisations actually last, and that our light cone already covers the whole galaxy's history over the past hundred thousand years.

Sohrab Rahvar, Shahin Rouhani. arXiv:2602.22252 (astro-ph.GA), 4 pages; accepted in Monthly Notices of the Royal Astronomical Society Letters. Read it.

Bears on: How long a civilisation stays detectable, in years; Ships in the night: civilisations are common but short-lived against the age of the galaxy, so almost none are alive at the same time; Self-destruction: civilisations destroy themselves soon after they learn to build the weapons that make them detectable.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. The bound is conditional on the other six pieces being large, so using it to narrow the lifetime range would count the silence twice; the headline deliberately does not update on the silence (docs/method.md), and the reading left the range alone.

The record is data/papers/rahvar-2026.toml in the public repository.

Reports QT45, an RNA molecule only 45 building blocks long, found by searching random sequences, that copies RNA from three-letter pieces in slightly alkaline ice: it made its own complementary strand from a random pool at 94.1% accuracy per letter and a copy of itself from defined pieces, each at a yield of about 0.2% over 72 days. The authors argue that so small a copying machine means such molecules are far commoner in the space of possible RNA sequences than assumed, which bears on how improbable the first self-copying step was. The two halves of the cycle have not yet been joined into full self-replication.

Edoardo Gianni, Samantha L. Y. Kwok, Christopher J. K. Wan, Kevin Goeij, Bryce E. Clifton, Enrico S. Colizzi, James Attwater, Philipp Holliger. Science 391 (6789), 1022-1028 (2026), doi 10.1126/science.adt2760; MRC Laboratory of Molecular Biology, Cambridge. Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it; There is no paradox: once you admit how little we know about each factor, an empty galaxy is the expected result, not a surprise.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the Crossref record, PubMed and Europe PMC: eight authors (now given in full), Science 391 (6789), 1022-1028, issue dated 2026-03-05. Corrected: the date field was 2026-02-13, the date of the press coverage; PubMed and Europe PMC give the paper's first online publication as 2026-02-12, and that is now the date. The first weekly reading (data/readings/2026-10-16.toml, never edited) still says 2026-02-13. The record no longer says anything about Philipp Holliger's role in the group; he is the paper's last author.

The record is data/papers/gianni-2026.toml in the public repository.

A summary of the symposium's discussion of the Great Filter some twenty-five years after Hanson proposed it: whether the idea is still useful, whether current thinking places the improbable step in our past or our future, and what each placement means for the search for life. No new evidence; a statement of where the argument stands.

Richard J. Wilman, Arwen E. Nicholson, Jacob Haqq-Misra. Proceedings of the International Astronomical Union 20 (S387), 178–192 (2026), Cambridge University Press. Read it.

Bears on: The Great Filter is ahead of us: civilisations get as far as we have all the time, and something stops nearly all of them from going further; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/wilman-2026.toml in the public repository.

Looks for the infrared glow of waste heat at the scale of whole galaxies, using the WISE all-sky survey on the galaxies of the 2MASS Redshift Survey with active nuclei and starbursts removed. No galaxy shows it: the typical limit per galaxy is about five to nine hundred million times the Sun's luminosity of waste heat at 150 to 600 kelvin, and no more than about one in 6,500 nearby galaxies can host such heat at the thresholds the sample can test. For a civilisation using a galaxy's worth of starlight at 300 kelvin the limit is about one galaxy in 6,000.

Bo-Lun Huang, Zhen-Zhao Tao, Tong-Jie Zhang. The Astronomical Journal 171 (3), 131 (2026), doi 10.3847/1538-3881/ae31e9; arXiv:2601.07297. Read it.

Bears on: The share of intelligent species that make themselves detectable; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect; We are listening the wrong way: an efficient message is indistinguishable from noise, and they may not be using radio at all; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the arXiv listing (https://arxiv.org/abs/2601.07297, submitted 2026-01-12) and the Crossref record: full names given; published in The Astronomical Journal 171 (3), 131 (online 2026-02-05, issue 2026-03-02). The date field was 2026-01 and is now the submission date, 2026-01-12.

The record is data/papers/huang-2026.toml in the public repository.

Counts every star that fell inside a radio telescope's beam during a large Breakthrough Listen survey, not just the stars it was pointed at, by simulating the galaxy's stellar population: about 6.2 million stars across 1,229 pointings out to 25 kiloparsecs. From the lack of detections it bounds continuously transmitting beacons of at least five times ten to the sixteen watts to fewer than about one in a hundred thousand stellar systems within 2.5 kiloparsecs, and argues that past surveys were less narrowly aimed than often claimed.

Louisa A. Mason, Michael A. Garrett, Andrew P. V. Siemion. arXiv:2511.20231 (astro-ph.IM, with astro-ph.GA), 8 pages. Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/mason-2025.toml in the public repository.

Reviews the equation and proposes replacing its last piece, the lifetime, with a term built from the energy a civilisation can spend, the area it can search and the entropy it generates, framed as the capacity needed to get past a civilisation's own climate and related crises. Concludes that a typical search spans a couple of decades, so a systematic search of about a hundred stars would take a few thousand years.

Orfeu Bertolami. arXiv:2511.11582 (physics.hist-ph, with astro-ph.EP), 15 pages. Read it.

Bears on: How long a civilisation stays detectable, in years; There is no paradox: once you admit how little we know about each factor, an empty galaxy is the expected result, not a surprise.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/bertolami-2025.toml in the public repository.

Re-analyses Cassini's 2008 flight through the plume of Saturn's moon Enceladus, when it hit ice grains only minutes out of the moon's ocean, and finds organic compounds in them: aromatic and oxygen-bearing groups already known from the older ring grains, and newly aliphatic and ring-shaped esters, alkenes and ethers, with nitrogen- and oxygen-bearing compounds tentatively. Because the grains are fresh, the chemistry comes from the ocean, pointing to hydrothermal organic chemistry under the ice. Not life, but the ingredients and the conditions, confirmed in a second ocean.

Nozair Khawaja, Frank Postberg, Thomas R. O'Sullivan, Maryse Napoleoni, Sascha Kempf, Fabian Klenner, Yasuhito Sekine, Maxwell Craddock, Jon Hillier, Jonas Simolka, Lucía Hortal Sánchez, Ralf Srama. Nature Astronomy 9, 1662–1671 (2025). Read it.

Bears on: The share of planets that could support life on which life actually starts; Life under ice: most living worlds may be oceans sealed under miles of ice, whose inhabitants never see the sky; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/khawaja-2025.toml in the public repository.

Works out, within ordinary general relativity, what it would take for a civilisation to live in a frame where time runs slowly: orbits near the photon radius of the galaxy's central black hole are tolerable at biologically bearable accelerations and within reach of a civilisation well below the level of harnessing a whole star, and one civilisation at that level could keep a galaxy-spanning society running with a time-dilation factor of ten thousand, so a trip across the galaxy fits in a lifetime as its citizens count it. Signals from such orbits would drift down in frequency. The authors also argue that fast ships are so vulnerable that the dark forest's caution has a basis.

Chris Reiss, Justin C. Feng. Acta Astronautica 246, 114–128 (September 2026); arXiv:2510.00377 (gr-qc). Read it.

Bears on: The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one; Transcension: advanced civilisations shrink inward into ever denser, faster worlds of their own making, and leave the sky behind; The dark forest: everyone is hiding, because announcing yourself gets you killed.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/reiss-2025.toml in the public repository.

Proposes that civilisations may advance so fast after reaching our stage that the phase in which their technology is detectable to us lasts mere decades, so the overlap between their signals and our instruments is tiny even if they are common and long-lived. On this view the silence says nothing about how many there are. The paper argues for searches that do not assume a technology: persistent large-scale traces such as waste heat and broadband leakage, and anomaly detection across many kinds of data.

Michael Garrett. arXiv:2509.23632 (astro-ph.IM), 15 pages; accepted by Acta Astronautica. Read it.

Bears on: How long a civilisation stays detectable, in years; The share of intelligent species that make themselves detectable; Ships in the night: civilisations are common but short-lived against the age of the galaxy, so almost none are alive at the same time; We are listening the wrong way: an efficient message is indistinguishable from noise, and they may not be using radio at all; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/garrett-2025.toml in the public repository.

The Perseverance rover found, in mudstone of the Bright Angel formation in Neretva Vallis, an old river valley on Mars (at a rock surface the team named Cheyava Falls, from which it drilled the core Sapphire Canyon), organic carbon together with tiny nodules and millimetre-scale reaction fronts made of iron phosphate (probably vivianite) and iron sulfide (probably greigite). The team reads them as the products of chemical reactions, after the mud was laid down and at low temperature, in which organic matter was used up as it stripped oxygen from iron and sulfate. On Earth such minerals often form as by-products of microbes feeding on organic matter. The paper weighs a route without life too: organic compounds can do the iron reaction on their own at 10 to 80 degrees Celsius, but making the sulfide without life would need temperatures above about 150 degrees or sulfide gas from a volcanic source, and the team saw signs of neither. It calls the features potential biosignatures, meaning features that fit biology but need more data before anyone can say whether life made them, and says measurements on the core in laboratories on Earth are what would settle it. The core is still on Mars, and Congress did not fund the mission to bring samples back for 2026.

Joel A. Hurowitz, M. M. Tice, A. C. Allwood, M. L. Cable, K. P. Hand, A. E. Murphy, K. Uckert, J. F. Bell III, T. Bosak, A. P. Broz, E. Clavé, A. Cousin, S. Davidoff, E. Dehouck, K. A. Farley, S. Gupta, S.-E. Hamran, K. Hickman-Lewis, J. R. Johnson, A. J. Jones, M. W. M. Jones, P. S. Jørgensen, L. C. Kah, H. Kalucha, T. V. Kizovski, D. A. Klevang, Y. Liu, F. M. McCubbin, E. L. Moreland, G. Paar, D. A. Paige, A. C. Pascuzzo, M. S. Rice, M. E. Schmidt, K. L. Siebach, S. Siljeström, J. I. Simon, K. M. Stack, A. Steele, N. J. Tosca, A. H. Treiman, S. J. VanBommel, L. A. Wade, B. P. Weiss, R. C. Wiens, K. H. Williford, R. Barnes, P. A. Barr, A. Bechtold, P. Beck, K. Benzerara, S. Bernard, O. Beyssac, R. Bhartia, A. J. Brown, G. Caravaca, E. L. Cardarelli, E. A. Cloutis, A. G. Fairén, D. T. Flannery, T. Fornaro, T. Fouchet, B. Garczynski, F. Goméz, E. M. Hausrath, C. M. Heirwegh, C. D. K. Herd, J. E. Huggett, J. L. Jørgensen, S. W. Lee, A. Y. Li, J. N. Maki, L. Mandon, N. Mangold, J. A. Manrique, J. Martínez-Frías, J. I. Núñez, L. P. O’Neil, B. J. Orenstein, N. Phelan, C. Quantin-Nataf, P. Russell, M. D. Schulte, E. Scheller, S. Sharma, D. L. Shuster, A. Srivastava, B. V. Wogsland, Z. U. Wolf. Nature 645, 332–340 (2025). Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the paper's own page on nature.com (https://www.nature.com/articles/s41586-025-09413-0) and the Crossref record: the byline is 89 names and is now given in full; Nature 645, 332-340, published online 2025-09-10. Read in full 2026-10-20 against the finding. Corrected: (1) the finding said the team found no strong evidence for a route without life; the paper says organic compounds could have done the iron reaction without life at 10 to 80 degrees Celsius, and that the difficulty for a route without life is the sulfide, which would need heating above about 150 to 200 degrees or a distant volcanic source, neither seen; (2) the finding said such minerals form on Earth most often with microbes involved; the paper says they are known to form as by-products of microbial reactions and gives no share, so it now says often; (3) Cheyava Falls is a rock surface in the Bright Angel formation mudstone, not the name of the mudstone, and the drilled core is Sapphire Canyon; (4) the closing sentence said the sample is sealed on the rover and the return mission unfunded, with no source; it now says only what a source shows: Congress declined to fund Mars Sample Return for 2026 and gave 110 million dollars to a Mars Future Missions line instead (Sky and Telescope, 2026-01-20, https://skyandtelescope.org/astronomy-news/nasas-budget-woes-are-over-for-now/). No sentence in the records says evidence of life about this rock; the paper's own term, potential biosignatures, is the one used.

The record is data/papers/hurowitz-2025.toml in the public repository.

Four transits of TRAPPIST-1 e, an Earth-sized planet in the habitable zone of a red dwarf 40 light-years away, observed with JWST, with the star's own variability contaminating every visit. The data give no strong evidence for or against an atmosphere: a bare rock fits, and so does a nitrogen-rich atmosphere with traces of carbon dioxide and methane; hydrogen-rich atmospheres are excluded and thick carbon dioxide atmospheres like Venus's or Mars's are weakly disfavoured. A follow-up campaign of fifteen more transits, paired with transits of the airless inner planet to cancel the star, is under way; its result had not been published by 2026-10-16.

Ana Glidden, Sukrit Ranjan, Sara Seager, Néstor Espinoza, Ryan J. MacDonald, Natalie H. Allen, Caleb I. Cañas, David Grant, Amélie Gressier, Kevin B. Stevenson, Natasha E. Batalha, Nikole K. Lewis, Douglas Long, Hannah R. Wakeford, Lili Alderson, Ryan C. Challener, Knicole Colón, Jingcheng Huang, Zifan Lin, Dana R. Louie, Elijah Mullens, Kristin S. Sotzen, Jeff A. Valenti, Daniel Valentine, Mark Clampin, C. Matt Mountain, Marshall Perrin, Roeland P. van der Marel. Astrophysical Journal Letters (2025), doi 10.3847/2041-8213/adf62e; arXiv:2509.05407. Read it.

Bears on: Planets per star that could support life; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks; We are early: the universe has barely started making civilisations, and we are among the first.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the arXiv listing (https://arxiv.org/abs/2509.05407, submitted 2025-09-05): the author list is 28 names and is now given in full, in the listing's order; title, date and DOI match.

The record is data/papers/glidden-2025.toml in the public repository.

Reports the third object ever seen passing through the solar system from another star, found on 2025-07-01: an orbit far too open to be bound (eccentricity about 6.1), arriving at about 58 kilometres a second, with a compact cloud of dust around it and a colour like other primitive small bodies. Everything measured fits a natural comet. Later work and the 2025 observations treated it as such; a minority argued for an artificial origin, and no evidence for one has been found.

Darryl Z. Seligman, Marco Micheli, Davide Farnocchia, Larry Denneau, John W. Noonan, Henry H. Hsieh, Toni Santana-Ros, John Tonry, Katie Auchettl, Luca Conversi, Maxime Devogèle, Laura Faggioli, Adina D. Feinstein, Marco Fenucci, Marin Ferrais, Tessa Frincke, Michaël Gillon, Olivier R. Hainaut, Kyle Hart, Andrew Hoffman, Carrie E. Holt, Willem B. Hoogendam, Mark E. Huber, Emmanuel Jehin, Theodore Kareta, Jacqueline V. Keane, Michael S. P. Kelley, Tim Lister, Kathleen Mandt, Jean Manfroid, Dušan Marčeta, Karen J. Meech, Mohamed Amine Miftah, Marvin Morgan, Francisco Ocaña, Eloy Peña-Asensio, Benjamin J. Shappee, Robert J. Siverd, Aster G. Taylor, Michael A. Tucker, Richard Wainscoat, Robert Weryk, James J. Wray, Atsuhiro Yaginuma, Bin Yang, Quanzhi Ye, Qicheng Zhang. arXiv:2507.02757 (astro-ph.EP); accepted for publication in the Astrophysical Journal Letters. Read it.

Bears on: They are already here: visitors or their machines have reached the solar system, now or long ago, and we have not recognised them; Berserkers: self-copying machines built long ago still roam the galaxy and destroy any civilisation that shows itself.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the arXiv listing (https://arxiv.org/abs/2507.02757, submitted 2025-07-03): the author list is 47 names and is now given in full, in the listing's order.

The record is data/papers/seligman-2025.toml in the public repository.

The fullest of the reanalyses, combining all of Webb's observations of K2-18 b from all three instruments and fitting them together. With the whole data set, the authors find no significant evidence for dimethyl sulfide or dimethyl disulfide; the feature in the 2025 paper can be fitted equally well by other gases, and the overall evidence for an ocean world remains weak. Their conclusion is not that the planet is lifeless but that the data cannot support the claim made for them. After this paper the working view among exoplanet astronomers was that K2-18 b has shown methane, probably carbon dioxide, and nothing yet that bears on the chance that life starts. For this project the episode is a lesson in how a factor's evidence arrives: as a claim, a correction and a long argument, not as a result.

Rafael Luque, Caroline Piaulet-Ghorayeb, Michael Radica, and 5 others. Astronomy & Astrophysics 700, A284 (2025); arXiv:2505.13407. Read it.

Bears on: The share of planets that could support life on which life actually starts; Planets per star that could support life; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the arXiv page and Crossref: eight authors (Rafael Luque, Caroline Piaulet-Ghorayeb, Michael Radica, Q. Xue, M. Zhang, Jacob Bean, D. Samra, M. E. Steinrueck), the analysis joins three instruments (NIRISS, NIRSpec and MIRI), the paper was published in Astronomy & Astrophysics 700, A284 (2025), and other gases fit equally well, not better. The Welbanks and colleagues paper (arXiv:2504.21788, 2025-04-30, 'Challenges in the detection of gases in exoplanet atmospheres') exists, but it is a methods paper that uses K2-18 b as a worked case, not a plain reanalysis; it could be logged separately.

The record is data/papers/luque-2025.toml in the public repository.

A second look at K2-18 b with a different Webb instrument, working at longer infrared wavelengths than the 2023 observation. The authors report a feature they attribute to dimethyl sulfide or the related dimethyl disulfide, at about three standard deviations of confidence and at concentrations thousands of times higher than Earth's, and say that no known process other than life produces so much. The announcement was reported worldwide as the strongest hint yet of life beyond Earth. Three standard deviations is below the level physicists treat as a discovery, the authors said so, and within weeks other groups had reanalysed the same data and found the feature could not be distinguished from noise or from other gases (the three entries that follow).

Nikku Madhusudhan, Savvas Constantinou, Måns Holmberg, Subhajit Sarkar, Anjali A. A. Piette, Julianne I. Moses. The Astrophysical Journal Letters 983, L40 (2025). Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16.

The record is data/papers/madhusudhan-2025.toml in the public repository.

A short note, out within about a week of madhusudhan-2025, that tests the new data without assuming any particular gas. Taylor asks whether the spectrum shows any feature at all beyond a flat line with a slope, and finds that it does not with any confidence: a featureless model fits the data about as well as one with dimethyl sulfide in it. The point is methodological and important. A three-standard-deviation signal for a specific molecule can appear when the data are fitted only against models that contain that molecule; a fair test asks first whether there is anything there to explain.

Jake Taylor. Research Notes of the American Astronomical Society 9, 118 (2025); arXiv:2504.15916. Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against Crossref: the note is volume 9, number 5, article 118 (published 2025-05-16), not 86; the arXiv identifier is right.

The record is data/papers/taylor-2025.toml in the public repository.

Maps 24,706 hot, massive stars within a kiloparsec of the Sun and from their numbers and ages works out how fast such stars form and how often they explode. Scaled to the whole galaxy that gives a star formation rate of 0.67 solar masses a year and a core-collapse supernova rate of 0.4 to 0.5 per century, both lower than earlier estimates, which the authors put down to a better census and newer models of how stars evolve. Nearby supernovae come about 2.5 times per billion years, close to the rate at which Earth has suffered mass extinctions with no settled cause, which the authors take as support for a supernova behind one or more of them (the paper names the late Devonian and late Ordovician events, 372 and 445 million years ago).

Alexis L. Quintana, Nicholas J. Wright, Juan Martínez García. Monthly Notices of the Royal Astronomical Society 538 (2), 1367-1383 (2025), doi 10.1093/mnras/staf083; arXiv:2503.08286. Read it.

Bears on: How many stars are born in the galaxy each year; A hostile galaxy: exploding stars and bursts of radiation have kept resetting life almost everywhere, and only recently stopped; Natural catastrophes: asteroid strikes, giant eruptions and nearby exploding stars end civilisations before they can last long enough to be noticed.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16. The star-formation figure here (0.67 solar masses a year, roughly 2 to 3 stars a year at Siegert 2023's conversion) sits inside the range this site uses, on its low side; Siegert 2023's figure sits above it. The two together are the spread of the current measurements. Checked 2026-10-17: the arXiv listing (https://arxiv.org/abs/2503.08286, submitted 2025-03-11) names three authors, the third being Juan Martínez García; the Crossref record gives Monthly Notices 538 (2), 1367-1383, doi 10.1093/mnras/staf083, published 2025-03-17; and the paper's own text (arXiv HTML version) says the near-Earth core-collapse rate of about 2.5 per billion years is consistent with one or both of the late Devonian and late Ordovician extinction events, 372 and 445 million years ago, citing Bond and Grasby (2017), so the extinctions are the paper's wording and not only the press release's.

The record is data/papers/quintana-2025.toml in the public repository.

Re-examines the model, going back to Carter 1983, on which the long wait for intelligence on Earth shows that several steps on the way were intrinsically improbable, and argues the record is explained without any hard step: the major transitions (complex cells, animals, and so on) waited for global environmental windows of habitability to open, so humans arrived on time relative to the biosphere's lifespan rather than early or late. If so, the argument that intelligence must be rare because it took so long here loses its force.

Daniel B. Mills, Jennifer L. Macalady, Adam Frank, Jason T. Wright. Science Advances 11, eads5698 (2025); arXiv:2408.10293. Read it.

Bears on: The share of planets with life on which it becomes intelligent; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it; We are early: the universe has barely started making civilisations, and we are among the first; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/mills-2025.toml in the public repository.

Compares how the sensory parts of the forebrain are built in a chick, a gecko and a mouse, by birth-dating neurons and mapping their genes, and finds that birds and mammals reach similar circuits by different developmental routes: the neurons are not inherited from a common ancestor but arrived at separately. One of a set of papers in the same issue (Zaremba and others on the bird forebrain's cell types) read together as showing that complex cognition evolved at least twice among vertebrates.

Eneritz Rueda-Alaña, Rodrigo Senovilla-Ganzo, Marco Grillo, Enrique Vázquez, Sergio Marco-Salas, Tatiana Gallego-Flores, Aitor Ordeñana-Manso, Artemis Ftara, Laura Escobar, Alberto Benguría, Ana Quintas, Ana Dopazo, Miriam Rábano, María dM Vivanco, Ana María Aransay, Daniel Garrigos, Ángel Toval, José Luis Ferrán, Mats Nilsson, Juan Manuel Encinas-Pérez, Maurizio De Pittà, Fernando García-Moreno. Science 387 (6735), eadp3411 (2025). Read it.

Bears on: The share of planets with life on which it becomes intelligent; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Checked 2026-10-17 against the Crossref record, PubMed and Europe PMC: 22 authors (now given in full); Science 387 (6735), eadp3411, issue dated 2025-02-14. Corrected: the date field was 2025-02-13, the date of the press coverage; the three catalogues give 2025-02-14, and the 2025-01-02 that one catalogue shows is a placeholder for the volume, not the paper's date. The companion paper is confirmed in Crossref: Zaremba and 15 others, 'Developmental origins and evolution of pallial cell types and structures in birds', Science 387 (6735), eadp5182, doi 10.1126/science.adp5182, same issue. The first weekly reading (never edited) still says 2025-02-13.

The record is data/papers/rueda-alana-2025.toml in the public repository.

An independent group reprocessed the 2023 Webb data on K2-18 b from the raw frames up, with several different pipelines, before the 2025 claim appeared. They confirm methane and find no statistically significant evidence for carbon dioxide, and no statistically significant sign of dimethyl sulfide: the hint in the original paper depends on choices made in reducing the data and does not survive other reasonable choices. The paper is the first of the reanalyses and set the pattern for the ones that followed: the planet has an atmosphere with methane in it, and the rest is not yet established.

Stephen P. Schmidt, Ryan J. MacDonald, Shang-Min Tsai, and 13 others. The Astronomical Journal 170, 298 (2025); arXiv:2501.18477. Read it.

Bears on: The share of planets that could support life on which life actually starts; Planets per star that could support life; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the paper and Crossref: Kevin Stevenson is not an author (the second author is Ryan MacDonald; sixteen authors in all); the title reads NIRISS+NIRSpec; it was published in The Astronomical Journal 170(6), 298, on 2025-11-05; and the final version finds no significant evidence for carbon dioxide, not weaker evidence.

The record is data/papers/schmidt-2025.toml in the public repository.

Simulates competing civilisations spreading between stars, each with a population, an environment and a technology level that grow at their own rates, and with battles when they meet. Civilisations whose three rates are out of balance collapse; balanced ones can come to dominate, but even the largest leave big gaps between their colonies, so small civilisations survive in the gaps. A model result in favour of settlement that spreads unevenly and leaves regions unreached.

Gregory Roudenko, Yurrian Pierre-Boyer. arXiv:2411.00061 (physics.soc-ph, with q-bio.PE). Read it.

Bears on: Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/roudenko-2024.toml in the public repository.

The brightest gamma-ray burst ever recorded had been called a once-in-ten-thousand-years event by extrapolating the known population; the authors argue it may instead belong to a separate population of narrow-jet bursts, in which case such bursts could come as often as one in 200 years across the sky. For our own galaxy that gives a burst bright enough to cause a mass extinction on Earth about once per 500 million years, from anywhere in the galaxy, which would make the galaxy a more sterile place than the earlier reckoning. Confirming the population would take more than a century of monitoring.

Justin Finke, Soebur Razzaque. arXiv:2407.15940 (astro-ph.HE), 15 pages; submitted to the AAS journals. Read it.

Bears on: A hostile galaxy: exploding stars and bursts of radiation have kept resetting life almost everywhere, and only recently stopped; Natural catastrophes: asteroid strikes, giant eruptions and nearby exploding stars end civilisations before they can last long enough to be noticed.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict.

The record is data/papers/finke-2024.toml in the public repository.

The first substantial reply to madhusudhan-2023 on the question of the ocean rather than the life. The authors model the planet's chemistry and find that the same mix of methane and carbon dioxide with no ammonia comes out naturally from a small, gas-rich Neptune-like planet with a thick hydrogen envelope and no surface anyone could stand on, and that an ocean world would in fact be expected to show more ammonia, not less. If so, the planet is not a candidate for life as we know it, whatever gases it has. A companion paper by Oliver Shorttle and colleagues (The Astrophysical Journal Letters 962, L8, 2024) argued the surface could as easily be magma as water.

Nicholas F. Wogan, Natasha E. Batalha, Kevin J. Zahnle, Joshua Krissansen-Totton, Shang-Min Tsai, Renyu Hu. The Astrophysical Journal Letters 963, L7 (2024). Read it.

Bears on: Planets per star that could support life; The share of planets that could support life on which life actually starts; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16.

The record is data/papers/wogan-2024.toml in the public repository.

Webb's first spectrum of K2-18 b, a planet about 8.6 times Earth's mass in the habitable zone of a red dwarf 124 light years away. The authors report methane and carbon dioxide in its atmosphere (methane at about five standard deviations of confidence, carbon dioxide at about three) and no ammonia, which they argue fits a world with a hydrogen atmosphere over a liquid-water ocean, a type they call 'hycean'. They also report a weak hint of dimethyl sulfide, a gas that on Earth is made only by living things, and are careful to call it tentative. The paper started the two-year argument, logged in the entries that follow, about whether this planet has an ocean at all and whether the hint was real.

Nikku Madhusudhan, Subhajit Sarkar, Savvas Constantinou, Måns Holmberg, Anjali A. A. Piette, Julianne I. Moses. The Astrophysical Journal Letters 956, L13 (2023). Read it.

Bears on: The share of planets that could support life on which life actually starts; Planets per star that could support life; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the paper and Crossref: the mass is 8.63 plus or minus 1.35 Earth masses (the record said about eight) and the online date is 2023-10-09; methane at five and carbon dioxide at three standard deviations are as in the abstract; the distance of 124 light years is the NASA Exoplanet Archive's (NASA's Webb release rounds to 120). The first of six K2-18 b entries logged together on 2026-10-16: this claim, the 2024 pushback on the ocean (wogan-2024), the 2025 stronger claim (madhusudhan-2025) and three reanalyses that found no significant signal (schmidt-2025, taylor-2025, luque-2025).

The record is data/papers/madhusudhan-2023.toml in the public repository.

The same measurement as greene-2023 for the second planet of TRAPPIST-1, which gets about as much light as Venus does. Its day side is around 380 kelvin, cooler than bare rock would be but far too hot for a thick carbon-dioxide atmosphere; the data allow at most a thin atmosphere or none. Together the two results suggest that the inner planets of the system either never had much atmosphere or lost it to the star's radiation early on, which is the pessimistic case for planets around red dwarfs. The habitable-zone planets further out (e, f and g) are the real test, and their measurements are harder and still coming in.

Sebastian Zieba, Laura Kreidberg, Elsa Ducrot, Michaël Gillon, Caroline Morley, and others. Nature 620, 746–749 (2023). Read it.

Bears on: Planets per star that could support life; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16.

The record is data/papers/zieba-2023.toml in the public repository.

The first measurement of heat from an Earth-sized rocky planet around another star. The James Webb Space Telescope watched the innermost of the seven planets of the red dwarf TRAPPIST-1 pass behind its star and measured the drop in infrared light. The planet's day side is about 500 kelvin (around 230 degrees Celsius), which is what bare rock heated by the star would be with no atmosphere moving heat around, and rules out a thick carbon-dioxide atmosphere like Venus's. For the equation this matters because most stars are red dwarfs and most habitable-zone planets orbit them; whether such planets can keep an atmosphere against their star's flares is the main open question about the third factor, and this was the first hard data point.

Thomas P. Greene, Taylor J. Bell, Elsa Ducrot, Achrène Dyrek, Pierre-Olivier Lagage, Jonathan J. Fortney. Nature 618, 39–42 (2023). Read it.

Bears on: Planets per star that could support life; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. TRAPPIST-1 b is too close to its star to be habitable in any case; it was measured first because it is the easiest, and as a test of whether rocky planets around red dwarfs keep atmospheres at all.

The record is data/papers/greene-2023.toml in the public repository.

Models the galaxy's glow of radioactive aluminium-26, which massive stars scatter and which gamma rays reveal through the dust, and finds the observed distribution is reproduced only with a star formation rate of 4 to 8 solar masses a year, two to four times the conventional figure of about 2, with a supernova rate of 1.8 to 2.8 per century. The paper gives masses only, not a count of stars. Science News's report of it (Ken Croswell, 2023-02-23) converted the masses to about 10 to 20 stars born a year; at half a solar mass per star, the conversion this site uses, it is 8 to 16. Either is above the top of the range this site used when the paper was logged (1 to 10; the top is now 20). The abstract notes that an unknown but strong foreground at 1.8 MeV could be present.

Thomas Siegert, Moritz M. M. Pleintinger, Roland Diehl, Martin G. H. Krause, Jochen Greiner, Christoph Weinberger. Astronomy and Astrophysics 672, A54 (2023), doi 10.1051/0004-6361/202244457; arXiv:2301.10192. Read it.

Bears on: How many stars are born in the galaxy each year.

What it moved:

Piece	What	From	To	Why
How many stars are born in the galaxy each year	estimate	four published estimates	six published estimates: the 2023 aluminium-26 measurement (8 to 16 stars a year) and the 2025 massive-star census (1.3 a year) added	The weekly reading of 2026-10-16 found two measurements the record did not carry and flagged the range for me. He ruled on 2026-10-16: add both to the estimate list and leave the range's top at 10 until the outside review of the seven ranges.
How many stars are born in the galaxy each year	range.high	10	20	Decided 2026-10-17 on the outside review (items 2 and 4 of docs/reviews/2026-10-17-rulings.md): the 2023 aluminium-26 measurement gives 8 to 16 stars a year, above the old top of 10, and the weekly reading of 2026-10-16 had flagged the same. The floor of 1 holds (the 2025 census gives 1.3) and the shape stays log-uniform.
How many stars are born in the galaxy each year	estimate	Drake 1961 source: Drake and Sobel (1992), pages 55 to 62. Siegert 2023 source: titled 'Galactic 26Al traces metal loss through hot chimneys'	Drake 1961 source: Pearman (1963), page 289. Siegert 2023 source: titled 'Galactic Population Synthesis of Radioactive Nucleosynthesis Ejecta'	Checked 2026-10-17 against Pearman's 1963 account of the meeting, the first printed one, which is openly readable at the Internet Archive; the source line now cites it with its page, in place of Drake and Sobel (1992), whose page numbers came from a secondary summary and which has not been read. The estimate's numbers are unchanged. The Siegert source line also gave the title of a different paper; it is corrected to the paper the publication record (siegert-2023) and the Crossref record for A&A 672, A54 describe.

Logged by the first weekly reading, 2026-10-16; the reading record says which records it was read against and with what verdict. Flagged for me: the estimate is above the range's top, and whether the range's top should rise is his call (the reading's limits, docs/weekly-reading.md). Moved the range on 2026-10-17: I ruled on the outside review that the top of the star-formation range rises from 10 to 20 (moved_range changed from false to true with a changelog line). Checked 2026-10-17 against the paper (arXiv HTML version of 2301.10192, https://arxiv.org/abs/2301.10192) and the Crossref record: the abstract gives a star formation rate of 4 to 8 solar masses a year, a supernova rate of 1.8 to 2.8 per century and the foreground remark, in solar masses only; there is no conversion to a number of stars in the paper. The conversion to 10 to 20 stars a year is Science News's (https://www.sciencenews.org/article/milky-way-star-formation-astronomy, which also gives the 'two to four times the conventional estimate'), and the 8 to 16 is this site's own arithmetic at 0.5 solar masses per star; the finding now says so. A&A 672, A54 (2023), published online 2023-03-30. The sentence in the finding that if some of the emission is local the rate sits at the low end of the new range is removed: it is not in the abstract.

The record is data/papers/siegert-2023.toml in the public repository.

The puzzle the paper starts from is that humans seem to have arrived early: most of the universe's habitable time lies ahead, around small stars that will burn for trillions of years, so an observer picked at random should expect to appear much later than 13.8 billion years in. The authors' explanation is that some civilisations expand ('grabby' ones), fill space at a good fraction of light speed, and make their volume uninhabitable for newcomers, so anyone who appears at all has to appear before the grabbers arrive. A three-parameter model fit to our own date gives their picture of the universe: grabby civilisations already control about 40 to 50 percent of the universe's volume, each will eventually control roughly 100,000 to 30 million galaxies, and our descendants could meet them in something like 200 million to 2 billion years; the volumes they control would look very large in our sky. The title's point is that if this is right, quiet civilisations that never expand cannot be much more common than loud ones.

Robin Hanson, Daniel Martin, Calvin McCarter, Jonathan Paulson. The Astrophysical Journal 922, 182 (2021); arXiv:2102.01522. Read it.

Bears on: The share of planets with life on which it becomes intelligent; How long a civilisation stays detectable, in years; The share of intelligent species that make themselves detectable; We are early: the universe has barely started making civilisations, and we are among the first; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the abstract (arXiv:2102.01522): the paper does not give a rate of one grabby civilisation per million galaxies (its 'one-in-a-million-galaxies' remark is about Earth being a rare Earth); the 200 million to 2 billion years is the time until we could meet them, not until they become visible.

The record is data/papers/hanson-2021-grabby.toml in the public repository.

The centre of the galaxy packs the most stars into the smallest patch of sky, so a single pointing covers tens of millions of them, and a civilisation wanting to be heard across the galaxy might reasonably signal from or toward it. Breakthrough Listen observed the central region for about 18 hours (7.0 with Parkes and 11.2 with Green Bank) between 1 and 8 gigahertz, and found no artificial signals, though it did pick up bursts from a known magnetar. The limits are coarse because the stars are so far away: only a transmitter far more powerful than anything humans have built would have been seen. The paper lays out the plan for a deeper survey of the region and is the first of a series.

Vishal Gajjar, Karen I. Perez, Andrew P. V. Siemion, Griffin Foster, Bryan Brzycki, and 21 others. The Astronomical Journal 162, 33 (2021). Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; Everyone listens and nobody transmits: if every civilisation makes the same cautious choice we have made, the galaxy is full of ears and empty of voices.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the paper's abstract: the observing time was about 18 hours, not seven, the frequency range is 1 to 8 gigahertz, and the paper has 26 authors (the 'Breakthrough Listen team' is not one). Later Breakthrough Listen results (a machine-learning search of 820 stars by Peter Ma and colleagues in Nature Astronomy, 2023, and the ongoing MeerKAT survey) should be logged by the weekly scan as it reads them.

The record is data/papers/gajjar-2021.toml in the public repository.

Carter's 1983 timing argument redone with modern dates for Earth's big steps (the start of life, oxygen-making photosynthesis, complex cells, animals, and intelligence) and with the Earth's remaining habitable time, which the authors take to be around a billion years, not five. If those steps were each easy, they would have happened early and quickly; that they took most of Earth's habitable span suggests each would typically take far longer than a planet stays habitable. Running the numbers, the authors find that the expected time for the whole sequence exceeds Earth's habitable lifetime by a large factor, and conclude that intelligent life is probably exceptionally rare, while being clear that the inference rests on Earth being an ordinary case.

Andrew E. Snyder-Beattie, Anders Sandberg, K. Eric Drexler, Michael B. Bonsall. Astrobiology 21, 265–278 (2021). Read it.

Bears on: The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; Complex cells are rare: life may start easily but then stay simple forever, because the one step that made big cells happened only once; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. A companion to carter-1983 and the quantitative counterweight to kipping-2020.

The record is data/papers/snyder-beattie-2021.toml in the public repository.

The Kepler team's final answer to the question the telescope was built for. Using the complete catalogue, the mission's measured reliability, and stellar sizes from the Gaia spacecraft, the authors estimate how many rocky planets (half to one and a half times Earth's radius) sit in the habitable zones of stars like the Sun. For the conservative habitable zone they find between about 0.37 and 0.60 such planets per star, with wide uncertainty (the lower figure's range runs from 0.16 to 0.85); for a more generous zone, 0.58 to 0.88. In plain terms, roughly half of Sun-like stars have a rocky planet where liquid water could exist, and the nearest such planet is probably within about 20 light years. This is the best current measurement of the equation's third factor for Sun-like stars; red dwarfs, which are more numerous, are a separate question.

Steve Bryson, Michelle Kunimoto, Ravi K. Kopparapu, Jeffrey L. Coughlin, William J. Borucki, and the Kepler team. The Astronomical Journal 161, 36 (2021); arXiv:2010.14812 (2020-10-28). Read it.

Bears on: Planets per star that could support life; The share of stars that have planets; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Checked 2026-10-16 against the paper: the four occurrence figures and the 0.16 to 0.85 range are as quoted, the nearest such planet is about 6 parsecs (about 20 light years), and the sample is G and K dwarfs rather than strictly Sun-like stars. The paper gives its results as a range between two statistical methods rather than a single number, which is why the finding quotes two figures for each zone.

The record is data/papers/bryson-2020.toml in the public repository.

A radio telescope pointed at one star also hears every other star in its beam. Using the Gaia catalogue of star positions and distances, the authors count how many stars were in the beams of price-2020's 1,327 target stars: 288,315, out to thousands of light years, most of them far more distant than the targets. Because a transmitter at a greater distance has to be more powerful to be heard, the limits are weaker for those stars, but there are two hundred times more of them. The paper's result is that no more than about 0.04 percent of stars within about 650 light years (200 parsecs) host a transmitter of at least 2.5 x 10^14 watts that is on most of the time, the first limit of its kind on a sample that large, and a reminder that the sky has been searched less than it feels.

Bart S. Wlodarczyk-Sroka, Michael A. Garrett, Andrew P. V. Siemion. Monthly Notices of the Royal Astronomical Society 498, 5720–5729 (2020). Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; Everyone listens and nobody transmits: if every civilisation makes the same cautious choice we have made, the galaxy is full of ears and empty of voices.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the abstract of the final (second) version: the count is 288,315 stellar objects up from 1,327 targets (not pointings), and the 0.04 percent limit holds within 200 parsecs (about 650 light years, not 1,000) for a transmitter of at least 2.5 x 10^14 watts; the paper also gives about 0.07 percent within 50 parsecs at 10^13 watts. The first version of the paper had different numbers.

The record is data/papers/wlodarczyk-sroka-2020.toml in the public repository.

Instead of guessing the biological factors, the authors assume Earth is typical: under their 'strong' assumption, intelligent life appears on a suitable planet between 4.5 and 5.5 billion years after the planet forms, as it did here, and only around stars with at least the Sun's share of heavy elements. Combined with the galaxy's star-formation history and a communicating lifetime of 100 years (ours so far), that gives at least about 36 communicating civilisations in the galaxy now, with a range of roughly 4 to 211, and a nearest neighbour at most around 17,000 light years away, too far to talk to within a lifetime at radio speeds. Under the weaker assumption (any time after 5 billion years) the count rises to hundreds or more. The headline number was widely reported; the paper's own point is that the answer hinges almost entirely on how long civilisations last.

Tom Westby, Christopher J. Conselice. The Astrophysical Journal 896, 58 (2020). Read it.

Bears on: The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; How long a civilisation stays detectable, in years; How many stars are born in the galaxy each year; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Checked 2026-10-16 against the abstract: at least 36 (minus 32, plus 175) civilisations, so 4 to 211, for a communicating lifetime of 100 years, and a nearest neighbour at most 17,000 (minus 10,000, plus 33,600) light years away. The paper frames both as limits, which the finding now says.

The record is data/papers/westby-2020.toml in the public repository.

Earth's life started quickly, within a few hundred million years of the planet becoming habitable, and intelligence took almost four billion years more. Kipping asks what those two dates say about how likely each step is on an Earth-like planet, taking care to assume as little as possible beforehand. He finds odds of at least about 2.8 to 1 that life starts quickly when conditions allow (rising to 8.7 to 1 if a contested chemical trace in ancient mineral grains counts), and weaker odds of about 3 to 2 that intelligence is rare even when life is common. In plain terms: the record leans toward life being easy and minds being hard, but neither lean is strong, and the paper is explicit that a single example cannot settle it.

David Kipping. Proceedings of the National Academy of Sciences 117, 11995–12003 (2020). Read it.

Bears on: The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Corrected 2026-10-16 against the abstract: the odds for rapid life are at least 2.8 to 1 (8.7 to 1 with the disputed zircon evidence), not 3 to 1; the 3 to 2 for rare intelligence is as the abstract says. Spiegel and Turner reached a more cautious conclusion with the same kind of analysis in 2012 (in this log as spiegel-2012); the difference is mostly in how much the two papers assume before looking at the data.

The record is data/papers/kipping-2020.toml in the public repository.

Breakthrough Listen, the privately funded ten-year radio search begun in 2016, reports its first large survey: 1,327 stars within about 160 light years, observed with the Green Bank and Parkes telescopes across a wide band of radio frequencies, with no signal found that could not be traced to human interference. The authors turn the silence into a limit: at 95 percent confidence, fewer than about 0.45 percent of the stars watched with Green Bank host a narrow-band transmitter above about 2 trillion watts, and fewer than about 2 percent on the Parkes data above about 9 trillion watts (all the stars lie within 50 parsecs). The limits apply only to the narrow kind of signal searched for, as the paper says, and to the moments the telescope was looking.

Danny C. Price, J. Emilio Enriquez, Bryan Brzycki, Steve Croft, Daniel Czech, David DeBoer, Julia DeMarines, Griffin Foster, Vishal Gajjar, Nectaria Gizani, Greg Hellbourg, Howard Isaacson, Brian Lacki, Matt Lebofsky, David H. E. MacMahon, Imke de Pater, Andrew P. V. Siemion, Dan Werthimer, and 6 others. The Astronomical Journal 159, 86 (2020). Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; Everyone listens and nobody transmits: if every civilisation makes the same cautious choice we have made, the galaxy is full of ears and empty of voices; Shouting is costly: a beacon that could be heard across the galaxy is expensive to run, so anyone sending one sends short pulses, rarely, and our searches almost never catch them.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Corrected 2026-10-16 against the paper (arXiv:1906.07750) and Crossref: the paper has no 0.1 percent figure (the limits are 0.45 percent for Green Bank above 2 x 10^12 watts, 0.37 percent for the Green Bank S-band, 2 percent for Parkes above 9 x 10^12 watts); the author list has 24 names, the six beyond those listed being James A. Green, Jane F. Kaczmarek, Ronald J. Maddalena, Stacy Mader, Jamie Drew and S. Pete Worden; and the date is the journal's online date, 2020-02-05. Supersedes Enriquez and colleagues' 2017 survey of 692 stars (The Astrophysical Journal 849, 104).

The record is data/papers/price-2020.toml in the public repository.

The authors define the space a radio search has to cover (where on the sky, how far, which frequencies, how strong, how often, when, and how the signal is shaped) as a many-dimensional haystack, and add up how much of it every search to date had looked through. The total comes to about six parts in 10 to the power 18 of the plausible volume, which they compare to somewhere between a large hot tub and a small swimming pool of water drawn from all the Earth's oceans; Jill Tarter's earlier comparison was a single glass. The paper is the standard answer to 'we have searched and found nothing': in a precise sense, we have hardly searched at all, and the silence so far is a weak constraint on the number of civilisations.

Jason T. Wright, Shubham Kanodia, Emily Lubar. The Astronomical Journal 156, 260 (2018). Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; We are listening the wrong way: an efficient message is indistinguishable from noise, and they may not be using radio at all.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the paper: its text gives a combined fraction of about 6 x 10^-18 (about 8,000 litres out of 1.3 x 10^21 litres of ocean), a little under one part in 10^17, and its abstract compares that to a volume between a large hot tub and a small swimming pool.

The record is data/papers/wright-2018.toml in the public repository.

Run with a range for each factor instead of a single value, the Drake Equation gives a median of 0.32 detectable civilisations in the galaxy, a 52% chance that we are alone in the galaxy and a 38% chance that we are alone in the observable universe (their Table 2, 'no update' row). Taking the silence itself into account raises those to between 53% and 99.6%, and between 39% and 85%, depending on what the silence is taken to show. The paper's point is that the paradox comes from treating very uncertain factors as if they were known.

Anders Sandberg, Eric Drexler, Toby Ord. arXiv:1806.02404 (physics.pop-ph), Future of Humanity Institute, Oxford; submitted to Proceedings of the Royal Society A. Read it.

Bears on: How many stars are born in the galaxy each year; The share of stars that have planets; Planets per star that could support life; The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; The share of intelligent species that make themselves detectable; How long a civilisation stays detectable, in years.

What it moved:

Piece	What	From	To	Why
The share of planets that could support life on which life actually starts	range.shape	every power of ten from 1 in 10 to the 22 up to 1 equally likely	a rate centred on one start per planet, spread over fifty orders of magnitude	I ruled on 2026-10-16: 'Use the 2018 paper's log-normal rate shape for the chance life starts, adding the third shape to the record layout.' The first pass had used a plain low-to-high range as a stand-in because the layout allowed nothing else; that stand-in put most of the weight on 'life is very rare', where the paper splits the weight evenly between 'life is easy' and 'life is hard'.

Founding entry. The project's method and its first-pass ranges start from this paper's Table 1 (log-uniform ranges for six factors and a very wide log-normal for the chance life starts). On 2026-10-16 I ruled that the chance life starts (data/factors/life.toml) takes this paper's shape instead of the first pass's plain low-to-high range, so the paper is named by a revision there and 'moved a range' is true. Figures checked against the paper's text on 2026-10-16.

The record is data/papers/sandberg-2018.toml in the public repository.

The authors drop the lifetime factor and ask a different question: not how many civilisations exist now, but how many have ever existed. With the exoplanet record fixing the first three factors, the product of the remaining biological ones would have to be below about 2.5 in 10 to the power 24 for humanity to be the only technological species the observable universe has ever produced, and below about 1.7 in 10 to the power 11 for us to be the only one in the Milky Way's history. In plain terms, unless the chance of a habitable planet producing a technological species is almost unimaginably small, we are not the first. The paper does not say anyone is out there now; it says the pessimists need a very strong claim to say nobody ever was.

Adam Frank, Woodruff T. Sullivan III. Astrobiology 16, 359–362 (2016). Read it.

Bears on: The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; The share of intelligent species that make themselves detectable; Planets per star that could support life; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; We are early: the universe has barely started making civilisations, and we are among the first; Civilisations burn out: the fast growth that makes a civilisation detectable also uses up its planet, and the ones that survive are the ones that slow down and go quiet.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Checked 2026-10-16 against the paper (arXiv:1510.08837): both thresholds are as written. They are set on different footing: 2.5 x 10^-24 is for a one-in-a-hundred chance that we are alone in the observable universe, and 1.7 x 10^-11 is for an expected one species in the Milky Way's history.

The record is data/papers/frank-2016.toml in the public repository.

Computation costs energy, and the cost of erasing a bit of information falls as the surroundings get colder. The universe is still warm from its birth and will cool for trillions of years, so a civilisation that cares most about how much thinking it can do in total would gain about a factor of 10 to the power 30 by waiting: do nothing now, sleep through the hot early universe, and wake when the cosmos is cold. The authors propose that advanced civilisations may be doing exactly that, dormant but present, which would explain why we see neither them nor their works. They note it predicts that someone should be guarding resources against waste, and that we see no sign of such guarding, which is the main argument against it.

Anders Sandberg, Stuart Armstrong, Milan M. Ćirković. Journal of the British Interplanetary Society 69, 406–415 (2016); arXiv:1705.03394 (2017). Read it.

Bears on: The share of intelligent species that make themselves detectable; Aestivation: advanced civilisations are asleep, waiting for the universe to cool so that their thinking will go further on the same energy; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Charles Bennett, Robin Hanson and Jess Riedel replied in 'Comment on the aestivation hypothesis' (Foundations of Physics 49, 820–829, 2019), arguing that the thermodynamic saving is smaller than claimed because the cold-universe computation could be done later anyway while still using resources now.

The record is data/papers/sandberg-2016.toml in the public repository.

A history paper that checks the story. Going back to the recollections of the people at the 1950 Los Alamos lunch where Fermi asked 'where is everybody?', Gray finds that Fermi was asking about interstellar travel, not doubting that other civilisations exist, and that the argument now called the Fermi paradox (they would be here by now, so they do not exist) was made by Hart in 1975 and Tipler in 1980. He also argues it is not a paradox but an argument with disputable premises, mainly that every civilisation would colonise and that colonisation would be fast and complete. The paper is the reason this project calls the thing the silence rather than the paradox where it can.

Robert H. Gray. Astrobiology 15, 195–199 (2015). Read it.

Bears on: The share of intelligent species that make themselves detectable; We have not looked long enough: all the searching ever done amounts to a glass of water dipped from the ocean; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Gray's own suggested name for the argument is the 'Hart-Tipler argument'.

The record is data/papers/gray-2015.toml in the public repository.

Gamma-ray bursts, the brightest explosions known, would strip a planet's ozone and kill most surface life if one went off close enough. The authors work out how often that happens as a function of where and when a planet sits. For Earth's position they find a 50 percent chance of a lethal burst in the last 500 million years and a very good chance, though not a certainty, over the last 5 billion. The galaxy's inner region, which holds about a quarter of its stars, is hit too often for land life to last; and because bursts were more common when the universe was young and poor in heavy elements, life as we know it was probably impossible anywhere at redshifts above 0.5, which is more than about 5 billion years ago, and is still unlikely in most galaxies. The paper is the quantitative basis for the idea that the galaxy keeps being reset.

Tsvi Piran, Raul Jimenez. Physical Review Letters 113, 231102 (2014). Read it.

Bears on: Planets per star that could support life; How long a civilisation stays detectable, in years; How many stars are born in the galaxy each year; Natural catastrophes: asteroid strikes, giant eruptions and nearby exploding stars end civilisations before they can last long enough to be noticed; We are early: the universe has barely started making civilisations, and we are among the first; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Corrected 2026-10-16 against the paper: the abstract gives a 50 percent chance in the last 500 million years and 'a very good chance (but no certainty)' over 5 billion, not a near certainty; the inner galaxy is described by the share of stars it holds (about a quarter), not as a quarter of the radius; and the 5-billion-year threshold is this project's conversion of the paper's redshift limit of 0.5. James Annis made a similar argument in 'An astrophysical explanation for the great silence' (JBIS 52, 19–22, 1999), proposing that the galaxy only recently became quiet enough for civilisations to appear, which is why nobody is ahead of us.

The record is data/papers/piran-2014.toml in the public repository.

Counting planets in Kepler's data with an independent search that also measured how many real planets the search would miss, the authors find that 22 percent (give or take 8) of Sun-like stars have a planet of one to two Earth radii receiving between a quarter and four times the light Earth gets, which they take as the habitable zone. That is the figure, 'one in five', that was widely reported and that most popular accounts of the equation have used since. The number depends heavily on where the edges of the habitable zone are drawn, as the paper says; narrower edges give a smaller share.

Erik A. Petigura, Andrew W. Howard, Geoffrey W. Marcy. Proceedings of the National Academy of Sciences 110, 19273–19278 (2013). Read it.

Bears on: Planets per star that could support life; The share of stars that have planets; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Superseded for the project's purposes by bryson-2020, which uses the full Kepler catalogue and the mission's final reliability figures, but kept because it is the number people remember.

The record is data/papers/petigura-2013.toml in the public repository.

Using only physics and engineering we already understand in outline (dismantling the planet Mercury for materials, solar-powered launchers, self-copying probes slowed by magnetic sails), the authors show that a single civilisation could launch probes to every galaxy within reach, millions of them, and that the whole launch would take about six hours of the Sun's energy output. The galaxy is therefore not the right scale for the paradox: if even one civilisation in a billion galaxies had ever chosen to expand, its probes should be here. That makes explanations that rely on every civilisation independently choosing not to spread much harder to hold, and pushes the weight onto explanations where civilisations are very rare or do not last.

Stuart Armstrong, Anders Sandberg. Acta Astronautica 89, 1–13 (2013). Read it.

Bears on: The share of intelligent species that make themselves detectable; The share of planets with life on which it becomes intelligent; How long a civilisation stays detectable, in years; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; The Great Filter is ahead of us: civilisations get as far as we have all the time, and something stops nearly all of them from going further.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. The title's six hours is the energy budget, not the travel time.

The record is data/papers/armstrong-2013.toml in the public repository.

Smart proposes that technological civilisations do not expand outward but inward: each step of progress packs more computation into less space, time, energy and matter, and the end of that road is something like a black hole, where a civilisation would find the densest possible computing and, perhaps, a way out of this universe altogether. If that is the universal pattern, every advanced civilisation disappears from view at roughly the same stage, shortly after ours, and the galaxy's silence is what a developmental law looks like from outside. He draws the practical conclusion that sending messages is pointless and that the search should look for the brief signatures of civilisations in transition.

John M. Smart. Acta Astronautica 78, 55–68 (2012). Read it.

Bears on: The share of intelligent species that make themselves detectable; How long a civilisation stays detectable, in years; Transcension: advanced civilisations shrink inward into ever denser, faster worlds of their own making, and leave the sky behind; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one; They became machines: the typical advanced civilisation is no longer biological, and machine minds live, think and travel in ways our searches do not expect.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. The idea is the one most often illustrated in fiction (Clarke, Banks, Vinge) and least often taken up in the research literature.

The record is data/papers/smart-2012.toml in the public repository.

When one star passes in front of a more distant one, its gravity briefly brightens the background star, and a planet around the nearer star adds a blip. Six years of watching for such blips, including the ones that did not happen, let the authors estimate how many planets stars have on average: about 1.6 planets per star in the range from five Earth masses to ten Jupiter masses at distances of half to ten times the Earth-Sun distance, and 62 percent of stars with a planet of a few Earth masses in that range. Because microlensing works for the dim red stars that make up most of the galaxy, which other methods miss, it was the first measurement that could fairly be said to cover the whole Milky Way: planets are the rule.

Arnaud Cassan, Daniel Kubas, Jean-Philippe Beaulieu, and the PLANET collaboration. Nature 481, 167–169 (2012). Read it.

Bears on: The share of stars that have planets.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. This is the source the planets factor's second estimate cites, and the schema's own example of a revision names this id; logged so that session A's factor file can link to it.

The record is data/papers/cassan-2012.toml in the public repository.

The common argument that life must be easy because it started so soon on Earth is weaker than it looks, the authors show, because we could only be here to notice if life had started early enough to leave time for us to evolve. Once that selection is taken into account, the early start is consistent with life being common and also with life being very rare, and which you conclude depends mostly on what you believed before looking. Their conclusion is that one planet's history cannot tell us the chance life starts elsewhere, and that finding a second, independent origin of life, on Mars or Europa, is the only thing that would.

David S. Spiegel, Edwin L. Turner. Proceedings of the National Academy of Sciences 109, 395–400 (2012). Read it.

Bears on: The share of planets that could support life on which life actually starts; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

The paper is the usual caution against reading a large value for the life factor off Earth's early fossils; Kipping 2020 (kipping-2020) is the main reply.

The record is data/papers/spiegel-2012.toml in the public repository.

The Kepler space telescope, launched in 2009-03, stared at one patch of sky containing about 150,000 stars and watched for the tiny dimming when a planet crosses in front of its star. This paper reports the first five planets from the first six weeks of data, all large and hot, and sets out the mission's real purpose: to count how common Earth-sized planets in the habitable zones of Sun-like stars are. Over the next nine years Kepler found more than 2,600 confirmed planets and showed that small planets are the most common kind, which turned the equation's second and third factors from guesses into statistics. Every occurrence-rate paper in this log rests on its catalogue.

William J. Borucki, and the Kepler team (David Koch, Gibor Basri, Natalie Batalha, Timothy Brown and others). Science 327, 977–980 (2010). Read it.

Bears on: The share of stars that have planets; Planets per star that could support life.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Checked 2026-10-16: NASA's Kepler mission page gives more than 2,600 planet discoveries and a field of about 150,000 stars.

The record is data/papers/borucki-2010.toml in the public repository.

The most careful modern survey of the explanations, by the astronomer who later wrote the book-length version (The Great Silence, Oxford University Press, 2018). Ćirković argues that the paradox has grown sharper, not weaker, since 1975, because we now know that planets are common, that the galaxy's habitable stars are mostly older than the Sun, and that spreading through it is cheap on cosmic timescales. He sorts the explanations into four families: solipsist (they are here and we are wrong about what we see), rare Earth (we are alone or nearly), neocatastrophic (civilisations keep being reset by disasters, natural or self-made), and logistic (there are limits on expansion or communication we do not yet understand). He rates most of the familiar explanations poorly and argues the catastrophic and logistic families deserve more work.

Milan M. Ćirković. Serbian Astronomical Journal 178, 1–20 (2009); arXiv:0907.3432. Read it.

Bears on: The share of intelligent species that make themselves detectable; How long a civilisation stays detectable, in years; The share of planets with life on which it becomes intelligent; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks; Natural catastrophes: asteroid strikes, giant eruptions and nearby exploding stars end civilisations before they can last long enough to be noticed; Self-destruction: civilisations destroy themselves soon after they learn to build the weapons that make them detectable; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; The zoo hypothesis: they know we are here and have agreed to leave us alone, the way a park leaves its wildlife undisturbed; The planetarium: an advanced civilisation is projecting an empty sky around us so that we believe we are alone; The simulation: the universe we observe is a computer model, and whoever is running it did not bother to include anyone else; Aestivation: advanced civilisations are asleep, waiting for the universe to cool so that their thinking will go further on the same energy; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

The catalogue's three families (never there, gone, unseen) are a plainer cut of the same material; Ćirković's four map onto them with his 'solipsist' and 'logistic' both falling under 'unseen'.

The record is data/papers/cirkovic-2009.toml in the public repository.

Hart's argument assumes that a civilisation keeps growing exponentially for millions of years. The authors point out that nothing on Earth has ever done that: fast-growing populations and economies hit limits and either collapse or level off. If the same holds elsewhere, civilisations that grow fast enough to fill a galaxy would be exactly the ones that burn out first, and the ones that last would be those that grew slowly or stopped, and therefore never spread far. The galaxy could then hold many old, stable, small civilisations and no empire. The paper's practical point is that the silence may be telling us something about the kind of growth that survives.

Jacob D. Haqq-Misra, Seth D. Baum. Journal of the British Interplanetary Society 62, 47–51 (2009); arXiv:0906.0568. Read it.

Bears on: How long a civilisation stays detectable, in years; The share of intelligent species that make themselves detectable; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; Civilisations burn out: the fast growth that makes a civilisation detectable also uses up its planet, and the ones that survive are the ones that slow down and go quiet; Self-destruction: civilisations destroy themselves soon after they learn to build the weapons that make them detectable; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16.

The record is data/papers/haqq-misra-2009.toml in the public repository.

Using models of how the galaxy built up its heavy elements and how often nearby stars explode, the authors map where and when in the Milky Way a planet could have held complex life long enough for it to matter. The answer is a ring between about 7 and 9 kiloparsecs from the centre (the Sun sits at about 8), which has been slowly widening as the galaxy aged: too close in and supernovae sterilise planets too often; too far out and there were not enough heavy elements to build rocky planets. Within that ring, three quarters of the stars that could host complex life are older than the Sun, by about a billion years on average. The paper is the standard reference for the idea that only part of the galaxy counts, and for the uncomfortable corollary that most habitable worlds had a head start on us.

Charles H. Lineweaver, Yeshe Fenner, Brad K. Gibson. Science 303, 59–62 (2004). Read it.

Bears on: How many stars are born in the galaxy each year; Planets per star that could support life; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks; Natural catastrophes: asteroid strikes, giant eruptions and nearby exploding stars end civilisations before they can last long enough to be noticed; We are early: the universe has barely started making civilisations, and we are among the first.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Checked 2026-10-16 against the abstract and text: the ring is between 7 and 9 kiloparsecs, 75 percent of its stars are older than the Sun, by about a billion years on average. The idea of a galactic habitable zone was introduced by Gonzalez, Brownlee and Ward in Icarus 152, 185–200 (2001); this paper is the quantitative version.

The record is data/papers/lineweaver-2004.toml in the public repository.

Bostrom argues that at least one of three things is true: almost no civilisation reaches the point of being able to run detailed simulations of its ancestors; almost none that reach it choose to; or we are almost certainly living in such a simulation, because the simulated people would vastly outnumber the real ones. The paper is not about the Fermi question, but it is the source of the simulation explanation for the silence: if our universe is a simulation, the sky may be empty because the simulators did not fill it, or because rendering other civilisations was not worth the cost. The argument's first two horns are themselves versions of the Great Filter and the inward turn.

Nick Bostrom. Philosophical Quarterly 53 (211), 243–255 (2003). Read it.

Bears on: The share of intelligent species that make themselves detectable; The simulation: the universe we observe is a computer model, and whoever is running it did not bother to include anyone else; The planetarium: an advanced civilisation is projecting an empty sky around us so that we believe we are alone; The inward turn: once a civilisation can build worlds of its own, made to measure and always rewarding, it stops bothering with the real one.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16.

The record is data/papers/bostrom-2003.toml in the public repository.

A palaeontologist and an astronomer argue that simple life may be common but that animals are not, because Earth's history depended on a long list of conditions that rarely come together: the right kind of star in the right part of the galaxy, a large moon that steadies the tilt, Jupiter sweeping up comets, plate tectonics recycling carbon, a magnetic field, and an unusually quiet several hundred million years in which complex cells could arise. They split the Drake Equation's habitable-planets factor into many smaller ones and conclude that the product is tiny. The book named the position, and every later argument about how rare Earth-like conditions are starts by agreeing or disagreeing with its list.

Peter D. Ward, Donald Brownlee. Copernicus Books (Springer), New York, 2000, 333 pages. Read it.

Bears on: Planets per star that could support life; The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks; Complex cells are rare: life may start easily but then stay simple forever, because the one step that made big cells happened only once; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

The usual replies are that several items on the list (a large moon, Jupiter's protection, plate tectonics) have since been shown to matter less or to be less rare than the book assumed; the exoplanet record since 2009 has been the main test of its claims about stars and planets.

The record is data/papers/ward-2000.toml in the public repository.

Hanson's essay gives the modern framing. Somewhere between dead matter and a civilisation that spreads across the galaxy there must be at least one step that almost nothing gets past, because if every step were easy the galaxy would be visibly full. He calls that step the Great Filter and asks the only question that matters for us: is it behind us or ahead? If life, or complex cells, or intelligence is the hard step, we are through it and the silence is good news. If the hard step is surviving technology, then every species gets this far and then stops, and the silence is a warning. He draws the practical conclusion that finding simple life on Mars or Europa would be bad news, because it would move the filter forward.

Robin Hanson. Essay published on the author's website, George Mason University, 1998 (first version 1996). Read it.

Bears on: The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; How long a civilisation stays detectable, in years; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it; The Great Filter is ahead of us: civilisations get as far as we have all the time, and something stops nearly all of them from going further; Self-destruction: civilisations destroy themselves soon after they learn to build the weapons that make them detectable; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; Complex cells are rare: life may start easily but then stay simple forever, because the one step that made big cells happened only once.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Nick Bostrom's 2008 essay 'Where Are They? Why I hope the search for extraterrestrial life finds nothing' (MIT Technology Review) restates the Mars point for a general reader.

The record is data/papers/hanson-1998.toml in the public repository.

Landis, a NASA physicist, models settlement of the galaxy the way physicists model a liquid seeping through rock. Each settled star either sends out settlers to its neighbours or does not, with some fixed probability; if that probability is below a critical value, the settled region stops growing by itself and leaves large voids, and if it is above, the galaxy fills. The point is that Hart's and Tipler's argument assumes every colony keeps colonising, forever. Relax that one assumption and a galaxy can hold many civilisations, some of them very old, with unsettled regions the size of our neighbourhood, without any of them choosing to hide.

Geoffrey A. Landis. Journal of the British Interplanetary Society 51, 163–166 (1998). Read it.

Bears on: The share of intelligent species that make themselves detectable; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

The model has been extended several times, notably by Carroll-Nellenback, Frank, Wright and Scharf in 'The Fermi Paradox and the Aurora Effect' (Astronomical Journal 158, 117, 2019), which adds moving stars and settlements that die out.

The record is data/papers/landis-1998.toml in the public repository.

The first planet found around an ordinary star other than the Sun: a body of at least half Jupiter's mass circling the star 51 Pegasi every 4.2 days, closer to its star than Mercury is to ours, found from the small back-and-forth wobble it gives the star's light. Nobody had predicted a giant planet so close in, and the discovery showed both that planets exist elsewhere and that other planetary systems need not look like ours. Until this paper the share of stars with planets, the second factor in the equation, was a guess; after it, it was a measurement waiting to be made. The authors shared the 2019 Nobel Prize in Physics for it.

Michel Mayor, Didier Queloz. Nature 378, 355–359 (1995). Read it.

Bears on: The share of stars that have planets.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Written 2026-10-16. Planets around a pulsar, a dead star, had been found by Wolszczan and Frail in 1992; this was the first around a star like the Sun.

The record is data/papers/mayor-1995.toml in the public repository.

Carter notices that intelligence on Earth took about four billion years to appear, and that the Sun will only stay fit for life for something like five to ten billion years in all. If intelligence usually took far less time than a star's life, we would expect it to have appeared early, not with the clock nearly run out; if it usually took far more, we would not be here at all. The near-match, he argues, is what you would see if the expected time for intelligence to evolve is much longer than a star's lifetime, and we are a lucky exception. From the timing he estimates that the path from simple life to observers has only one or two very improbable steps, and that intelligent life is correspondingly rare. This is the argument behind every 'hard steps' calculation since.

Brandon Carter. Philosophical Transactions of the Royal Society of London A 310, 347–363 (1983). Read it.

Bears on: The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; Rare life: getting from chemistry to the first living cell may be so unlikely that it almost never happens; Complex cells are rare: life may start easily but then stay simple forever, because the one step that made big cells happened only once; The Great Filter is behind us: somewhere on the road from dead matter to a galaxy-spanning civilisation there is a step almost nothing passes, and we have already passed it.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

The Snyder-Beattie, Sandberg, Drexler and Bonsall 2021 paper in this log applies the same reasoning with modern dates for Earth's transitions.

The record is data/papers/carter-1983.toml in the public repository.

The first full survey of the question, written by an astrophysicist who went on to be a novelist. Brin coins 'the Great Silence' for the fact that needs explaining, lays out the two camps (those who say the galaxy should be full and conclude it is empty, and those who say the search has barely started), and then catalogues the explanations offered to that point: colonisation that stalls, deadly self-replicating probes, worlds covered by water whose inhabitants never build radios, civilisations that keep quiet out of fear, zoos, and self-destruction. He does not pick one. His own contribution is the observation that a galaxy with even one dangerous species in it would teach every other species to be quiet, which is the dark forest argument twenty-five years early.

Glen David Brin. Quarterly Journal of the Royal Astronomical Society 24, 283–309 (1983). Read it.

Bears on: The share of intelligent species that make themselves detectable; How long a civilisation stays detectable, in years; The share of planets with life on which it becomes intelligent; The zoo hypothesis: they know we are here and have agreed to leave us alone, the way a park leaves its wildlife undisturbed; Berserkers: self-copying machines built long ago still roam the galaxy and destroy any civilisation that shows itself; Self-destruction: civilisations destroy themselves soon after they learn to build the weapons that make them detectable; The dark forest: everyone is hiding, because announcing yourself gets you killed; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached; Rare Earth: simple life may be common, but a planet that can carry complex life needs a long list of lucky breaks; Shouting is costly: a beacon that could be heard across the galaxy is expensive to run, so anyone sending one sends short pulses, rarely, and our searches almost never catch them.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Most of the catalogue's explanations that predate 1983 are gathered here, and the paper is the usual starting point for any later survey.

The record is data/papers/brin-1983.toml in the public repository.

Tipler sharpens Hart's argument by removing the need for anyone to travel. A civilisation with our level of rocketry plus a machine able to copy itself from raw materials (the self-reproducing probe that John von Neumann described) could send one probe, which builds more at each star, and have a probe at every star in the galaxy within about 300 million years for an initial outlay smaller than running a radio beacon for a few centuries. Since no such probes are in the solar system, he concludes that no other technological civilisation has ever arisen in the galaxy, and that the radio search is founded on a mistake. The paper is where the self-copying probe entered the debate, both as the reason the galaxy should be full and, in later hands, as the thing that might be keeping it empty.

Frank J. Tipler. Quarterly Journal of the Royal Astronomical Society 21, 267–281 (1980). Read it.

Bears on: The share of planets with life on which it becomes intelligent; The share of intelligent species that make themselves detectable; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; Berserkers: self-copying machines built long ago still roam the galaxy and destroy any civilisation that shows itself; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here; They are already here: visitors or their machines have reached the solar system, now or long ago, and we have not recognised them.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Checked 2026-10-16 against the scanned paper: the 300 million years is his bound for present-day rockets (his faster probe speeds give about 4 million years); the cost is his, not 'a single launch' (about 3 billion dollars, a tenth of Apollo, he says). The ADS 'full' link returned an empty page on the check; the scan is at articles.adsabs.harvard.edu/pdf/1980QJRAS..21..267T. Sagan and Newman answered it in 1983 ('The Solipsist Approach to Extraterrestrial Intelligence', same journal, 24, 113–121), arguing that a civilisation that would build such probes would not survive them.

The record is data/papers/tipler-1980.toml in the public repository.

The paper that turned Fermi's lunchtime question into an argument. Hart starts from one observation, that there are no intelligent beings from elsewhere on Earth now, which he calls Fact A, and argues that a civilisation able to travel between stars at even a tenth of the speed of light would cross most of the galaxy in about 650,000 years (about twice that if there are pauses between voyages), so any civilisation that began exploring more than about 2 million years ago would already have reached us. That is a blink against the galaxy's age. He sorts every way out into four kinds (they cannot come, they chose not to, they have not had time, they came and we missed them) and rejects each, and concludes that we are the first civilisation in the galaxy and that searching for radio signals is probably a waste of time and money.

Michael H. Hart. Quarterly Journal of the Royal Astronomical Society 16, 128–135 (1975). Read it.

Bears on: The share of planets with life on which it becomes intelligent; The share of intelligent species that make themselves detectable; How long a civilisation stays detectable, in years; We are early: the universe has barely started making civilisations, and we are among the first; Rare intelligence: evolution has no goal, and of the billions of species that have lived on Earth only one built anything; Interstellar travel is impractical: the distances are so great and the costs so high that nobody ever goes anywhere, so nobody has come here; The zoo hypothesis: they know we are here and have agreed to leave us alone, the way a park leaves its wildlife undisturbed; They are already here: visitors or their machines have reached the solar system, now or long ago, and we have not recognised them; Expansion stalls: civilisations do spread between stars, but settlement spreads like a stain, not a flood, and leaves large regions, perhaps ours, never reached.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Corrected 2026-10-16 against the scanned paper (articles.adsabs.harvard.edu/pdf/1975QJRAS..16..128H; the ADS 'full' link returned an empty page): at a tenth of light speed most of the galaxy is crossed within 650,000 years; the two million years is the time since a civilisation would have had to start exploring to have reached us. The 'Fact A' framing is Hart's own and most later surveys cite it. Nearly every explanation in the catalogue's 'unseen' family is an answer to this paper.

The record is data/papers/hart-1975.toml in the public repository.

A three-page note from a radio astronomer at Harvard's observatory proposing that the galaxy may hold civilisations far older than ours that have deliberately set our region aside, as a wilderness or a zoo, and avoid any contact that would disturb it. Ball's point is that this is the only explanation he can find that fits both a galaxy old enough to have been settled many times over and the fact that nobody is here; he admits it is unfalsifiable and unpleasant, and says so. The paper named the idea that Star Trek had already put on television as the Prime Directive, and it remains the reference every later 'they are watching' explanation cites.

John A. Ball. Icarus 19, 347–349 (1973). Read it.

Bears on: The share of intelligent species that make themselves detectable; The zoo hypothesis: they know we are here and have agreed to leave us alone, the way a park leaves its wildlife undisturbed; The planetarium: an advanced civilisation is projecting an empty sky around us so that we believe we are alone; No interest in contact: they exist, they could reach us, and they simply do not care to.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Ball returned to the idea in 'Extraterrestrial Intelligence: Where Is Everybody?' (American Scientist 68, 656–663, 1980).

The record is data/papers/ball-1973.toml in the public repository.

Drake wrote the equation on a blackboard as the agenda for a three-day meeting of eleven people: the number of communicative societies in the galaxy equals the rate at which stars form, times the share with planets, times the number of planets per star suitable for life, times the share on which life starts, times the share where intelligence appears, times the share that are communicative, times how long they stay so. The meeting's own values, as its first printed account reports them (Pearman, 1963), were about 1 star a year, with as many as 10 if some older stars are counted; 0.2 to 0.5 for the share with planets, all called weak arguments; 1 to 5 planets per star suitable for life; about 1 for life starting and about 1 for intelligence following; 0.1 to 0.2 for the share that are communicative; and a lifetime that might have two humps, under a thousand years for societies that destroy themselves and hundreds of millions for those that do not. The account says the lifetime decides the answer. The equation was never meant as a calculation; it was a way to organise an argument, and it has done that job for sixty-five years.

Frank Drake, and the Green Bank meeting: Dana Atchley, Melvin Calvin, Giuseppe Cocconi, Su-Shu Huang, John Lilly, Philip Morrison, Bernard Oliver, Carl Sagan, Otto Struve, J. Peter Pearman. Order of Dolphin meeting, National Radio Astronomy Observatory, Green Bank, West Virginia, 1961-11-01 to 1961-11-03; first published account in J. P. T. Pearman, 'Extraterrestrial Intelligent Life and Interstellar Communication: An Informal Discussion', in Interstellar Communication, ed. A. G. W. Cameron (Benjamin, 1963), 287–293. Read it.

Bears on: How many stars are born in the galaxy each year; The share of stars that have planets; Planets per star that could support life; The share of planets that could support life on which life actually starts; The share of planets with life on which it becomes intelligent; The share of intelligent species that make themselves detectable; How long a civilisation stays detectable, in years.

What it moved: nothing. It was read against the ranges it bears on and none of them changed. Logging it anyway is how the record shows what has been considered.

Corrected 2026-10-16: the headcount in the finding was eleven; the author field names ten and the secondary accounts say about ten (Drake himself says he invited twelve). Founding entry. There is no 1961 paper by Drake presenting the equation; the earliest printed version is Pearman's 1963 account, and Drake's own first published form is in his 1965 chapter 'The Radio Search for Intelligent Extraterrestrial Life' in Current Aspects of Exobiology. The link is the SETI Institute's page on the equation. Checked 2026-10-17 against Pearman's chapter itself (read in full at the Internet Archive, https://archive.org/details/interstellar_communication_1963): it runs on pages 287 to 293 of Interstellar Communication (the next chapter, Oliver's, begins on page 294), so the page range is confirmed. Corrected: the chapter lists eleven attendees, Struve (chairman), Atchley, Calvin, Cocconi, Drake, Huang, Lilly, Morrison, Oliver, Sagan and Pearman, so the headcount is eleven as the finding first said, and the author field lacked Giuseppe Cocconi (named on the volume's contributors list); the finding is rewritten from the chapter, whose values for each factor are given on its pages 289 to 292 and are now the source of the factor records' 1961 estimates. The star-formation question is answered there: about 1 a year, up to 10 with older stars counted. The finding's earlier sentence that the values made the number of civilisations roughly equal to the lifetime, between 1,000 and 100 million, is removed: it came from a secondary summary, and the chapter's own figures for the number of civilisations could not be read in the scan's text. Drake and Sobel (1992) and Drake, Vakoch and Dowd (2015) retell the meeting and have not been read here; they remain worth a cross-check when the books arrive.

The record is data/papers/drake-1961.toml in the public repository.